Last week we witnessed the biggest product launch of the first half of 2026 in AM. Formlabs unveiled the Fuse X1, and it represents a breakthrough on several levels at once.

First, this is finally the company’s genuine entry into the industrial sector. For years, the Fuse 1 was marketed and sold as a desktop and low-cost machine, even though it had long been operating in an entirely different league. The X1 puts that pretense aside.

Second, we’re getting quite interesting technology at a price point that looks very attractive compared to the competition. That’s a rare combination in this segment.

Third - and this is where things get interesting - the X1 poses a serious threat to EOS, HP, and 3D Systems in their powder-bed polymer businesses. It reaches even further, challenging Chinese industrial manufacturers such as Farsoon and TPM3D by making their path into Western markets considerably more difficult. And for Sinterit, it’s simply a death sentence.

All of this sounds like the beginning of something great for Formlabs. At first glance, the future looks wonderful and bright.

And yet, I find it hard to be optimistic.

Because from the perspective of someone who has been watching this market for over thirteen years, the Fuse X1 launch looks less like a triumph and more like a great escape forward.

An escape from the growing pressure of FFF giga-farms that are already delivering, at scale, what SLS and MJF have been promising for a decade - and never fully achieved.

The machine is excellent.

The question is: what exactly is it running from?

What’s this Fuse all about?

Joris Peels has once again published a great analysis on 3DPrint.com that goes beyond the standard copy-and-paste press release. So instead of writing about the printer itself (I assume most of you already know everything there is to know about it), I’d encourage you to read Joris’ article.

I’ll also break down some of his most interesting observations, because they’re worth a closer look.

One of the more interesting observations is his argument that the X1’s low price will intensify competition in polymer PBF and should expand the market as a whole. It’s difficult to disagree. The Fuse X1 automatically puts significant pressure on EOS and even HP, while also making life more difficult for Farsoon, which is already handicapped in Western markets due to its Chinese origins.

Alongside the launch of the new printer, Formlabs also introduced its own manufacturing service. Peels sees this as a strong “try before you buy” model and recommends that everyone adopt something similar because it makes scaling easier for customers.

He does, however, add a caveat: Formlabs’ service competes directly with service bureaus, which may not be thrilled about the idea. His proposed solution is a partnership model where service providers operate Formlabs equipment while the service helps fill their excess capacity.

Personally, I think it’s a great idea for a 3D printer manufacturer and a terrible one for customers. Who, after buying the machines, will find themselves competing against their own supplier. But that’s just my opinion. I can only assume that it’s very tempting from a revenue perspective, and that certain people have decided it’s worth more than the criticism and complaints posted on social media and in comment sections.

Equally interesting is the fact that Formlabs officially disclosed its revenue for the first time. The company generated more than $250 million in revenue in 2025. That’s an impressive figure for a company of this type.

Peels interprets this transparency as a clear signal that an IPO may be on the horizon. And here’s the most unexpected twist: he suggests that the listing could take place in China.

Formlabs already manufactures its machines there and could potentially achieve a significantly higher valuation. If that happened, the implications for the global AM industry would be enormous.

Well, I’d say, we’ll see...

And then there’s the most important topic: materials.

Peels claims, that if the system is open to third-party materials and process parameters, Formlabs wins. If the ecosystem remains closed, it loses.

Delaying openness may preserve high short-term margins on powder sales, but it also creates opportunities for HP and emerging Chinese competitors. A closed ecosystem would also stifle advanced research because nobody could modify process parameters or integrate their own software.

That’s Peels’ view, and I recommend reading the full article.

My view is different and a bit pessimistic.

The biggest threat to Formlabs and the entire powder-bed industry is material cost.

Bigger than EOS and HP combined. Bigger even than the debate about open versus closed systems.

Will the best powder system lose to the FFF giga-farm?

For three decades, 3D printing has excelled at one-off parts and short production runs. Whenever production scales up, economics deteriorate dramatically.

This is described by the Second Market Law of Additive Manufacturing:

I’ve also written about this before as well in context of change of sentiment in AM production:

Today, that sounds almost heretical because it challenges a paradigm that the industry spent years building:

Rapid manufacturing = powder bed fusion.

which was reinforced by stereotype:

Desktop printers = cheap, low-volume junk.

The problem is that this stereotype has been dead for more than three years, and the old paradigm is beginning to crack.

Business owners are asking a simple question:

What’s the better investment?

• one SLS machine - or one hundred Chinese FFF printers?

• one build chamber with stacked parts - or one hundred build plates with no stacking required?

Young entrepreneurs entering additive manufacturing often don’t even consider SLS or MJF. To them, these technologies seem as distant and prohibitively expensive as anything else outside their budget.

They simply buy FFF.

If 3D printing is to make sense in serial production, low material cost is critical. Without it, scaling is impossible.

The machines themselves must also become cheaper.

China understands this, and prices across the ecosystem continue to fall.

That’s how you end up with quotations measured in tenths of a RMB per gram of processed material - something that appears completely absurd from the perspective of a Western AM company.

The Czech Guy can cry a river that all of this is unfair and driven by government subsidies, but regardless of whether he’s right or wrong, all he can really do is wipe the tears with a tissue bought on TEMU.

Either way, this is where things become uncomfortable for Formlabs.

The company started losing ground to Chinese desktop competitors already during the Fuse 1 era and, more recently, in the resin segment as well.

Where certification matters: dental, medical, regulated industries, where both process and material require approval, or where specialized materials are needed, such as jewelry casting and investment casting applications, Formlabs remains strong.

But in mainstream industrial production, both SLS and resin have increasingly lost ground to FFF farms: cheaper, easier to scale, and more frequently “good enough” in terms of quality.

That’s why I interpret the Fuse X1 as an escape forward into more specialized and advanced industrial applications.

In sectors where Chinese desktop FFF systems cannot easily enter due to regulatory, procedural, or process-related barriers - defense being an obvious example - the Fuse X1 will be highly competitive.

But mass adoption? That’s a different story.

The only real solution

There is only one solution, and it isn’t easy:

Everyone understands that SLS will never achieve cost parity with FFF, and powdered PA will never cost the same as PA filament (which itself can often be replaced by ordinary PETG as „good enough”).

But the difference between paying €100-150 per kilogram for PA powder and paying 30-40% of that amount for filament becomes enormous when production scales.

The conclusion is simple, and uncomfortable.

In the long run, Formlabs has no choice but to move further into industrial manufacturing because the desktop segment has already been largely captured by Chinese manufacturers.

The Fuse X1 is a good move.

Given the rules of this game, it’s a very good move.

But it remains an escape.

The real battle for high-volume manufacturing is already being fought on a different front.

Blue Gravity Capital has invested tens of millions of złoty into Spectrum Group and acquired a majority stake. Founder and CEO Michał Żołądek remains at the helm. Within roughly 12 months the company plans to double its production capacity.

Spectrum is probably the biggest European filament manufacturer.

It is also a profitable, growing Polish company built entirely through organic growth. It has spent the last five years in a state best described as: “we have more orders than we can produce.”

A permanent backorder situation. And that is precisely why it is bringing in outside capital. Its own success had started to become a threat...

I’ve known Michał for more than 10 years. I was the one who published the article announcing Spectrum’s market debut on my former portal, Centrum Druku 3D.

It was June 25, 2015. I arrived in a small village - Pęcice near Warsaw, on a rainy evening, where two guys welcomed me with beer and a table full of snacks while convincing me that their PLA was the best filament in the world.

One of those guys, Michał Żołądek, is now my dear friend.

The filament market created by Chinese printers

Just a few years ago, a 3D printer required patience and humility. Then the Chinese added lidar, flow calibration, and error detection to inexpensive machines. All of a sudden, a hobbyist device costing €300 to €1,000 began outperforming industrial equipment costing €10,000 to €20,000 in many applications.

Sales sky-rocketed.

The figures cited by the parties involved in the transaction are impressive. Chinese exports of desktop 3D printers reached 3.7 million units in 2024, 5.2 million in 2025, and are projected to hit approximately 8 million in 2026.

And every new printer consumes filament.

Blue Gravity currently values the filament market at around $1 billion and assumes annual growth of roughly 20%, while noting that this may be a conservative estimate because it does not fully account for what is happening in the printer market. The fund adds that CAGR could rise to 30%, and periodically even 40%.

Virtually all demand growth is currently concentrated in low-cost desktop machines. In the second quarter of 2025, entry-level printer revenues grew by more than 20%, while professional, mid-range, and industrial segments declined.

In other words, the fastest-growing part of the market is also the one that consumes the most spools.

This brings us to the first uncomfortable fact: China dominates filament manufacturing as well. Its advantages are scale and price.

However, China has one weakness, and Spectrum knows how to capitalize on it.

Shipping from Asia is expensive, and freight rates can fluctuate dramatically from quarter to quarter. Manufacturing in Europe means a shorter supply chain, shorter lead times, and less exposure to freight market volatility.

What is not so important in 3D printers, becomes very important in filament.

How Spectrum went so far

Spectrum launched in 2015 with ordinary PLA and extraordinary colors.

Everything that followed was built internally: through operations, leasing, and bank financing. No investment rounds. No venture capital. Not once during the entire decade.

Between 2018 and 2025, the company increased production volume by an average of around 50% annually, growing faster than most of its European competitors.

Step by step, it progressed from being Poland’s largest filament manufacturer, to the largest in Central and Eastern Europe, and eventually to becoming one of the largest in Europe.

Today’s scale: 1,500 tonnes of filament annually, more than 100 materials, sales in over 40 countries, and approximately 80% of revenue generated from exports.

But let’s be honest: the gap between Spectrum and the Chinese filament manufacturers in terms of production volume is enormous.

That is why Spectrum competes on different factors: European manufacturing, shorter delivery times, consistent quality, and a portfolio that goes far beyond commodity PLA into technical, engineering-grade, specialty, and compostable materials.

According to Żołądek, the company can already match Asian competitors on price in standard materials.

It is also pursuing acquisitions. Hungary’s Filaticum (to be relaunched) and Poland’s Prografen, known for niche graphene-based filaments, are already part of the group, while another Polish brand is currently in the pipeline.

It is no coincidence that Żołądek openly states that he does not rule out further market consolidation in Europe in the near term.

The operational plan for the next year is remarkably straightforward: double production capacity. No new factory for now, just a reorganization of the existing facility and the addition of new production lines. The company openly admits that at this pace it will run out of space soon, but that is a problem for later.

Every large filament company stared at the same wall

Every member of China’s filament Big Four faced exactly the same obstacle at some point.

More orders than capacity. No more room on the factory floor. Customers who constantly hear: “please wait.”

You don’t break through that wall with engineering talent or founder determination alone.

You break through it with capital.

Money allows a company to scale ahead of demand instead of spending five years chasing it. It’s the only path I’ve ever seen, and I don’t know of any exceptions.

Spectrum is doing exactly that.

It is giving up a majority stake, taking in tens of millions, and buying itself a head start instead of another decade of apologizing for lead times.

Will Spectrum catch up with the Chinese Big Four in terms of production volume?

Probably not in this decade. Maybe never.

But that is not particularly important. Because there will always be room in the global market for at least one strong European filament manufacturer.

And if anyone on this side of the map has the credentials to become that manufacturer, it is the company that has spent the last five years struggling to keep up because it simply has too many customers.

The rest is a matter of capacity. And it has just bought itself more of it.

Atomic Layer from the Past:

Two layers today:

13 years ago, 3D Systems acquired 80% of Phenix Systems, a French leader in DMLS metal 3D printers. The deal closed in July 2013, with the remaining 20% bought later. Phenix’s technology served aerospace, automotive, and medical sectors. This acquisition complemented 3D Systems’ portfolio, strengthening its presence across all major additive manufacturing markets.

Read all:

11 years ago, GAIA Multitool (now 3DArtech) released the GAIA Multitool MAXX, a large-format ceramic 3D printer standing 187 cm tall. Featuring delta kinematics and a ⌀45 x 105 cm build area, it included ten interchangeable tools: ceramic extruders, FFF heads, engraver, cutter, CNC mill, and laser engraver.

Read all:

News & Gossip:

#1

Formlabs just dropped the Fuse X1, and this is one of the biggest stories not of the year but of the last few years!

So I’ll take it apart properly in Monday’s 3DP War Journal.

For now, just the basics (you probably know):

• it’s Formlabs’ first SLS built for industrial series production

• build volume 330 × 330 × 565 mm, 61.5 liters, a full unit done in roughly 24 hours, 120 W laser, build-unit swaps in 5 minutes, plus Adaptive Thermal Control and Print Intelligence vision monitoring

• the company promises up to 3x throughput and half the per-part cost of classic industrial systems.

#2

I mentioned this one already in a LinkedIn post, but it is significant enough to highlight it here again.

BMF reshuffles its top team: Bryan Ferrand becomes President, Donna Kelly steps up to COO, and John Kawola slides into a strategic advisor seat.

For the record, the organisational changes at BMF maybe isn’t the headline, but Kawola is. The man is a genuine industry legend, and watching him step back from the wheel is significant story.

The official line calls it confidence in the new team. Yeah, well, probably true. Still a shame to see him go.

#3

TDK is buying San Diego’s Fabric8Labs for up to $400 million, part upfront, part multi-year earnout. The prize is ECAM, an electrochemical metal 3D printing process aimed at thermal management parts for data center cooling.

Not a coincidence: TDK Ventures backed the company at seed, so this is a quiet bet maturing into ownership. And read the room. A profitable electronics giant pays nine figures for a metal AM firm because of where it points, data center heat, not because metal printing itself prints money. The part that sells is the cooling problem. The printing is just how you solve it.

#4

Photocentric spun its space division into a separate company, CosmicMaker, after testing an LCD-based printer under shifting gravity.

The pitch: print spare parts and functional components off-world, no resupply rocket from Earth required.

In April three identical printers ran on parabolic flights from 2g to 0g, printing silicon carbide, aluminum oxide and two thermoset plastics, parts came out to spec, all backed by UK Space Agency and ESA money.

The clever bit: in microgravity the ceramic slurries actually segregated less, and the sealed liquid chamber kills the need for supports. Years from orbit, sure. But the physics is cooperating, which is more than most space-AM stories can say.

#5

HeyGears unveils the G1, a desktop printer that promises full-color models straight off the bed, no hand-painting, no six-figure industrial system. The G1X runs an Epson i3200 head with 3,200 nozzles, eight ink channels, 3.9-picoliter droplets and over ten million mixable colors, plus water-soluble supports and modules that print on 400-plus substrates from metal to textiles.

There's the obligatory AI bit too: text and image-to-3D, 500 templates, days-to-minutes.

Two catches worth holding onto. No price yet, just a discounted waiting list. And HeyGears admits the units shown may be prototypes or renders. Promising spec sheet.

Now show me the parts…

#6

MyMiniFactory wants to bring Thingiverse back from the dead.

The platform that basically invented the 3D-model directory has been passed around since its MakerBot days, and its current owner admits the previous ones barely touched it, so designers walked.

The fix: an ad-free version (ads were the number-one complaint), premium paid features to fund it, and a rebuilt monetization system so creators earn more than spare-change tips.

Anthropic has unveiled Claude Fable 5. At the same time, the official Claude YouTube channel released a video showing the model building its own design tool: a browser-based CAD editor called VibeCAD, complete with a built-in assistant.

Then, inside the very environment it had just created, it designed a small lighthouse for a desk.

All from a single sentence of description!

The model added a base, a tower with colored stripes, a lantern room with columns and a dome. Along the way, it even noticed that the object was supposed to fit next to a monitor and adjusted the proportions accordingly.

It’s impressive. Really.

Geometry, code, and design in a single workflow, without jumping between five different programs and two file formats.

And then you scroll down to the comments.

Man, I can’t even afford to run Opus.

And suddenly, all the magic disappears...

Because Fable 5 is the most powerful model Anthropic has released to the public so far. The problem is that it’s even more expensive to use than Opus - a model that most users had already found difficult to afford.

So now we have a designer that can create objects on demand and even think about where the monitor will sit on the desk. Yet there’s one question the video never addresses.

What’s the point of having such a designer if so few can afford to use it?

What is it about?

Let’s start with what actually happened, because it is still a major achievement.

The model wasn’t given a ready-made editor. It first wrote the tool for itself, and only then started designing the object within it.

It iterated through several inputs, saving intermediate stages in the interface, until a rough description became a closed solid model. This is precisely the design layer into which the industry has been pouring money and hope for years.

Pure software.

Of course, we don’t know whether the design is actually functional. Can it be printed? Anthropic is showcasing its best model in the best possible light, while openly admitting that it doesn’t know whether the file will produce a usable object or a pile of spaghetti.

But that’s not the point.

Those issues can be refined and improved.

What matters are the numbers.

For rich people only

Fable 5 costs $10 per million input tokens and $50 per million output tokens. Until June 22, it’s included at no extra cost in the Pro, Max, and Team plans. Starting June 23, users will need to pay with credits unless Anthropic extends the promotional period.

In other words, the free designer comes with an expiration date. After that, it moves into a pricing tier above Opus.

And that’s where the real question emerges:

3D printing makes money where parts are unique, inexpensive, and needed immediately. A desk lighthouse is a textbook example of that category. A small object of the kind designers create by the hundreds.

But if generating the file costs more than drawing it manually, then we’re looking at a very expensive solution to a very cheap problem.

So how much does one such lighthouse actually cost in tokens?

Let’s estimate, because Anthropic did not provide exact figures in the demo, and I’m not gonna invent them.

A session that first builds an entire CAD editor with an assistant and then spends several rounds refining a design is a lengthy agentic workflow. At every step, the model rereads context, causing input token consumption to balloon.

Realistically, we’re talking about output in the range of 100,000 to 300,000 tokens, with input usage expanding into the low millions.

Using Anthropic’s pricing, that works out to somewhere between a dozen and several dozen dollars for a single lighthouse.

To be clear: that’s an estimate.

But the order of magnitude is realistic.

Now let’s look at the other side of the table.

A designer from virtually any global freelance marketplace could model the same object in thirty minutes, maybe an hour. For a comparable fee, often less. And they won’t bill you for three failed iterations where the geometry didn’t close properly.

The commenters under the video figured this out faster than any marketing department could.

You can literally do this in an hour with actual usable models.

Another commenter asked directly why Anthropic keeps adding features nobody requested before fixing its pricing and token structure.

A third summed up the company’s strategy: prioritizing raw power over efficiency will end badly, in their view, and they wished Anthropic luck with the small number of customers who will actually be able to afford it.

It’s easy to dismiss these reactions as complaints.

That would be a mistake.

AI is genuinely moving into the design layer, and it’s getting better at it at an astonishing pace. The direction is absolutely correct. A model that builds its own tool and then designs a printable object inside it is a preview of what design-for-additive-manufacturing workflows may look like in a few years.

The problem is today’s arithmetic

Once again, the vision arrived before the economics.

We’ve seen this before in the additive manufacturing industry, when consumer 3D printing was being promised a decade before Chinese manufacturers made it genuinely affordable.

First comes what we can do. Only later comes what we can afford.

Today, Fable 5 is a showroom exhibit. It is still far from the factory floor.

A brilliant, expensive exhibit that people admire from behind the rope, because after June 22, very few will be able to afford the admission ticket.

For now, the cheapest human designer still outperforms the world’s most expensive designer.

There are things we never imagine ourselves doing. Things that seem completely abstract. For example, I never thought I would be testing and walking around in custom orthopedic insoles made on a 3D printer.

For the past few days, I’ve been breaking in a pair that arrived from Hungary. Printed in TPU, finished with an elegant leather top layer, with my name put on them.

This is one of those things that don’t look 3D printed at all. You’d need someone with several years of experience in the industry to even notice.

An insole like this isn’t a 3D print - it’s a finished, functional product.

I became their owner somehow by accident. I was in Budapest attending an industrial trade show together with 3Dee Hungary. On the last day of my trip, I went to see Batz - one of the Hungarian shoe companies that uses Bambu Lab 3D printers in its daily operations.

I walked out with my feet scanned and a promise that two weeks later I would receive my own custom insoles.

Which is exactly what happened.

And I found that remarkable. Because after 14 years of covering this industry, I’ve learned one thing: real-world applications of 3D printing rarely look the way startup presentations promised they would to potential investors.

The market for 3D printed insoles is changing rapidly right now. Quietly, without press releases. More and more pairs are being produced on standard FFF printers using increasingly specialized TPU materials. And hardly anyone is talking about it.

So let me ask a question that’s been on my mind ever since I returned from Budapest.

What if, in a few years, it turns out that the world’s entire fleet of FFF 3D printers is used primarily to print shoes, or insoles for shoes?

After a decade of promises about the next industrial revolution, after all the battles over who should control the openness of these machines, we might discover that FFF printers are simply on-demand footwear manufacturing systems.

Wouldn’t that be wonderfully ironic?

Visiting Batz

In mid-May, I was in Budapest. I spoke at Industrial Days about the four eras of additive manufacturing. And on the final day, together with 3Dee Hungary, the local authorized Bambu Lab reseller, visited one of their customers: Batz.

Batz is a Hungarian footwear manufacturer. Slippers, sandals, sneakers, ankle boots, direct-to-consumer sales, a retail store network, more than 11,000 reviews with an average rating of 4.78, collaborations with local fashion designers.

A classic comfort footwear brand, the kind you can find throughout the region.

With one important detail: Batz prints insoles on 3D printers.

In the production area stands a fleet of H2D Pro machines. They run continuously and perform one task: producing personalized orthopedic insoles.

The service is integrated from start to finish. First comes a foot scan and a pressure map analysis. In reality, that’s the product, not the plastic.

Using this data, a custom insole model is created for the specific foot of a specific person. Then comes TPU printing and leather finishing, followed by personalized markings. In my case, my name and a unique identification code.

Two weeks later, the package arrived in Poland. Well packed, visually appealing. Insoles that look like a finished retail product.

I’m only in the first few days of wearing them, and honestly, the beginning is strange. Dominik Marczell from Batz had warned me about that. Your foot needs time to adapt to the new support structure and gradually correct its position, so for a while it feels like something is wrong before it starts feeling right.

I already know I’ll order additional pairs for my other shoes. And that’s the most interesting part from a business perspective: I don’t need to travel anywhere or have my feet scanned again. Batz already has my model. I click, they print, the package ships.

And now to the reason I’m writing this at all. Batz is not an exception. FFF printing is quietly becoming the default method of insole manufacturing worldwide.

What the world prints when nobody is watching

Let’s start with an uncomfortable truth. Insoles are the perfect product for 3D printing, and for reasons that don’t sound particularly exciting from a conference stage.

Every foot is different. The production run is exactly one pair. The geometry is not excessively complex, the required accuracy is moderate, and the material needs to be flexible rather than structural.

This is exactly the kind of application where additive manufacturing outperforms traditional production methods: low volumes and complete customization. Insoles fit that description perfectly.

For years, there was one problem: FFF technology simply couldn’t do it well.

Affordable printers were slow and inconsistent, while TPU was flexible mostly in marketing materials. In practice, the result was a hard, slippery rubber-like object with very little actual flexibility.

Anyone who tried printing shoes or insoles at home just a few years ago knows exactly what I’m talking about.

But that has changed. And it changed from two directions at once: machines and materials.

The machines have matured. Fast, reliable FFF printers equipped with direct-drive extruders no longer choke on flexible filaments. Continuous printing now makes it possible to produce insoles one after another, day after day, without human intervention.

Plasmisc – an Austrain company using its Feetneeds’ solution, manufactures them in continuous production mode from recyclable foamed TPU, using a honeycomb structure whose hardness can be varied within a single insole. From a soft Shore A 30 to a firm Shore A 90.

A traditional insole simply cannot do that: different stiffness under the heel and under the forefoot, all in a single piece.

The material has caught up with the machines. Today’s TPU is not a single filament but an entire category of materials. Harder 95A grades for support, softer 85A grades for cushioning, and foamed variants that expand after printing into something much closer to foam than plastic.

Filament manufacturers have developed dedicated product lines specifically for footwear and orthotics. And that’s the key point: progress in TPU materials for FFF has done more for adoption than any new printer launch ever could.

Rigid orthotic shells remain the domain of SLS and MJF, most commonly produced from PA11, valued for its strength and biocompatibility. Some labs use resin-based systems. But economics is pushing volume toward FFF, because the difference in machine and material costs is substantial, while the resulting quality is more than sufficient.

The most significant developments, however, are happening in software and automation. This is a textbook IVth Era 3D printing business model: the machine is inexpensive and interchangeable, while the real value lies in software that converts a foot scan into a production-ready file without manual intervention. Clinics print locally. Orthotic labs reduce waiting times from weeks to hours.

And the shoes...

And then there’s the other side of the same coin: the hype surrounding printed shoes.

Here it’s important to be careful, because it’s easy to lump everything together. The most famous example, Adidas 4D and 4DFWD, is neither FFF nor TPU. The midsoles are produced by Carbon using its DLS photopolymer technology, and according to Adidas, the latest generation delivers 23% more cushioning.

That’s an entirely different technological story than my insole from Budapest.

But alongside it, a movement much closer to what I’m describing is growing. Zellerfeld prints entire shoes as a single-piece structure made from a single material, and in 2026 entered a high-profile collaboration with Nike on the Air Max 1000.

Startups such as ARKKY, FORMISM, FUSED, KOOBZ, or PRESQ are pushing printed footwear toward price points comparable to premium running shoes.

The common denominator across this entire wave is the use of single-material construction, most often TPU: recyclable, locally manufactured, and printed on demand. According to industry forecasts, the 3D printed footwear market is expected to reach USD 2.61 billion by 2029, with double-digit annual growth rates.

And now for the most important observation. It sounds obvious, yet it is rarely emphasized.

None of the companies genuinely pushing this market forward are 3D printing companies.

Batz is a footwear company. Adidas is a footwear company. Zellerfeld is a footwear brand that happens to own printers.

The market does not perceive them as additive manufacturing companies, and it never will. For them, the printer is simply a tool, no different from an injection molding machine or a sewing machine.

And that is exactly what technological maturity looks like: the moment FFF starts genuinely transforming an industry, it disappears from view as “3D printing.” What remains is simply the way a footwear company makes its shoes.

Which brings me back to the question I asked at the beginning.

After a decade of promises that this was the machine that would print everything and eliminate factories. After years of ideological battles over who should control the openness of these devices (because truly, nothing intensifies a philosophical debate about open-source software quite like the prospect that it is ultimately being fought over machines that print insoles).

That would be very funny.

Because this entire industry spent years selling a science-fiction revolution. What it got instead was something far more modest—and far more real: a product that people actually want to buy, manufactured locally and tailored to an individual foot.

I’m wearing exactly that product as I write this.

The first few days are strange, just as they told me they would be. But it gets better every day, and something tells me the same is true of this technology.

Post Scriptum

Of course, I don’t actually believe that 3D printers will end up printing only shoes and orthopedic insoles.

They will continue printing drones - and more and more of them. They will print toys. Dental printers will produce teeth, prosthetics, and aligners. They will continue creating jewelry patterns for lost-wax casting and countless other applications.

These machines will stop being devices for everything. Instead, they will develop into sector-specific production platforms, each optimized for a particular application and market segment.

And that is where their greatest strength lies in terms of deployment volume and commercial adoption.

For an entire decade, the industry has been selling the idea that metal 3D printing would finally reach full-scale production as soon as machines became faster, larger, and cheaper.

The machines became faster, larger, and cheaper. And yet you still cannot simply take a metal 3D-printed part and install it in a jet engine.

The reason is both trivial and incredibly difficult to solve. Before anyone allows a printed part to fly, it must be qualified. Qualification in metal additive manufacturing means mountains of physical testing to generate B-basis and C-basis data.

Because, as Dyndrite’s Harshil Goel puts it: the machine is a black box. You put powder in, press a button, and get a part whose properties can only be fully understood after destroying it in a mechanical testing machine. Again and again, hundreds of times.

At its core, the problem is uncertainty.

And in aerospace and defense, uncertainty is paid for with time and money. Jonathan Cohen of Mimo Technik states it plainly:

the biggest barriers to scaling metal additive manufacturing has been the time and cost associated with qualification and materials testing.

That’s it. Everything else is engineering detail.

What is actually changing

America Makes and NCDMM have announced the winners of a $2 million project call funded by the ManTech Office (OSW ManTech). The AIM-4AM (Artificial Intelligence for Material Allowables in Additive Manufacturing) program aims to create an AI-driven framework that identifies and quantifies risk in the current approach to generating material allowables for LPBF-produced 17-4PH stainless steel in the H1025 condition.

The winning team is led by Dyndrite, a software company built around its GPU-powered ACE geometric engine. Along stand Mimo Technik responsible for controlled builds and test coordination, and RTX serving as the transition partner for aerospace and defense applications. Together they will develop a framework that models the relationships between process parameters, material structure, and material properties.

It will then identify which physical tests provide the greatest amount of information.

The practical goal is straightforward: safely reduce the number of required tests and associate every such reduction with a calculated probabilistic risk level. The demonstration will focus on a single material and a single process: LPBF.

The team will generate preliminary qualification datasets for 17-4PH H1025 and evaluate whether the model’s predictions match the actual results of tensile and fatigue testing.

If they do, some physical testing can be replaced by computation. If they do not, that will be valuable information as well.

Here, qualification is created through mathematics and statistics.

Before anyone opens the champagne

Now for some cold water, because without it this article would become a press release itself.

A $2 million budget is not an overwhelming amount of money.

The program focuses on a single grade of steel in a single heat-treatment condition.

At this stage, it is merely a framework demonstrator, far removed from becoming a standard that could be incorporated into aerospace specifications tomorrow.

The path from “we demonstrated that it works for 17-4PH” to “we trust it for a critical component made from a more challenging material” is long and paved with the caution of people whose job is to make sure airplanes do not fall out of the sky.

Goel himself does not hide this reality. He openly states that he fully understands the hesitation surrounding the generation of material allowables through new methodologies and that trust must be established through rigorous validation.

Such honesty is rare in an industry that often promises superpowers.

The verdict

The chaos in the 3D printing industry continues. Bankruptcies, consolidations, and announcements of “breakthroughs” that age faster than the milk we forgot to put back in the refrigerator this morning.

Yet, at the same time, something genuinely important and potentially transformative is happening in the background: someone is finally tackling the least glamorous part of the entire equation - trust, statistics. the ability of a machine to tell us how much confidence we can place in it.

Goel sees an even broader opportunity: accelerating adoption through intelligent parameter selection and by building confidence in the manufacturing process itself. He is right, although the road ahead is longer than this budget might suggest.

For more than a decade, the industry has been trying to win by selling increasingly capable printers, while the real bottleneck has been somewhere else entirely.

It is encouraging to see someone finally taking that bottleneck apart, even if, for now, they have only $2 million and a single piece of steel to work with.

In the end, it is simply mathematics and physics.

Atomic Layer from the Past:

9 years ago Formlabs launched the Fuse 1, its first selective laser sintering (SLS) machine, priced at $9,999 - 20x cheaper than industrial rivals. The printer supported Nylon PA 12/11. Formlabs also unveiled Form Cell, an automated SLA production system with robotic gantry for 24/7 digital factory operations.

Read all:

And 8 years ago, 3D Systems launched two metal printers: DMP Flex 100 for R&D and small, complex parts, and DMP Dental 100 for labs - printing 90 crowns in under four hours as an entry-level solution.

Read all:

News & Gossip:

#1

Formlabs put up a teaser: a wooden crate, a spray-painted logo, reading “Something big is coming,” and a countdown to June 9, 15:00 CEST on YouTube. No name, no specs, no process, no price. “Big” obviously means large format, so my bet is a bigger printer. Most likely SLS-type (mark may words).

#2

UnionTech just booked 120 systems across two deals: 100 SLA machines to Dongguan Fohan, 20 metal printers to Dongguan Huanya. Fohan already runs 1,600+ UnionTech SLA units, so this is a regular topping up his fleet, not a revolution. The real story is the metal push: three SLM machines since 2023, a four-laser model built for shoe molds, and 150 million RMB sunk into a Jinjiang project.

#3

UnionTech orders look big. But in bigger picture they’re a rounding error. China just flipped 3D printing from spec focus to capacity focus. BLT is sinking 3.1 billion RMB into Phase IV plus a billion more on powder. Farsoon wants 3.91 billion. FFF print farms in Jiangsu reportedly hit 20,000 machines, and one buyer grabbed 15,000 Bambu units at once. Service prices crashed to 0.2 RMB per gram. Sounds absurd, but read the insider’s report on this.

#4

Meanwhile in the West, Sandvik just sold its entire AM unit to investment firm Mimir, after two decades in the business. The deal covers metal powders, MIM, and hot isostatic pressing, and it lands barely a year after Sandvik launched a patent-pending MAR 55 powder and signed on with Additive Industries.

Same outfit that dumped its BEAMIT stake in 2024 and “refocused on powders.”

Now it’s refocusing on the exit, booking a 230 million SEK impairment on the way out. CEO calls it positioning for the “next growth phase.”

For someone else, evidently.

#5

Stratasys opened a 200,000-square-foot Americas HQ in Minnetonka. There was official ribbon-cutting, there were the Crumps, and even two members of Congress. ARCH bundles engineering, R&D, and customer collaboration “under one roof,” which mostly means moving teams that already existed into a nicer building.

The press release leans hard on jobs, national security, and a passed ISO 14001/45001 audit. Real estate and ISO certs aren’t a strategy, but with Sandvik exit, a Western company opening new HQ still counts as a win.

#6

Axtra3D also took a bigger 17,000-square-foot site in Vicenza. Company wants to anchor its European operations: materials, engineering, manufacturing, and support under one roof.

The difference from the others in this thread is the demand behind it: 55%+ repeat-customer growth and rising installs since 2021, with Toyota, LPE, and Protolabs on the reference list.

Founder Gianni Zitelli swears it’s “not simply about adding space,” which is exactly what everyone adding space says. But repeat buyers are the one signal that’s hard to fake. Five years in, that’s a company growing into its building, not decorating it.

During a podcast episode, Thomas Pomorski, Director of Additive Manufacturing at Ursa Major, asks an AI agent to write a new hatch strategy and, twenty minutes later, prints it on an EOS machine. This is what industrial metal 3D printing is starting to look like.

Pomorski opens VS Code, logs into an AI agent, and enters a prompt. He asks for a new laser path strategy. One that fills hatch tiles from the outer edge of the part toward the center because it distributes heat more evenly.

The agent reads the slicer’s repository, creates a plan, writes the function, edits the configuration file, and assigns it to a specific part on the build plate.

Pomorski clicks “Run.”

The slicer imports the geometry, automatically generates supports, and slices the build for an EOS machine. Twenty minutes later, that same pattern is being laid down layer by layer inside the printer.

All of this happens live during an episode of the “Additive Snack” - podcast hosted by Fabian Alefeld from EOS.

No editing. No “we prepared this earlier.”

Pomorski runs the additive manufacturing facility at Ursa Major, a company that prints rocket engines for hypersonic systems, solid-fuel propulsion, and space propulsion applications.

For him, this is just a normal Tuesday at work…

The most interesting things are no longer happening in hardware

For more than twenty years, metal additive manufacturing has been driven forward by mechanical engineering. Better lasers, better gas flow, better optics, larger build chambers.

And to give the industry credit, that work has reached a level that would have seemed almost unimaginable a decade ago.

Today, Ursa prints the Hadley engine with roughly 80% of its mass produced on EOS equipment. The same engine has flown hypersonically more than nine times, returned to Earth, undergone inspection, and flown again.

The EOS M450 machines installed at Ursa Major feature four nLight lasers, enhanced gas flow systems, and upgraded optics.

So yes, the hardware is ready.

The problem is that hardware alone is no longer enough. All the fundamental additive manufacturing technologies have already been invented.

And when that happens, competitive advantage shifts to where there is still plenty left to optimize: process preparation, process control, and validation.

Those are software domains.

The anatomy of a single prompt

Let’s break down this demonstration…

The slicer used by Ursa Major is called Polaris and is being developed in collaboration with Dyndrite. At its core are TOML configuration files. These contain all input parameters: layer thickness, angles, top and bottom skin parameters, hatch settings, contours, and more. Engineers modify laser settings simply by editing text files, without touching the underlying code.

The second component is part-name-based segmentation. Every item on the build plate: “sample 1,” “sample 2,” and so on. It receives its own dedicated parameter strategy, automatically assigned through configuration.

One build plate. A dozen different process recipes. Zero manual clicking through individual parts.

So when Pomorski asks the agent for a new hatch strategy, the agent is not “drawing” anything. It is writing a function into the slicer repository and then connecting it to the selected part through a TOML file.

The toolpath strategy becomes a piece of code that can be version-controlled in Git, compared against previous versions, rolled back, and audited.

The process is no longer a collection of settings stored in the head of an operator and scattered across internal network drives. It becomes a formal, programmable description that can be analyzed and reproduced.

A company that controls only the hardware is standing on thin ice. Ursa’s demonstration shows why.

Without a programmable slicer, a request such as “create a new hatch strategy and deploy it to the machine” would have nowhere to land.

AI multiplies the capabilities of those who already understand the process

It would be easy to draw the wrong conclusion from this demonstration, that anyone with a laptop can now print a rocket engine.

Pomorski dismantles that idea himself, and he does it elegantly.

AI is a force multiplier for people who already possess expertise. If you can tell the agent that you care about the thermal history of a part at a 20-degree angle, and explain why, you’ll get excellent suggestions and working code.

But if you’re a junior engineer who doesn’t yet understand which factors influence a thermal profile, you’ll get far less value from the tool. The real advantage emerges when experience meets speed.

And speed is where this becomes powerful.

Ursa’s software development team consists of four people, yet its effective output resembles that of ten or more.

Pomorski describes a two- to threefold increase in the number of experiments and research plans that can be pushed through the machines.

He achieves all this while spending perhaps five percent of his time on software development because his primary job is running the factory, not writing code. Which naturally raises an interesting question: what happens when someone like that finds time for the remaining ninety-five percent?

There is another level of honesty here that I appreciate:

Pomorski prototypes broadly with AI agents, and then the software team rebuilds the solution properly for production use, assigning the model small, precise tasks.

For context, Ursa is working with an older model version that complies with AWS GovCloud requirements, several releases behind what much of the rest of the world currently uses.

Even with that limitation, they are accomplishing things that were out of reach just a year ago. That says a great deal about where this curve is heading.

And all of this eventually shows up in the cost sheet.

Three or four years ago, very few people in the industry would have argued that 3D printing could be cheaper than conventionally producing an engine.

Today, Ursa is targeting a complete hypersonic vehicle at around one million dollars, while custom-built alternatives have historically cost between $20 million and $50 million per unit.

Part of that difference comes from better printers. But increasingly large portion comes from software: fewer failed builds, faster iteration cycles, and a four-person team performing the work of ten.

Why this is good news

First, we’re talking about “a corner of the market” where the West still holds a meaningful advantage.

Consumer FFF belongs to Chinese manufacturers, who control roughly 95% of the market below the $2,500 price point.

Industrial metal additive manufacturing is protected by certification requirements, process control expertise, service relationships. And now increasingly by software.

North America accounts for more than 37% of the global AM market, with defense applications becoming an increasingly powerful driver.

(Velo3D’s recent recovery has been fueled largely by defense-related orders)

So yeah, this is not a segment in decline…

Second, this story has allies:

Just look at that crew...

A company focused on applications. A software company. A metal 3D printer manufacturer. A laser manufacturer.

That said, a degree of skepticism is appropriate.

A successful demonstration on a podcast hosted by EOS, featuring a partner from the same alliance, is still a family gathering - not an independent audit.

And no AI agent changes the fundamental reality that AM remains a complement to injection molding, casting, and machining rather than a replacement for them.

But nobody here is arguing otherwise.

Ursa prints components as replacements for existing platforms quite deliberately because the industry is not yet ready for widespread adoption of double-wall geometries and lattice structures.

(Even though Ursa has already tested such designs successfully)

Maturity means understanding what your tools are capable of today. Hardware in metal additive manufacturing is ready.

This is still the beginning of the curve.

But for this troubled industry, that is a genuinely substantial reason for optimism.

How many times have we already seen the same scenario in the AM industry? A brilliant, innovative technology, enormous potential, promises of major disruption and... bankruptcy, collapse, assets acquired for pennies on the dollar. Markforged has just joined that unfortunate and ever-growing list.

This is the 100th episode of 3DP War Journal. One hundred consecutive weeks without a single break 🥳

I could celebrate with a cake, launch fireworks, and write an emotional motivational post about discipline, persistence, and consistency (and you would applaud me and leave heart emojis 🥰😘😍)

But maybe I’ll do that another time. This topic is too good to waste on fireworks and public ego-stroking.

Because man, just look at these numbers...

Nano Dimension sold Markforged to Stratasys for $42.5 million. Less than two years earlier, it had paid $116 million for the company.

Back in 2021, after going public, Markforged was valued at over $2 billion. Revenue in 2025? Seventy million dollars.

The only consolation is that things turned out better than they did with Desktop Metal. Other than that, it’s a disaster.

The whole story is actually funny in a way. It unfolded like a South American soap opera.

In 2023, everyone was trying to buy everyone else. Nano wanted to launch a hostile takeover of Stratasys and got rejected. Stratasys came close to acquiring Desktop Metal, but shareholders - including Nano itself - stopped the deal. Then Nano went ahead and bought Desktop Metal and added Markforged to the portfolio. Two companies that hated each other.

Now, three years later, Stratasys does not own Desktop Metal, but it does own Markforged, while Nano Dimension is barely hanging on (well, technically it’s alive, but what kind of life is that...)

Overall, my view is that Stratasys scored big this time and seized a once-in-a-lifetime opportunity. Markforged had developed two intriguing technologies - continuous carbon fiber 3D printing (CCF/CFR) and metal filament 3D printing. It had an excellent distribution network. It was the strongest alternative to Stratasys itself. And now it has landed in Stratasys’s hands for next to nothing.

I wrote about it here:

But there’s a completely different story hidden underneath all of this. Well, actually, there are three...

The first is that Markforged invented two genuinely great manufacturing technologies, spent a decade telling the entire industry about them, and never managed to earn enough from them to stand on its own feet.

Continuous carbon fiber printing, followed by metal printing - these were genuinely exciting technologies. Everyone talked about them.

Very few people used them on a daily basis.

(compared to desktop FFF folks, of course)

The second story is that when it was time to invest the money they had raised into growth, scalability, and cost optimization, the company ended up burning cash on lawyers. Not necessarily by choice, but the fact remains.

Markforged wasted a significant amount of money unnecessarily. It was pointless.

And finally, the third story: once the founder left, everything started falling apart.

A classic.

Damn, did Markforged ever really have a chance?

What was under the hood

Markforged was founded in 2013 by Gregory Mark, an MIT engineer who had previously built aerodynamic ballast systems for race cars at Aeromotions. As a result, he knew composites extremely well.

He came up with the idea of automating fiber placement: carbon fiber, fiberglass, Kevlar, and integrating it into a 3D printer.

In January 2014, at SolidWorks World in San Diego, he unveiled the Mark One for $4,999. It was the world’s first printer capable of laying continuous carbon fiber, embedding centimeter-long strands inside nylon. The resulting parts could be stronger than aerospace-grade aluminum.

On a PowerPoint slide, it looked like the end of CNC machining.

Then the company moved into metal.

In 2017, at CES, it introduced ADAM technology. You print using a filament loaded with metal powder and binder, remove the binder, sinter the part in a furnace, and end up with a metal component without needing a laser powder bed fusion machine costing hundreds of thousands of dollars.

Affordable metal manufacturing for anyone with a furnace and a bit of patience.

Both CCF and ADAM worked.

And both promised a revolution that never materialized because each turned out to be too niche to carry that revolution on its own.

Because right next to the promise stood mathematics, ruining the whole party.

Continuous fiber meant expensive materials and long print times.

Metal X meant a furnace and hours of processing, while the finished part still came out inferior to a machined component.

On the other side of the desk sat something cheaper and good enough.

To move from one-off parts to tens of units, a good product is enough. But to move from tens to hundreds or thousands, prices must begin to decline accordingly.

If a perfect 3D-printed part is absurdly expensive, people choose the cheaper option.

The infamous “good enough” principle.

And for most applications, a spool of PA-CF or PETG-CF - a standard filament reinforced with chopped fiber, works perfectly well on a conventional printer at a fraction of the cost.

Yes, it’s weaker. But in most cases, nobody notices the difference.

And when strength and precision truly matter, you go to CNC machining anyway, because it has been delivering those qualities for decades, more cheaply and more reliably.

Markforged built two such technologies simultaneously.

Continuous fiber lost to desktop printing with PETG-CF spools and to CNC machining. Metal X lost to the rest of the workshop.

Both technologies were better. Both were too expensive. And the market, as always, chose “good enough.”

And this is not a problem unique to 3D printing. This is how every manufacturing market works. Paying a premium for a genuine advantage only succeeds where that advantage is truly necessary, and there are always fewer of those situations than investor presentations assume.

Burning cash

It is also worth remembering one uncomfortable fact. Markforged did not invent continuous fiber printing on its own, nor did it hold an uncontested claim to the concept. There was Continuous Composites. There was Anisoprint. And there was that pesky Desktop Metal. Each company took a slightly different path toward the same idea: embedding continuous fiber into a printed structure.

And this is where the second parallel story begins - one that slowly consumed the company from within. Metal X dragged Markforged into a war with Desktop Metal.

Desktop Metal was founded by Ric Fulop, the same investor who provided some of Markforged’s earliest funding in 2013 and sat on its board of directors. He left in 2015, launched his own company, and developed a process strikingly similar to Metal X.

From 2018 onward, the two sides bombarded each other with lawsuits. In the end, the court cleared Markforged of the allegations, but the victory cost years of wasted energy and millions of dollars.

During that time, both companies spent more effort chasing each other than chasing the market. They built patent arsenals while preparing for a metal printing boom that never arrived.

For years, Markforged and Desktop Metal exhausted themselves fighting one another until both began running short of cash.

Then, in 2021, Continuous Composites sued Markforged for patent infringement. One claim survived, the jury sided with Continuous Composites, and Markforged ultimately settled for $25 million, with $18 million due by the end of 2024.

A company that was already laying off employees and moving into a smaller office received that bill at the worst possible moment.

Later, in an attempt to survive, Markforged and Desktop Metal ended up under the same roof at Nano Dimension.

Like two soaked and exhausted dogs sharing the same shelter.

Not long afterward, Desktop Metal entered Chapter 11 bankruptcy proceedings, and now Markforged has been sold to Stratasys.

Now imagine a world where none of this ever happened.

Imagine both companies had spent that money on development, optimization, manufacturing, and lowering costs – literally on anything except lawyers.

Looking back today, tell me: who benefited from all of this besides the lawyers?

What exactly did Markforged or Desktop Metal gain from those isolated courtroom victories?

A fatal change of leadership

And finally, there is one more factor that may, or may not have mattered.

At the end of 2021, just as Markforged was at its peak, Gregory Mark left the company.

Shortly afterward, he founded an AI startup called Backflip, which aims to revolutionize design for 3D printing and 3D scanning.

There is a recurring pattern in the additive manufacturing industry, one that recently became visible again in the Nano Dimension story:

A founder builds a technology. The company grows. Departments appear, funding rounds arrive, and the founder spends more and more time behind the CEO’s desk and less and less time in the laboratory.

Eventually, a crossroads appears.

Either the founder abandons R&D and becomes a full-time executive, or they bring in an experienced manager and return to inventing.

Mark chose a third path.

He left and founded a software startup.

The people who remained were specialists in operations and financial metrics.

On paper, it was a sensible move. By then, Markforged had long outgrown the garage-startup mentality of 2013. In practice, however, a familiar law of the additive manufacturing industry once again came into effect.

Outside executives rarely understand the unique dynamics of the additive manufacturing market, because truly successful CEOs rarely build their careers in 3D printing. As a result, they tend to run these companies according to rules borrowed from other industries.

The outcome is usually the same.

In additive manufacturing, you have to live and breathe the industry every day.

You need to understand that customers buy reliability and cost-effectiveness, not a strength chart from a presentation slide.

You need to know where the real-world value of continuous fiber ends and where the trade-show brochure begins.

The dispute with Continuous Composites, the shrinking workforce, the smaller office, the Nano Dimension acquisition, and the eventual sale to Stratasys - it is impossible to say whether Gregory Mark could have kept the company alive.

What is certain is something else.

After he left, nobody possessed both an inventor’s instinct and a business operator’s touch at the same time.

And in this industry, one rarely works without the other.

And the punch line?

For a decade, Markforged positioned itself as a challenger to Stratasys.

Now, Stratasys bought Markforged for $42.5 million - less than the company generated in revenue during a single year.

Two brilliant manufacturing technologies ended up exactly where good ideas often end up in this industry when they are valued higher than the market is willing to pay for them:

And perhaps that is simply how this industry works.

Markforged leaves behind two manufacturing technologies whose technical value nobody disputes, a distribution network that competitors envied, and a brand recognized by engineers from Boston to Shenzhen.

All of it now belongs to Stratasys for the price of a single funding round.

The technology will survive.
The patents will continue to generate value.
The printers will continue laying fiber.
Metal filament will continue to find its niches.

Only the company that invented it is gone.

And that is how the story ends. With a quiet wire transfer and a new logo on the machine’s enclosure.

Additive manufacturing in a nutshell…

In 2023, Nano Dimension attempted a hostile takeover of Stratasys. It spent millions of dollars on the effort, hired investment banks, and submitted multiple offers.

Stratasys said: no.

Three years later, Nano Dimension is selling Markforged to Stratasys. For $42.5 million.

The same company it had paid $116 million for just a year earlier.

If the 3D printing industry had its own Oscar for Best Ironic Screenplay, the jury wouldn’t need much time to deliberate. But the 3D printing industry doesn’t have Oscars. What it does have is a long history of companies buying too much, too expensively, and too quickly - only to later sell to whoever still had cash and patience left.

Stratasys had both.

Sometimes, all you need to do is wait

To understand what actually happened in this deal, you only need to look at the numbers on both sides of the table.

Stratasys closed Q1 2026 with $237.8 million in cash and zero debt. A year earlier, it completed a $120 million investment round from Fortissimo Capital with a declared objective: acquisitions. It already acquired Nexa3D’s IP portfolio and hardware assets after the company effectively ceased to exist. Markforged is the next move in that sequence - but a far more calculated one.

Markforged is a company founded in 2013 that became the first in the world to commercialize continuous carbon fiber (CCF) printing. That’s the historical part. What matters now is what Stratasys actually gets.

It gets CCF technology and the Digital Forge platform - an integrated ecosystem of hardware, materials, and print management. It gets a foothold in the defense sector, including at least one X7 system installed aboard a US Navy submarine. It gets customer relationships that Stratasys either didn’t have or had only partially.

And it gets all of this for $42.5 million from a company burning roughly $15 million in cash annually and effectively forced to sell.

Nano Dimension retained only the Metal Binder Jetting division from the original Markforged package - the business acquired in 2022 from Sweden’s Digital Metal. Everything else goes to Stratasys. The company that not long ago wanted to control Stratasys is now handing over assets at roughly sixty-three cents on the dollar.

Yoav Zeif, CEO of Stratasys, talks about an “immediate acceleration of revenue growth.” It sounds like every other post-acquisition press release.

But this time, the math is actually on his side: Markforged generated $70 million in revenue in 2025, and Stratasys acquired it - excluding MBJ - for less than that annual revenue figure.

All those left behind…

Now to the real point: what this transaction actually changes. Because this is not really about the numbers. It’s about who will now serve Markforged customers - and how.

For years, Stratasys and Markforged built separate partner networks. Separate reseller lists, separate integrators, separate regional distributors.

Two channels that spent a decade growing in parallel while competing for the same customers in the same verticals: automotive tooling, aerospace, defense. In many geographic markets, especially across Europe and Asia-Pacific, the same partners either currently represent or previously represented both companies simultaneously.

That now becomes the first real post-acquisition problem.

Stratasys has one of the broadest distribution channels in industrial AM. Markforged built its own network around a different sales philosophy: smaller systems and a lower barrier to entry for customers. Integrating those two channels without triggering partner conflict is a task that will take at least two years in the best-case scenario.

In the worst case, part of the Markforged reseller network - especially smaller partners - will simply disappear from the ecosystem because Stratasys will neither have the time nor the incentive to actively maintain them.

For Markforged customers, especially those who built workflows around Digital Forge and CCF materials, the key question is simple: will Stratasys maintain the product roadmap, or will it absorb Markforged and gradually phase it out?

The history of AM acquisitions does not inspire optimism. MakerBot under Stratasys became a textbook example of a brand slowly fading away - intentionally or not. MakerBot under Ultimaker looked more like placing the body directly into the coffin.

This time the segment is different. These are industrial customers, not consumers, and they operate on contracts rather than enthusiasm.

But we’ll only see how this plays out over time.

The real value of this acquisition will become visible in about 18 months. If Stratasys actually integrates the FX10 metal kit into its broader product strategy, and if Digital Forge reaches customers through existing Stratasys contracts, then $42.5 million may turn out to be one of the best acquisitions in AM history.

If the distribution channel fractures and the CCF roadmap gets frozen, then Stratasys simply eliminated a competitor at a discount price.

Meanwhile, Nano Dimension is now essentially a Metal Binder Jetting company without commercially proven technology, without Desktop Metal, and without Markforged.

Its CEO talks about “Phase 3 of the strategic plan.” I don’t know what Phase 3 contains. But I do know that $42.5 million and $15 million in annual savings is not an infinite runway.

The consolidation of the IVth Era of AM is far from over. Assets are simply migrating toward whoever still has cash - and the nerves to wait.

This time, Stratasys won.

Next time, we’ll see who else managed to survive long enough with their hands still in their pockets.

Atomic Layer from the Past:

8 years ago, Ilan Levin announced his resignation as CEO of Stratasys, a position he had held since June 2016. He had succeeded David Reis, the architect behind the merger with Objet and the acquisition of MakerBot. Chairman Elan Jaglom temporarily stepped in as interim CEO, and after an eighteen-month search, Yoav Zeif was appointed as the new CEO — a role he still holds today.

Levin, who had been associated with Objet for 15 years, stepped down amid weak financial results. Q1 2018 brought losses and declining sales. In media statements, he expressed disappointment but predicted that the trend would eventually reverse. In the end, that turnaround came only after his departure.

Read all:

News & Gossip:

#1

Bad timing is a Sinterit specialty lately, and the BIANCO2 launch proves it. The company rolled out a compact CO₂ SLS system with a colorable-material twist, 137 open parameters, and a €47,000 price tag, right into the news cycle that Stratasys swallowing Markforged completely buried. So the headline got eaten before anyone read it…

The machine itself is fine: open parameters for R&D crowds, white and natural powders, Q4 2026 deliveries. The problem is that a company on a cold streak couldn’t even land the launch date. In AM, timing is half the product.

#2

Another day, another metal LPBF box promising to “bridge the gap” between prototyping and production. Mastrex’s MX300 brings dual 500W lasers, a 300 x 300 x 350 mm build volume, the usual alloy menu (titanium, Inconel, cobalt-chrome), and a $185,000 sticker aimed squarely at machine shops priced out of metal AM until now.

The pitch is aerospace, defense, and medical, designed and built in the US, which matters more in 2026 than it used to. First adopter Solomon MFG calls it a competitive edge. Fair enough.

Now it has to survive the market...

#3

Speaking of which - China’s Bright Laser Technologies is quietly running 20 lasers at once. Its production case study prints a full bed of robot arm housings on the BLT-S800, an 800mm-format machine, cutting build time from 119.8 hours to 35.2. That’s a full bed in 35 hours, one part every 58.6 minutes, in AlAM400 aluminum at 120μm layers.

These are BLT’s own figures, so apply the usual skepticism. But the trajectory is hard to ignore: the laser-count arms race is happening, and it isn’t happening in Texas.

#4

ROBOZE and Swiss research institute SUPSI are jointly chasing additively manufactured carbon-carbon and ceramic matrix composites. You know, the materials you reach for when metal alloys give up: hypersonics, fusion systems, anything running hot enough to humble Inconel...

The interesting part isn’t the printing. It’s the downstream thermal conversion that turns a printed shape into a real C-C or CMC part. That process chain, not the printer, decides whether this works.

Press-release quotes is filled with usual “milestones,” “performance,” “reliability,” and “durability,” so temper expectations. Promising chemistry, years from a qualified component.

Worth watching, not worth betting on yet.

#5

WASP, another Italian AM veteran, is letting users poke at the beta of its revised app, a browser tool for parametric modeling. You can define a shape with adjustable values, change one number, the whole model updates. No local install, no CAD license.

The 2.1.0 Beta adds a free-form tool, manual control-point editing, and a base generator, plus a faster, rebuilt layout.

It’s a sensible niche play. Lower the barrier to designing printable geometry and you sell more printers.

#6

And finally, Decathlon’s running brand Kiprun launched its first 3D printed shoe. A knit upper on an HP Multi Jet Fusion midsole, printed in a proprietary TPA, with a coral-red variable-density lattice tuned for 75% energy return versus the 50-65% of normal EVA foam.

Developed in under six months in Shenzhen, sold only in select Chinese stores for about $250.

Nike and Adidas have already been here, so the novelty is thin. And nobody will call Decathlon a 3D printing company for this. They’ll call it a shoe. The midsole is just supply chain, doing its job quietly, exactly as AM should.

An engineer launches a thermal simulation. Monday morning, laptop open, “run” clicked, and then… waiting begins. Three days later, the result arrives. If the outcome says “your print will fail,” only then does the engineer start changing parameters and running the simulation again. Another three days.

That has been the reality of industrial AM for two decades. And it is precisely why most DED, WAAM, and large-format polymer integrators barely use thermomechanical simulation at all. Too expensive and too slow to fit into a daily workflow.

Aibuild has just introduced FETS - Finite Element Thermomechanical Simulation. The tool simulates stresses, distortion, thermal, and thermomechanical effects across five AM processes: DED, WAAM, AFSD (friction stir), FGF, and FFF. It works for both metals and polymers. Cloud-based, accelerated with NVIDIA CUDA-X GPUs, and validated by NIAR (National Institute for Aviation Research) at Wichita State University.

Aibuild claims the solution is up to 10,000 times faster than conventional FEA. That it achieves an average prediction accuracy of 96.8% on real-world 17-4PH stainless steel WAAM prints. That results which previously took days are now available in seconds or minutes on standard hardware, without multi-million-dollar HPC investments.

Aibuild’s CSO Guy Brown summarizes the issue directly:

The numbers sound impressive. “10,000x faster” feels like a classic sales presentation slogan. But once you separate the marketing from the engineering, it becomes clear that FETS is solving a very specific problem in a segment that, until now, largely had no simulation capabilities at all.

Aibuild, FETS, NIAR

Aibuild is a UK-based software company that has spent years delivering CAM solutions for robotic DED and large-format polymer systems. Their customers are mainly integrators: companies building custom AM systems for highly specialized applications, such as turbine components transported once a year on dedicated heavy-duty trucks.

This is an environment where every printer is different, and every part is often unique. Desktop DfAM rules do not translate here at all.

A few months ago, Aibuild expanded its offering with Aibuild OS - an AI layer sitting across the entire workflow. FETS is the next step in that same direction.

So what exactly does FETS do?

Four things simultaneously. It simulates thermal flow throughout the geometry during deposition. It models the material’s thermomechanical response to that heat flow. It predicts distortion - where and by how much geometry will deviate from design intent - and residual stresses, meaning where the material will attempt to deform after cooling.

All based on the exact toolpath that the machine actually executes on the shop floor.

FETS is CAM-agnostic. In practical terms, that means if you generate a toolpath in third-party software - for example, an integrator’s proprietary platform that you do not want to replace - Aibuild can still import the path and run the simulation. No migration to their CAM ecosystem required.

The NIAR validation is important and deserves separate commentary. NIAR is a research institute deeply involved with the aerospace industry. Validation from them is not a symbolic gesture. They measure real prints on real machines.

In the case of FETS, on 17-4PH WAAM samples, the tool achieved 96.8% average prediction accuracy.

That number is strong. But underneath it sits one narrow application, one laboratory, and one material.

Generalizing the same accuracy to titanium, Inconel, or high-performance polymers will require additional validation rounds.

Aibuild already had its own thermal analysis tool integrated into the CAM environment. That tool is fast and can generate thermal envelopes within minutes, but it relies on simplifications that work primarily for low-conductivity materials, such as carbon-fiber-reinforced polymers. For steel, titanium, and geometries requiring full physics fidelity, those simplifications are insufficient.

FETS is the full-rigour version: complete FEM simulation without approximation shortcuts, based on the actual thermal history of every deposited layer.

For clarity, FEA simulation in AM is nothing new. Ansys, Altair, and Abaqus have offered FEA tools for years, some with GPU acceleration.

What do they all have in common? Simulation operates separately from CAM. It is a different workflow, different software, often even a separate engineering department. Every change in toolpath requires manual remeshing and another simulation iteration.

What FETS does differently is integration.

Simulation exists directly inside the CAM workflow. Change the toolpath, and the simulation updates automatically. No exporting data into separate software. No waiting for a dedicated FEA team.

That matters because most Aibuild customers do not have FEA departments.

These are DED integrators building custom robotic systems, large-format polymer robot integrators, aerospace suppliers producing one-off components. They are not going to purchase million-dollar Abaqus licenses just to simulate a single turbine once a year.

For them, FEA simulation was inaccessible not because of software pricing, but because the workflow itself required specialized expertise.

What is really happening here

Back to the original question: is FETS truly a breakthrough tool, as Aibuild claims?

The short answer is yes.

The long answer depends on how we define “breakthrough.”

“10,000x faster” is marketing-important, but from an engineering perspective, the real question is: faster compared to what?

Conventional FEA on a 15-year-old laptop? Fully nonlinear Abaqus simulations with manual remeshing after every layer? GPU-accelerated Ansys workflows?

Aibuild most likely benchmarked itself against the slowest of these scenarios, which is standard practice in the software vendor space. It creates an impressive number. It creates a clickable headline.

The 96.8% accuracy figure is legitimate, but still based on one material and one process. Aerospace companies will appreciate that Aibuild performed validation at all. Aerospace companies will also ask whether the same accuracy holds for Ti-6Al-4V and Inconel 718.

The answer is simple: we will see in a year.

The real value of FETS lies elsewhere.

Until now, Aibuild was fundamentally a CAM company - software sitting between the designer and the machine, translating geometry into toolpaths. A narrow but well-defined role.

Adding FEA simulation into the same workflow changes the category entirely. Now Aibuild also controls process validation, failure prediction, and eventually, qualification workflows.

For a DED integrator that only recently evolved from building welding robots into building AM systems, this means they can offer customers simulation capabilities without hiring an FEA department.

Their aerospace customer can print parts without building in-house simulation expertise.

The customer’s customer receives a validated process, not just a machine.

There is, however, one caveat.

The more workflow layers controlled by a single vendor, the harder it becomes to leave that ecosystem. Aibuild OS plus Aibuild CAM plus FETS creates an increasingly deep dependency for customers.

That is not unique to Aibuild. It is the logic of the entire AM software segment. Integrators should approach it with open eyes.

At the end of this launch, one thing remains clear:

Everything else - the performance multipliers and benchmark numbers - remains marketing detail until independently validated on additional materials.

But in my view, this is one of the most significant software moves in large-format AM in quite some time.

Joris Peels is one of the best AM journalists and market analysts in the world. I can’t say he’s the best, because I’m still here too (hehe 😜).

At the end of 2023, he published a series of articles about the condition of the industrial AM sector which, in hindsight, turned out to be prophetic. A few months later, in the summer of 2024, the market collapsed. Joris, Alex Huckstepp, and I were describing it before it became real. But it was Joris who started that uncomfortable, though truthful narrative.

Now Joris is back with another series of bitter pills to swallow, which - just like in 2023 - CEOs of the world’s leading AM companies would probably prefer to ignore.

But ignoring facts will not protect you from them eventually catching up with you.

3DPrint.com has published four articles from Peels’ new series, The Additive Chicken Coop. Debunking, cold, and at times brutally funny.

The series portrays the AM industry as a self-contained system, dependent on a narrow customer base and increasingly incapable of inventing anything new beyond what already exists.

The third part is weird. Joris dedicates it entirely to bananas. Seriously, bananas. But the fourth part ties everything together and is, in my opinion, the most important analysis of the AM industry this year.

Below is my summary and commentary. And at the end is what Joris didn’t write: how to get out of this. How to escape the chicken coop.

The Additive Chicken Coop - the coop that locked itself

Peels begins with the metal 3D printing sector. The first part describes how industrial LPBF systems spent years functioning as million-dollar Petri dishes. Machines where every parameter could be changed, but which were not optimized for producing anything specific. They were bought mainly by universities. For six years, everyone studied Inconel properties. Everyone in their own way. Everyone starting from scratch.

In the following sections, Peels moves toward a market observation: the AM industry is obsessed with itself. Constantly looking at its own tools instead of customer problems. Costs are rising, opportunities are shrinking, and companies are entering the same markets with the same machines.

Then comes the third part.

Peels writes about bananas. Literally.

The whole thing revolves around Cavendish bananas, Panama Disease Tropical Race 4, Dole and Chiquita, clones, and monocultures. It feels slightly surreal to read on a 3D printing website, but the meaning becomes clear in the fourth part.

Because that’s where Joris delivers the diagnosis.

The metal AM industry is doing exactly what the global banana industry did: most companies focus on the same sectors - aerospace, defense, medical - offering similar machines and depending on the same funding sources. One regulatory mistake, one high-profile implant incident, one Pentagon budget shift, and the entire market starts wobbling.

Peels reminds readers of the collapse of metal dental printing: everyone was printing the same crowns for the same customers. The market shifted toward zirconia, and companies suddenly discovered they had been standing on very short legs.

On the desktop side, the situation is different, but not better. Everyone is trying to copy Bambu Lab. That’s the only idea.

Peels sees no serious player thinking about anything else: cheaper machines, more durable systems, products designed for completely different applications. Strategic cloning packaged as innovation.

And where are the untouched markets? Specialized services for nuclear energy, maritime infrastructure, subsea industries. No simple CAD platform for children beyond Tinkercad. No complete plastic-to-filament recycling solutions. No platform where you can order large-format metal printing online.

Markets worth billions of dollars. With almost no competition. Nobody is looking there.

Peels’ thesis is simple and unpleasant: the AM industry is running toward a cliff like a herd of lemmings.

Strategic replication and obsession with the same narrow customer groups have turned it into a fragile system. And fragile systems break under the first serious obstacle.

Fine. Joris is right. So what now?

How to escape the chicken coop

Peels delivers the diagnosis. The cure (at least for now) he does not prescribe. So I decided that part is my job.

Software first. That is the only strategy that works long term in AM.

Formlabs survived II Era and III Era of AM in good condition because it built an ecosystem: hardware, firmware, materials, and a closed, optimized workflow all in one.

Bambu Lab built an entire platform, and the printer is merely its interface. MakerWorld is the foundation of user loyalty, not just a website with free models.

Companies controlling only the hardware layer today are in a bad position. Hardware margins are shrinking because Chinese factories can manufacture the same thing faster and cheaper. The real moat is software, data, certifications, and customer trust. That is also the only space where European and American companies still have a chance for long-term advantage.

Joris Peels says something even more important: go where there is no crowd. I wrote exactly the same thing two years ago. It’s exactly what I say during every meeting with my resellers in Central Europe.

Go where nobody expects to see a 3D printer.

Nuclear energy. Maritime. Subsea. Repair of metal energy infrastructure. Markets where trust is currency, margins are different, and entry barriers come from specialized expertise, not just capital. Where one properly implemented project builds a reputation for ten years. Where a “supplier” quickly becomes a “strategic partner.”

There are also other places nobody takes seriously, but which are quietly doing something genuinely interesting.

Desktop FFF and orthopedic sole manufacturing. Sounds unimpressive. Several European companies are printing insoles and soles on demand using farms of cheap FFF printers, with short production runs, full customization, and zero warehousing. On-demand 3D-printed footwear is one of those unexpected directions worth watching closely. Small, agile, profitable.

And then there are those stupid colorful “toys” - snakes, dragon eggs, fidgets. The things every advanced 3D printer user looks at with disgust and contempt.

EVERYONE - from open-source fanatics to engineers optimizing laser paths in metal PBF. All of them get equally pissed off that these idiots make so much money from it while other idiots actually buy it.

Meanwhile, the big, “serious” AM industries should be learning from them. These people ask very simple questions: what does the customer need, when do they need it, and how much will they pay?

They do not debate whether 3D printing will replace injection molding. They simply print and make money.

The Straight Edge 3D Printing Manifesto

In May 2024, in an article published - nomen omen - on 3DPrint.com, I wrote three slogans that I believe are the answer to Peels’ chicken coop.

Be true – Be different – Share and spread.

In case you haven’t noticed, since day one these slogans are part of the logo of The 3D Printing Journal.

• Be true. Stop selling promises the technology still cannot deliver. Talk about what genuinely works, for whom, and under what conditions.

• Be different. Doing the same thing as everyone else stopped being enough when The III Era of AM ended. Look for markets where your expertise is unique and competition is practically nonexistent.

• Share and spread. Knowledge drives the industry faster than capital. Companies that share what they have learned build trust that pays dividends for years.

The AM chicken coop is comfortable. Familiar. Full of people with the same machines and the same customers. It’s easy to stay there.

Easy enough not to notice when the doors close.

Escape the chicken coop. Why? Because the coop is the place where you die...

Three weeks ago, I wrote about how the stock market carousel in the 3D printing industry keeps spinning regardless of what happens to revenues and profits.

Farsoon valued at roughly fourteen times more than 3D Systems. Velo3D with $45 million in revenue and a valuation above $400 million.

The market is pricing narratives, not fundamentals - and so far, nothing suggests it is ready to break that habit.

This week, another major player officially entered the stage. And not just any player - Creality.

The company officially launched its IPO on the Hong Kong Stock Exchange. That meant disclosing all of its financial documents. So let’s start with the numbers, because they tell the whole story...

Creality generated revenue of:

• RMB 1.3 billion ($187 million) in 2022

• RMB 1.9 billion ($262 million) in 2023

• RMB 2.3 billion ($318 million) in 2024.

In 2025, revenue climbed further to $453 million, representing annual growth of around 20–30%.

At the same time, Creality reported net profit of:

• RMB 104 million ($14.5 million) in 2022

• RMB 129 million ($18 million) in 2023

• RMB 88.66 million ($12.3 million) in 2024.

Despite rapidly growing revenue, profits kept shrinking. Then came 2025.

In 2025, the company posted a net loss of RMB 182.4 million. Operating cash flow also turned negative.

Now, to understand why Creality is losing money, you need to understand what Bambu Lab did to the market.

In May 2022, Bambu Lab launched the X1, bringing high-speed and multicolor printing to the consumer market. The result? In 2025, Bambu Lab overtook longtime leader Creality to become the global leader in entry-level 3D printer shipments.

Creality - the company that defined affordable 3D printing for years - became number two. In 2024, Bambu Lab shipped 1.2 million printers, while Creality shipped around 700,000 units.

• Bambu Lab: RMB 5.5 billion in revenue, RMB 2 billion in profit.

• Creality: RMB 3.1 billion in revenue, RMB 182 million in losses.

Creality’s revenue growth in the 3D printer segment over recent years came mainly from higher selling prices, not volume. Between 2022 and 2025, the average price of a Creality printer increased from RMB 1,306 to RMB 2,404, while unit sales declined from 842,000 to 742,000 units. In other words: the company is selling fewer printers at higher prices, because the cheaper alternative disappeared from the market, displaced by a better one.

Against this financial backdrop, Creality is going public in Hong Kong. And this is where things get really interesting.

The company is selling only 15.73% of its shares, with 84.27% remaining in private hands.

Based on the IPO price and total share count, the company’s valuation comes to approximately $7.13 billion 😲

That valuation would be comparable to the combined market value of two other Chinese metal AM companies: Farsoon and BLT.

What are investors buying for that money? Hope.

If those projections prove accurate, Creality - as one of the largest players in the segment - could eventually be worth much more. This is the classic mechanism I described earlier in relation to Farsoon and Velo3D: the market is buying the future, not the present.

And finally - is the IPO an attractive financing option or a necessity?

Looking at the numbers, the answer seems fairly obvious: yes, the IPO was necessary.

In 2025, operating cash flow turned negative, with an outflow of RMB 64 million ($9.3 million). That means the company was no longer generating enough operational cash to fully cover its ongoing needs.

A company with negative operating cash flow, rising inventories (inventories reached RMB 634 million ($93.3 million) at the end of 2025) and a net loss exceeding RMB 180 million.

Additionally, the agreement with Series A investors required the company to complete an IPO by December 31, 2025. Failure to meet that deadline would have allowed investors to demand a buyback from the controlling shareholders. A third agreement, signed in July 2025, extended the deadline by six months - to June 30, 2026.

So yeah, for Creality, this is the last call…

Atomic Layer from the Past:

Once again, today we have two:

14 years ago, at the RAPID Conference in Atlanta, Objet Geometries unveiled the Objet30 Pro, based on proprietary PolyJet technology. It offered seven materials, including transparent and high-temperature-resistant options.

Founded by Gothait, Bonen, and Miller, at that time the Israeli company had already shipped over 3,378 industrial systems. Priced at $19,900, the Objet30 Pro targeted small businesses. The launch came just before Objet’s formal merger with Stratasys, finalized in December 2012. Thus, the Objet30 Pro became one of the last products under the independent Objet brand.

Read all:

Polish startup Zortrax unveiled the M200, a desktop FDM printer that would revolutionize the industry. Unlike open, cable-filled machines made of wood or cheap metal, the M200 featured a professional industrial design and solid components. Its revolutionary Z-SUITE software was closed, blocking parameter changes like temperature and forcing use of Zortrax’s own filaments. This seemingly suicidal strategy succeeded, guaranteeing simplicity, high quality, and repeatability.

Priced at just 1899, the M200 became a desktop alternative to industrial systems like Stratasys. A highly professional Kickstarter campaign raised $179,471 from 144 backers, launching an 11-year market-leading run in Europe.

Read all:

News & Gossip:

#1

Zellerfeld has launched a new 3D printed sneaker, the OD Easy PZ, developed with OverDose - a brand founded by former NBA star player Baron Davis. The post-workout recovery shoe features a monolithic, laceless design and is made from recyclable TPU.

Customers scan their feet via smartphone and order made-to-fit pairs for $199. Available in multiple colorways, the shoe embodies OverDose’s “analog to AI” ethos.

#2

GoEngineer has acquired SKA, Latin America’s leading SOLIDWORKS reseller. From now on, it will operate as a GoEngineer company. Founded in Brazil in 1989, SKA serves thousands of customers in manufacturing, aerospace, medical, and industrial sectors, with expertise in MES, machining automation, and additive manufacturing.

This acquisition follows GoEngineer’s 2025 expansion into Canada and marks its first presence in Latin America and the Southern Hemisphere.

#3

14 Trees and Tvasta have launched Cedar, a large-format gantry-style construction 3D printer for remote environments. Tvasta, an Indian automation firm, partnered with 14 Trees - a joint venture including Holcim and Amazon’s Climate Pledge Fund.

Cedar uses AI for material characterization and works with regular concrete, lowering adoption barriers. It offers a 240 m² build area, 10 m height, and pumps up to 5 m³/h at 60 bar.

The collaboration creates a new competition for COBOD in the growing construction 3D printing market.

#4

Lynxter has launched SIL-004, a liquid silicone for direct 3D printing in food-processing environments. It meets FDA CFR 21 177.2600 standards for repeated food contact and is free from BPA and PFAS - claimed as the world’s first certified 3D printable silicone.

Operating from -50°C to 250°C with 6.12 MPa tensile strength, SIL-004 targets custom seals, molds, and conveyor components. Compatible with Lynxter’s S300X and S600D printers, it produces parts in hours without tooling.

#5

Phrozen has previewed the Mighty Revo MAX, a large-format resin 3D printer featuring a 14-inch 16K LCD for high detail on bigger models. It includes a 2,000 ml resin tank for extended printing, dual temperature control for viscous resins or cool environments, and a tool-free quick-change foil. A leak-proof base prevents resin damage, while an integrated 1080p camera with AI monitoring detects print errors.

Phrozen has not yet announced build volume, pricing, or release date.

#6

And finally… Ultimaker!

UltiMaker has launched the Factor 4 Plus, a new variant of its industrial desktop 3D printer. Targeting manufacturing and defense users, it delivers up to twice the speed of the Factor 4. A reinforced gantry reduces vibrations, while the Cheetah motion planner smooths directional changes.

New AA+ and CC+ high-throughput print heads support materials like PLA, ABS, and PPS-CF. Key feature TRACE (Technical Reporting And Certification Engine) logs print parameters and generates quality reports for every part.

A month ago I wrote about Bambu Lab introducing the Color Mixing feature in Bambu Studio - the ability to mix filaments directly inside the slicer. Now a few words about what the community is actually doing with this feature in practice - and it turns out they are doing quite a lot.

Color Mixing allows users to define a filament in Bambu Studio as a blend of two or three other filaments in specific proportions. Instead of switching between colors, the printer literally mixes them inside a single nozzle - and that blended material is what gets deposited onto the print bed.

The result is a color no manufacturer ever had to produce.

To achieve the best results, it helps to use predefined filament sets inspired by the classic CMYK palette from the printing industry. In the case of Bambu Lab, the company offers sets expanded with white as an additional color (Cyan, Magenta, Yellow, Kontour Black, and White).

To use this feature, you need either an X2D, H2D, or H2C printer equipped with the Vortek system, or two AMS units connected together.

One MakerWorld user - VedicFutura, has probably conducted the most comprehensive tests of this feature so far. The result of that work is a catalog containing exactly 292 unique printable colors.

The method was simple, but time-consuming. The author generated blends in groups of a dozen or so per file, printed them on a Bambu Lab H2C at a 0.12 mm layer height, and labeled each swatch with the exact percentage composition of its components.

Go to MakerWorld

The color names - and this is something I really liked - were generated using Google’s Gemini. The result? A palette that sounds like something out of a professional paint studio catalog: not “70% magenta, 20% yellow,” but names like warm sienna or dusty coral.

What can you do with it? Print the catalog, hang it next to your printer, and treat it like a paint reference guide.

If you paint 1:6 scale figures and are searching for the right skin tone, instead of guessing you simply grab the swatch book and recreate the formula one-to-one. It sounds simple because it is simple. And that is exactly why it works.

It is also worth pausing on what turned out unexpectedly well in these tests: skin tones. The author himself admits he was surprised - blends of magenta, yellow, and white in different ratios produce an impressively wide spectrum, ranging from light, almost porcelain-like tones to warm browns and deeper, darker shades.

This is an area that has traditionally been problematic in 3D printing - either a plastic-looking beige straight out of the box, or manual painting after printing. Now there is a third option.

That said, not everything is perfect. The author openly points out two things users need to keep in mind.

First, Bambu Studio “lies” in its color preview - what you see on screen does not match what actually comes out of the printer. The screen renders something that looks like a mathematical interpolation, but real filament mixing inside the nozzle behaves differently. Bambu Lab itself mentioned this in the update notes. Without printing a sample, you cannot be certain what result you will get.

Second, many blends show visible banding - stripes caused by uneven color blending between layers. In close-up prints and on flat, uniform surfaces, this can be noticeable. On figures viewed from a distance, however, it is far less of an issue.

There is also a technical limitation worth knowing about: Bambu Studio allows a maximum of 32 colors or blends in a single 3MF file. The author worked around this limit by splitting the catalog into several separate files, each containing 22-26 colors. Slightly inconvenient, but manageable - and still better than having no catalog at all.

Overall, this is quite an interesting direction for desktop FFF 3D printing - especially when you remember what things looked like 10–15 years ago. Back then, this would have been considered a major revolution. Today, it feels more like an obvious development that quietly passes by without much attention.

But it is still intriguing.

For the past few months, white-collar workers have been in a state of deep panic. Everyone is staring at reports of yet another round of layoffs in Silicon Valley and hundreds of other corporations scattered across the world. Everyone’s asking themselves the same question: does my profession even still exist?

AI is taking jobs from good people. It optimizes processes, simplifies workflows, eliminates corporate bureaucracy. It cuts costs, restructures organizations, improves profitability.

It increases value (to shareholders).

Except most of that is either false or only half true.

And none of it is actually new. We’ve already seen this before - dozens, if not hundreds, of times. The exact same cycle happened in 3D printing twice.

So before Skynet takes over the world, it first has to politely pass through the Gartner hype curve. And right now, it hasn’t even reached the first truly critical stage.

I remember 2013. Do you?

Magazines were writing that 3D printing would change everything. That everyone would have a printer at home and print their own shoes, kidneys, cars or rockets that will take to space. The stocks of Stratasys and 3D Systems were breaking records. Bre Pettis was the second Steve Jobs coming.

And then came the great disappointment.

But not for long, because just around the corner another crew was already waiting with another bubble to inflate.

This time „3D printing” was replaced by „Additive Manufacturing”. Basically the same thing - but „pretty serious” this time. Desktop Metal, Nexa3D, Velo3D, Shapeways (the US one), Carbon, and others promised an industrial revolution the world had never seen before.

And the world never saw it.

Just like with the first wave, very little of those promises actually materialized.

3D printing / Additive Manufacturing has become a disappointment, people with money turned away from it, and today companies with „3D” in their name are avoided like a single mother with three on a dating site.

And now AI is following the exact same path…

What will kill the first wave of AI hype is tokenmaxxing

Tokenmaxxing is one of the most grotesque inventions to come out of Silicon Valley. Companies started measuring employee productivity by the number of tokens they “burned” through language models. The more tokens you consumed, the more productive you were considered.

It sounds like a joke. It’s not a joke.

The result is that employees began running multiple AI agents in parallel, stretching prompts, automating things - not because any of it was useful, but because they wanted to win the internal token rankings.

It’s a bit like measuring a chef’s efficiency by the number of liters of water used. You can outperform everyone else simply by leaving the tap running during the meeting.

And this is where the more cynical side of the story appears. Token budgets do not grow, they do not accumulate value, they do not appear in salary negotiations. Tokenmaxxing is hustle culture degenerated into absurdity - a race for the appearance of productivity, not actual productivity.

At the same time, analysts are starting to do the math. It turns out that in many cases it’s still cheaper to hire a junior developer to write lines of code than to spend money on the tokens required for a model to do it.

Oh - and in both cases, a senior engineer still has to check whether it actually works.

But that’s not something tech bros want to talk about today. For now, everyone is just pushing tokens.

So tell me honestly: in an environment where your value is measured by the number of tokens purchased from OpenAI, do you really feel secure?

Because if compute is doing the actual work, and your role is simply “coordinating compute,” then the question of how many coordinators are actually needed becomes very uncomfortable.

This is exactly what the peak of hype looks like.

And I have good news and bad news for you.

• The bad news: AI is not going away.

• The good news: what you are seeing right now is not the true picture of this technology. The AI sector is moving toward what Gartner calls the Peak of Inflated Expectations.

And do you know what comes after the peak?

Trough of Disillusionment

The Gartner Hype Cycle is one of the most useful tools for reading technological history.

The pattern is always the same: a technology appears, the media falls in love with it, investors pour money into it, everyone says it’s the end of the old world and the beginning of a new one.

Then reality arrives. The bubble bursts. Headlines appear: “technology X has failed,” “sector Y was a scam,” “Z is nothing more than a useless toy.”

But behind the scenes, the technology quietly matures, finds applications nobody predicted, and eventually becomes an industry.

The internet in 2000. Blockchain in 2017. 3D printing in 2013 and additive manufacturing in 2020. Now generative AI.

I will never say AI is a scam. I’m saying you are witnessing the peak of inflated expectations. And right after that peak, people become especially vulnerable to making terrible career decisions.

Either you throw yourself into the chaos and burn enormous amounts of energy on something that, a year from now, will look completely different.

Or - and this is the option nobody talks about out loud - you look at the other side of the same curve and ask: what’s waiting there, beyond the valley?

The answer is Additive Manufacturing

AM went through this curve honestly. At the height of the hype, we were supposedly going to print guns, cars, houses, even internal organs for transplantation.

Basically, we were supposed to print everything.

Then the valley came. Companies either consolidated or went bankrupt. (mostly the latter)

But technologically, the work quietly continued in the background. Development never stopped.

Today, additive manufacturing is entering the slope of enlightenment - the phase where a technology begins delivering what it originally promised, just in a different place and at a different pace than everyone expected.

Dentistry, where 3D-printed orthodontic aligners have reached cost parity with CNC manufacturing while simultaneously offering a level of personalization impossible to achieve through traditional methods.

Medical implants, prosthetics, orthotics, surgical planning models tailored to individual patients.

Automotive, which has used 3D printing practically since the beginning. Rail and maritime industries. Aerospace and space technology, where printed engine components are now standard and nobody even asks anymore whether it makes sense.

Defense, where the ability to print spare parts in the field - without supply chains, without warehouses, without waiting for a container from Shenzhen - has strategic significance.

I’m saying this to highlight the fundamental difference between AM and AI at this particular moment in history.

AI primarily produces digital content: text, code, images, analysis. Important things - but from a market perspective - easily configurable, easily replicated, and difficult to price.

Additive manufacturing, meanwhile, produces physical things.

You can touch AM

You cannot “generate” a dental implant with a single prompt. You cannot type into ChatGPT: “create a defense drone component with 1200 MPa tensile strength and deliver it to a base in Romania within 48 hours.”

That requires expertise, machines, materials, processes - and, importantly - people who understand how all of it actually works.

And this is where we arrive at the core of the issue.

AM skills are rare.

Not because the industry is closed off, but because the intersection of engineering knowledge, additive process expertise, materials science, and design-for-additive-manufacturing capabilities is an extremely narrow path.

And there’s one thing you cannot ignore in this equation:

Generative design algorithmically creating geometries optimized for 3D printing is one of the most beautiful examples of synergy between AI and physical manufacturing.

Topology optimization supported by machine learning algorithms produces parts that humans would never be able to design manually. AI-assisted print-process simulations reduce validation time and costs by orders of magnitude.

In other words: if you understand AM, AI becomes a tool for you. If you only understand AI all you’re left with are tokens.

There’s another argument that rarely appears in discussions about the future of work, yet is becoming absolutely critical right now.

The pandemic showed how fragile global supply chains really are. The ability to manufacture locally, on demand, without warehousing and without dependence on factories in the Far East, is a strategic advantage - not just a business advantage.

AM enables exactly that.

A digital design can be transmitted in seconds. Printed locally. Delivered locally. No ship, no container, no six-week wait.

If you have AM expertise, you can become the operator of this “factory inside a laptop.” And that factory cannot be replaced by a prompt.

The Gartner curve is not just a chart for analysts. It’s a career map, if you know how to read it.

Right now, you are witnessing the peak of AI hype.

In one, two, three years - the valley of disappointment. Layoffs at companies that rebuilt entire business models around generative AI without solid foundations. A correction of expectations. New questions about what AI actually can and cannot do.

And AM?

AM is already on the other side of that valley. It is entering the plateau of productivity. The technology has matured. Machine and material costs are falling. Industrial applications are documented and scalable. Skilled people are scarce.

Anyone entering AM today is entering at the beginning of a growth phase.

Not at the top of a bubble. At the beginning of something that is finally starting to work.

I’m not saying you should ignore AI. I’m saying you should stop being afraid and start thinking strategically.

And it is precisely this industry that the coming decades will need.

Your tokens can be taken away from you. Your expertise cannot.

3D Systems has released its financial results for the first quarter of 2026. At first glance, they may not look particularly spectacular, but as is often the case in the AM industry - the devil is in the details.

Revenue reached $95.5 million, representing 1% year-over-year increase, which is hardly impressive on its own. But once we account for the divestiture of several key software businesses in 2025 - Geomagic, 3DXpert, and Oqton - the picture changes dramatically.

On an organic basis, revenue grew by 11%. And that is not a difference that can be dismissed lightly.

The company continues to distance itself from assets that failed to generate sufficient value and is focusing on businesses that make strategic sense.

The Healthcare segment grew 21% year-over-year, reaching $50.1 million - and for the first time in the company’s history, it approached the scale of the Industrial segment. The strongest growth drivers were Dental and Med Tech, both posting growth rates above 20%.

The Industrial segment declined by 15%, although after adjusting for divestitures it showed modest 2% growth. Aerospace and defense remain the largest and one of the fastest-growing subsegments within Industrial.

Profitability also showed clear improvement. Adjusted EBITDA turned positive for the first time in years, reaching $2.1 million compared to negative $23.9 million a year ago.

Net loss narrowed dramatically: from -$37 million down to just -$4.4 million.

It may be too early to call it a breakthrough, but it is equally difficult to ignore the scale of improvement achieved within a single year.

It is also worth mentioning an event that illustrates where 3D Systems sees future growth opportunities. The new Cadillac Formula 1 team - the newest entrant in the FIA championship - has deployed seven SLA systems from 3D Systems. Formula 1 remains the ultimate laboratory for the automotive industry and, at the same time, a global showcase for advanced manufacturing technologies.

Still, this mood of cautious optimism must be viewed within a broader context that leaves little room for euphoria.

Guidance for Q2 2026 projects revenue in the range of $93-95 million and a return of Adjusted EBITDA to negative territory: between negative $4 million and negative $2 million.

In other words, Q1 was seasonally strong, and the company itself warns that this result should not be interpreted as a straight-line trajectory forward. Management pointed to global logistical disruptions - including war in the Middle East - as a risk factor for supply chains in the coming months.

The question that has hovered over the AM industry for years remains unanswered: when will revenue growth become sustainable enough to justify reinvestment at a scale that enables genuine expansion?

3D Systems still has cash - $86.5 million at the end of March - and is expanding its Littleton facility by 80,000 square feet to support production of metal aerospace components.

In the 3D printing industry, every quarter without a creeping crisis counts as a success. Q1 2026 was something more - the first moment in quite some time when it became possible to say that the company may have finally found a direction that genuinely works.

This is not yet a story of triumph, but it is no longer merely a story of survival either.

Atomic Layer from the Past:

13 years ago, Formlabs launched PreForm software for its Form 1 SLA printer. Unlike costly industrial alternatives, PreForm simplified resin printing with intuitive supports and orientation tools. This milestone made desktop SLA accessible to hobbyists and small businesses, sparked a resin printing revolution, and established Formlabs as an industry leader.

Read all:

News & Gossip:

#1

Unfortunately, not all of the industry’s leading companies performed equally well in the first quarter of the year… Nano Dimension began 2026 with significantly higher Q1 revenue, reaching $29.7 million, but it was mainly due to the inclusion of Markforged.

At the same time, the company reported quite a significant net loss of $69.7 million, largely tied to a $40.4 million writedown of Markforged’s value 😱

Excluding Markforged, Nano’s core revenue declined -12% year-over-year. The company suspended its 2026 guidance while pursuing strategic alternatives, including potential mergers and asset sales.

Nano recently sold its AME and Fabrica businesses to reduce annual cash burn, while focusing future growth on Markforged and Essemtec operations (I was describing this here)

#2

Chinese Peopoly launched the Giga 800, a large-format FGF 3D printer with an 800 mm build volume priced at just $15,000. It uses industrial pellets instead of filament, reportedly extruding 3 kg/hour.

It is based on Klipper firmware - same as all current filament desktop printers. Designed as “air-gap ready” for defense and aerospace, the printer targets applications like molds and fixtures. Early access is open for commercial partners.

#3

Endless Industries has integrated its continuous fiber reinforcement technology into BigRep’s large-format IPSO 105 printer after nearly two years of development.

The result? Parts that can be up to 20x stronger than standard thermoplastics while remaining significantly cheaper than traditional automated fiber placement technologies 👀

The system uses recyclable materials such as PA and PP and is aimed at aerospace, medical, and tooling applications - sectors where strength-to-weight ratio is critical.

Joint sales across Europe are scheduled to begin in summer 2026, with global expansion planned through 2028.

#4

Samuel, Son & Co. has officially exited the 3D printing industry by selling its AM subsidiary, Burloak Technologies, to i3D MFG - a division of BTX Precision.

Burloak operates more than 30 metal 3D printers across two facilities, and after the acquisition the combined organization now controls over 60 systems focused on aerospace, defense, and medical manufacturing.

Another sign that consolidation in metal AM is far from over

#5

Some of you may have heard about the drama currently unfolding across social media and YouTube regarding Bambu Lab and the company’s supposedly negative approach to open-source.

Unfortunately, due to my contractual obligations, I can’t reveal much… which is a shame, because as usual, the situation has a far more interesting background than what the clickbaiters are presenting.

It’s obvious that Bambu Lab made mistakes here - but not necessarily in the areas people are pointing to.

Besides… there still isn’t a better alternative anyway 🤣

#6

Because one of the potential alternatives got carried away on social media and, taking advantage of its competitor’s stumble, published a long post about the “Evil China”:

But people are already noticing how awkward it is, and they are calling it out:

I wish the man had better advisors. Or start listen to them if he has.

BumpMesh is one of those tools that feels so obvious in hindsight that it’s surprising nobody built it earlier. At least not in this form.

And even though the software first appeared at the beginning of April - so it’s not exactly brand new anymore - it still deserves a dedicated article in this series. Better late than never, especially when the tool is genuinely this good.

So what exactly is BumpMesh?

Imagine designing a microphone grip or terrain for a tabletop game and wanting the printed surface to have texture - not the random kind created by fuzzy skin in a slicer, but something controlled, predictable, and custom.

In parametric CAD, that simply isn’t possible. Blender theoretically allows it, but Blender is its own universe, complete with its own vocabulary and a learning curve capable of killing motivation instantly.

BumpMesh fits perfectly into that gap: it’s a free browser-based tool for applying displacement textures to 3D models.

BumpMesh

No installation, no registration, no subscription fees. Files are processed locally on the user’s own computer.

Who’s behind it?

The creator is Stefan Hermann, better known from his YouTube channel CNC Kitchen. Today it’s one of the largest and most respected channels dedicated to 3D printing.

Hermann is known for approaching every topic like an engineer - with testing, measurements, and conclusions.

He’s not a 3D artist, and he openly admits Blender defeated him: it was opened, half a UV unwrapping tutorial was watched, confusion followed, and the software was closed. Instead of continuing to fight a tool that didn’t match his way of thinking, he did what makes sense in 2025 - he asked AI for help building his own solution.

The entire BumpMesh project was created in roughly a week and reportedly cost about twenty dollars in Copilot subscriptions. The result is impressive.

How does it work in practice?

The workflow starts at bumpmesh.com: upload an STL, OBJ, or 3MF file and begin. The interface is intuitive - there are dark and light themes, switching between English and German is possible, and the entire workflow is designed not to get between the user and the final result.

The key idea is that texture is generated from a grayscale image. BumpMesh already includes twenty-four built-in patterns, but custom images can also be uploaded - even something as simple as a product photo found online, cropped, desaturated, and boosted in contrast.

Pixel brightness directly determines how deep or how raised the texture becomes on the model surface.

To wrap the texture correctly around geometry, several projection modes are available. Planar works well on flat surfaces, Cylindrical on bottles and handles, and Triplanar on more cube-like forms. This is one of the most important parameters, and experimenting with it is worthwhile because a poor choice becomes visible immediately.

If texture should only appear on selected parts of the model, surface masking is available. There is a brush tool as well as a bucket fill tool operating on the model’s faces. Areas can be selected to receive texture, or excluded entirely - for example the underside of a model, to preserve bed adhesion. That part is handled by Angle Masking, which automatically avoids surfaces positioned at a specified angle. It is especially useful for preserving clean bridges and overhangs.

The “symmetric displacement” toggle determines whether the texture protrudes outward or sinks inward. Everything happens in real time - results appear immediately.

In his video, Hermann demonstrates something less obvious: the right texture can actually improve print strength. Textured surfaces also hide layer lines and seams very effectively - imperfections that can become irritating on smooth models. And for anyone printing terrain, props, grips, or any object where tactile surface control matters, BumpMesh essentially introduces an entirely new category of possibilities.

The only issue Hermann himself mentions is the need for repeated exports when applying multiple textures to a single model. One texture is applied, exported to STL, reimported, then another texture is added. One export cycle per texture.

He’s already working on improving that workflow, but considering this is practically the only real inconvenience, it’s difficult to complain.

The code is open-source and available on GitHub. Feedback can be submitted through Printables, MakerWorld, and GitHub. For something built in a week, it’s an impressively solid starting point.

If bumpmesh.com hasn’t been visited by you yet, this is a good reason to change that.

Last week, Materialise announced its financial results for the first quarter of 2026. Along the way - almost casually - the company mentioned a rather surprising yet highly symbolic decision.

The company is transferring its eyewear business to the management team running that operation.

This comes just a few weeks after RapidFit, another branch of the Belgian company’s activities, met exactly the same fate.

Two business sales within a single quarter is a fairly distinct signal. The official statement talks about “focusing capital on key areas,” but I would like to pause for a moment and consider what is really behind it.

Especially in the context of what I have described for years as The I Market Law of Additive Manufacturing.

3D printed glasses

Materialise spent years building its eyewear business - and by AM industry standards, it actually did it very well. The company was a pioneer in applying 3D printing to eyewear frame production, operating the largest dedicated eyewear manufacturing facility in Europe.

It developed proprietary materials - both PA12 and bio-based PA11. It created validated production workflows, the Eyewear Fitting Suite digital fitting system, finishing technologies, dyeing processes, and post-processing capabilities.

For years, it genuinely set the standard for the entire industry.

The technology worked and the products were of very high quality. The project was well thought out, developed consistently, and supported by global partners. And despite all that - Materialise decided to let it go...

So let’s look at the numbers, because numbers are always the most honest commentator.

The latest financial results

In the first quarter of 2026, Materialise reported revenues of EUR 66.3 million - essentially flat year-over-year (EUR 66.4 million in Q1 2025). At first glance, this looks stable, but the structure of those results is far more revealing.

The Medical segment grew by 6.7% to EUR 33.2 million and generated an adjusted EBITDA margin of 27.8%. Software - despite a slight revenue decline of 1.4% to EUR 9.6 million - dramatically improved profitability: adjusted EBITDA jumped 87.4% to EUR 1.1 million.

Meanwhile, the Manufacturing segment, which included both eyewear and RapidFit, recorded an -8.1% revenue decline to EUR 23.5 million, with adjusted EBITDA barely above zero - EUR 281 thousand versus a loss of EUR 377 thousand a year earlier.

This is the real context. Two “healthier” businesses are pulling results upward, while one - tied to custom manufacturing, physical production, consumer markets, and automotive - is dragging them down.

As a whole, the company returned to net profitability: EUR 1.82 million compared to a loss of EUR 535 thousand a year earlier. Margins improved, debt was reduced, and the cash position remained solid.

But it is entirely possible that management finally asked itself a difficult question: why continue carrying businesses that do not generate meaningful returns when you already have a medical segment delivering nearly 28% EBITDA margins?

The answer to why the divestiture happened is therefore complex. In part, it is classic portfolio rationalization - focusing on what generates margins and predictable growth.

Medical remains one of the most resilient and least cyclical segments across the entire AM industry. Software, meanwhile, generates recurring revenues that can scale without proportional increases in cost.

And eyewear? That is physical manufacturing, with a significant manual component (hand-finishing), a customer base consisting of fashion brands and optical retailers, and a market governed by completely different dynamics than medicine or industrial production.

And this is where my first Law of the 3D Printing Market enters the stage in full glory.

The I Market Law of Additive Manufacturing

The first law states:

This law says nothing about product quality, technological competence, or the correctness of a strategic vision. It speaks solely about market perception - how a company is viewed by customers, partners, and investors outside its own ecosystem.

For years, Materialise operated its eyewear business not as an “eyewear brand,” but as a contract manufacturer for other brands. And that is precisely the definition of operating within the boundaries of its own market - the company printed frames just as it prints surgical guides or prosthetics, meaning it was fundamentally providing a 3D printing service.

But the ambitions of that business reached further - proprietary fitting software, its own design studio, its own network of optical retail partners, and the Yuniku system as a direct touchpoint with end customers.

These were attempts to build something more than just contract manufacturing. Something closer to a true eyewear brand or platform.

And this is exactly the moment when the first market law begins to operate.

The eyewear market does not perceive Materialise as an eyewear player. It perceives it as a 3D printing company that happens to print frames. That is a fundamental difference.

For an eyewear brand looking for a contract manufacturer, Materialise is an excellent partner. But for a consumer buying glasses, Materialise does not exist as a brand - it exists merely as technology operating in the background. There is no independent consumer identity here, no direct relationship with the end user.

There is no loyalty to “Materialise glasses,” because from the perspective of the eyewear market, such glasses effectively do not exist.

This law is ruthless precisely because it operates independently of quality. Interestingly, the law is not a death sentence for the business itself - it is a sentence imposed on the brand.

An eyewear business spun out as a separate company, managed by people who have worked in it for years, without the burden of the “3D printing company from Leuven” label, suddenly has very different opportunities. It can build its own identity. It can develop relationships with customers and commercial partners without constantly having to explain that it is not simply another contract manufacturing division.

Materialise understands this - and that is why it is retaining a minority stake in the new company, believing in the business’s potential, but not necessarily in its ability to thrive within the existing corporate structure.

The first market law is ultimately a realistic observation about how additive manufacturing is perceived by other industries - and by the broader world itself.

Materialise will continue developing competencies where it has historically been genuinely strong - software and medical applications, where being a “3D printing company” is an advantage rather than a limitation.

Eyewear, meanwhile, is beginning a new journey. How will it end? We will see. But it will certainly be a fascinating story to watch.

A landmark deal in the French and European AM industry!

3D Prod has announced the acquisition of Sculpteo - one of the pioneers of digital additive manufacturing.

This marks a symbolic return of Sculpteo to French ownership after several years under the control of the German chemical giant BASF.

Sculpteo was founded in 2009 by Clément Moreau and Eric Carreel. Alongside Belgium’s Materialise and the Netherlands’ Shapeways, it was one of the first companies to build a sophisticated platform for ordering 3D-printed parts for both industrial companies and individual users.

The company developed a digital platform enabling users to upload files and order parts in more than 75 materials and finishes. In 2015, Sculpteo opened an office in San Francisco, entering the U.S. market, and three years later it was already serving more than 100,000 customers worldwide.

The company specialized in SLS and SLA technologies, and its platform also allowed users to edit and optimize models directly in the browser.

In November 2019, BASF - one of the world’s largest chemical corporations - acquired Sculpteo through its subsidiary BASF New Business GmbH.

BASF wanted to build a distribution channel for its advanced 3D printing materials developed under the Forward AM brand, while Sculpteo brought to the equation a broad customer base and a ready-made digital platform.

For Sculpteo users, this meant access to an expanded portfolio of industrial-grade materials. At the time, the deal appeared to be a win for both sides.

Reality, however, proved more complicated. In 2022, both founders left the company. In 2023, Sculpteo shut down its marketplace for independent designers, focusing exclusively on professional industrial services. The decision made business sense, but it was painful for the designer community that had spent years building their portfolios on the platform.

The key turning point came in July 2024, when BASF announced the divestment of the entire Forward AM business unit, including Sculpteo, effectively withdrawing from the 3D printing sector. Sculpteo became part of an independent organization while waiting for a new owner.

3D Prod comes from a very different background. Operating from Raon-l’Étape in the Vosges department, the company specializes in industrial 3D printing and custom manufacturing solutions. One of its strengths is the support of Platex - a company specializing in plastic injection molding.

This gives 3D Prod a unique advantage: the ability to offer comprehensive solutions ranging from prototyping to serial production. It is a company with strong industrial manufacturing roots, closely aligned with the needs of French and European manufacturers.

The merger creates a group employing 100 people, with combined revenue of €17 million and the ambition to reach €20 million by 2027.

The new entity operates two production facilities - in the Vosges region and in Villejuif near Paris - and serves more than 7,000 corporate customers across 62 countries. Particularly impressive is its machine fleet, which includes one of the world’s largest installations of HP MJF printers.

The group’s strategy is divided into two phases. Until 2027, the priority will be integration and leveraging the complementary expertise of both companies.

Starting in 2027, the second phase will focus on investments in modernizing the machine fleet, recruitment, and accelerating serial production, in line with the projected market growth exceeding 10% annually.

Atomic Layer from the Past:

Today, my friends, we have two!

14 years ago, Stratasys launched the Mojo 3D Printer, the cheapest professional-grade system at $9,900. It featured FDM technology and easy cartridge-like material loading. However, its tiny build volume and high price made it uncompetitive against cheaper alternatives like MakerBot. Discontinued quickly, Mojo symbolized a failed attempt to dominate the growing desktop 3D printing market.

Read all:

And on the very same day, but 9 years ago, Formlabs launched Form Wash and Form Cure at RAPID+TCT. These post-processing devices automated process of resin cleaning and UV curing, replacing messy manual workflows. Form Cure enabled medical-grade biocompatibility. Both streamlined the Form 2 ecosystem, greatly speeding production.

Read all:

News & Gossip:

#1

Backed by Lockheed Martin Ventures and In-Q-Tel, Firestorm Labs raised $82 million in Series B funding, to scale its xCell containerized 3D printing platform. The system produces combat-ready drones directly in the field, addressing contested logistics in the Indo-Pacific.

#2

In Q1 2026, Materialise swung to a net profit of €1.82 million, reversing a €535,000 loss from Q1 2025. Revenue held flat at €66.28 million. Gross profit rose 3.2%, and adjusted EBIT margin expanded to 3.7% from 1.0%. Net cash grew to €72.83 million. The Medical segment grew 6.7%.

The company reaffirmed 2026 revenue guidance of €273–283 million and adjusted EBIT of €10–12 million.

#3

SHINING 3D is another Chinese company willing to get some extra money from stock exchange. It filed for a Beijing Stock Exchange listing, aiming to raise 550 million yuan (75.7 million) by issuing up to 23 million shares.

The company posted strong 2025 results: 221 million revenue (+31% YoY) and net profit surging 190.9% in the first nine months. SHINING 3D met the exchange’s first listing standard based on 2024 net profit of 159.2 million yuan.

#4

Roboze acquired key assets from Dimanex to build an AI-powered, cloud-connected network for distributed manufacturing. Dimanex technology will integrate into Roboze’s Pandora and SlizeR software, enabling real-time optimization and digital inventory management.

#5

SprintRay received FDA clearance to 3D print porcelain dental crowns, entering the restorative dentistry market. CEO Amir Mansouri claims that they are able to print crowns in-office within 10–20 minutes, avoiding weeks-long lab delays.

#6

UltiMaker has boosted the performance of its S6 and S8 3D printers with the integration of the “Cheetah” motion planner. The upgrade enables print speeds up to 500 mm/s and acceleration of 50,000 mm/s². Beyond speed, UltiMaker boasts its extensive materials ecosystem: over 301 materials with ready-to-use print profiles.

Split3r emerged as a response to one of the most frustrating problems in desktop 3D printing: what to do when a model is simply too large for a printer?

Philippe Boichut - founder of the French company Qualup SAS, with over a decade of experience in industrial FFF printing - developed software that intelligently splits models into printable parts, complete with precise joints and alignment pins.

The project debuted on Kickstarter in September last year:

The campaign was successful: 2459 backers pledged a total of over €111,000. It was a clear signal that the market had been waiting for exactly this kind of solution.

Since then, the Qualup team has been steadily releasing new versions of the software. The latest - 1.5.0 - opens up entirely new possibilities for the first time.

Full color!

Until now, Split3r worked with STL files. It handled cutting, joints, and export very well - but color was irrelevant, because STL simply doesn’t store it.

If you wanted to split a colored model intended for printing on a machine with AMS or another filament-switching system, you were on your own - which in practice meant painful and boring manual repainting of each part after cutting.

With version 1.5.0, that changes.

Split3r can now import a colored 3MF file exported from Bambu Studio, split it into parts that fit the printer’s build plate, and return ready-to-use 3MF files - with colors preserved on every fragment. Each part goes straight into the slicer, without manual repainting or guessing which region had which color.

It sounds simple, but the challenge was in the details. Blender - which powered Split3r’s geometry engine from the very beginning - could not handle color data from the 3MF format.

For months, Boichut worked on replacing Blender’s engine with his own implementation.

The solution - as is often the case with the hardest engineering problems - came suddenly, at night, as a complete concept. A few days later, version 1.5.0 was up and running.

The new geometry engine doesn’t just support color - it is also dramatically faster. Cuts are performed seven to eight times faster than in Blender. The engine operates directly on mesh surfaces without volumetric reconstruction, which means typical models are processed in seconds rather than minutes.

Importantly, it does not require a perfectly watertight mesh. Open shells, zero-thickness walls, overlapping surfaces - all are handled correctly, without the need for pre-repair.

Additionally, the mathematical core uses constrained triangulation for all cutting surfaces, ensuring dimensional accuracy without rounding or smoothing artifacts.

Blender remains available as an optional engine during the beta phase, although without color support.

Colored cross-sections: no more gray joints

Another practical improvement which is introduced is that cutting surfaces and tenons (alignment connectors) now inherit the dominant RGB color of the part instead of defaulting to neutral gray. In practice, this means fewer filament changes during printing, as the printer no longer needs to switch colors just to produce internal joints.

AutoShell and Smart Selection for colored models

Two features available in the Split3r Add-on have also been updated for color support. AutoShell - a tool for creating hollow shells - now generates an outer layer with preserved colors and an inner wall in the dominant color. The result can be re-split in Split3r without any color loss.

Smart Selection and Extract now work directly with colored 3MF files. If you want to isolate a specific region of a multi-color model, you can do it in just a few clicks - and the color remains intact.

Export the entire workspace to 3MF with one click

A new File menu option allows you to convert all PLY and STL files in the active workspace to 3MF in a single operation. Colored PLY files are exported with full RGB data, while non-colored PLY and STL files are converted as standard geometry. The entire folder structure is preserved in an export_3mf subfolder.

Split3r 1.5.0 is currently in testing and is expected to reach a stable release within a few weeks. The software is available via the developer’s store at https://shop.qualup.com.

Available versions include Split3r Basic (€60.00 excl. tax), Split3r Add-on (€35.00 excl. tax), and Split3r Full - which includes both Basic and Add-on - for €95.00 excl. tax.

According to the logic of the capital markets, Farsoon today is roughly fourteen times larger than 3D Systems and more than six times larger than Stratasys. It’s just a pity that this isn’t reflected in revenue.

This is the state of the additive manufacturing industry on stock exchanges in 2026. One might ask: will these people ever learn…?

Because consider this: Farsoon Technologies - a Chinese manufacturer of laser-based 3D printers headquartered in Changsha, with annual revenues of around 715 million yuan (just over $100 million) - is valued by the market at approximately 33 billion yuan (about $4.55 billion).

Meanwhile, 3D Systems, the company that literally invented 3D printing and created the entire AM market nearly 40 years ago, generates $387 million in annual revenue and is valued at just $346 million.

Numbers that don’t lie - and numbers that lie to everyone

Let’s put the data side by side. Farsoon closed 2025 with revenues of 715 million yuan and net profit of 69 million yuan. Annual revenue growth reached 45.43%, while profit increased by only 2.68%. Its market capitalization at the turn of April and May 2025 hovered around 33 billion yuan ($4.5 billion).

BLT, or Xi’an Bright Laser Technologies - the world’s second-largest metal 3D printer manufacturer by revenue - reported 667 million yuan in revenue in the first half of 2025, with 17% year-over-year growth. Its market capitalization rose from around $1 billion at IPO to over $2 billion.

And let’s not forget the miraculously revived Velo3D, recently relisted on the NYSE. The company closed 2025 with $45 million in revenue and a modest 12% growth, yet its market cap exceeds $400 million.

On the other side of the mirror, we have 3D Systems, which closed 2025 with $386.9 million in revenue - a decline of over 12% year-over-year. Its market capitalization: $346 million.

Stratasys, creator of FDM and PolyJet technologies, with $551 million in revenue in 2025, is valued at just under $768 million.

Materialise, the Belgian provider of 3D printing software and services, with revenues exceeding €267 million, is valued at merely $329 million.

So we are looking at companies generating three to five times higher revenues, yet valued five to ten times lower.

At present, Farsoon and BLT individually have a higher market value than all other publicly listed AM companies combined.

The legacy players vs “the new breed”

3D Systems, Stratasys, and Materialise are companies that built the additive manufacturing industry from the ground up - literally and figuratively.

3D Systems and Stratasys both serve customers in aerospace, defense, medical, dental, automotive, and consumer sectors, holding certifications across all key areas. Materialise, although smaller than the other two, has for over 30 years delivered Magics software - effectively the industry standard for 3D print file preparation - and built certified supply chains for the medical and aerospace industries. Its certifications and regulatory know-how are assets that cannot be bought overnight at any price.

Farsoon, BLT, and Velo3D are different - focused, specialized companies targeting specific segments. Farsoon specializes in laser sintering systems for polymer and metal powders, emphasizing open system architectures. BLT is effectively a specialized manufacturer of metal laser printers with a dominant position in China’s aerospace sector. Velo3D is a niche producer of metal AM systems and service provider, aimed at the space and defense industries.

All three are valuable and technologically interesting companies. But none come close to the scale of portfolio, market reach, and technological diversification of their “legacy” counterparts.

Where do these discrepancies come from

The fundamental principle is this: the capital market values not what a company is today, but what investors believe it will become tomorrow.

The problem arises when this “future potential” is valued not on verifiable business fundamentals, but on narrative, macro trends, geographic exoticism, and herd behavior.

The additive manufacturing industry is particularly prone to this dysfunction for several reasons.

First, AM technology is inherently a story about the future. A manufacturing revolution, the end of traditional machining, digital factories - these narratives have real foundations, but their time horizon is constantly shifting.

This creates a space where valuation can grow independently of performance, because “the real breakthrough is still ahead.”

Second, companies listed on Asian stock exchanges operate in an investor environment with very different cultural and regulatory parameters. P/E ratios of 400–500x, which would be a red flag for Western investors, can be considered normal for companies with strong narrative growth potential.

Third, liquidity and trading volumes in these markets may be more susceptible to short-term speculative impulses.

Why is this dangerous? Because high market valuations have real consequences - for companies, investors, and the entire industry.

An overvalued company makes it harder to raise capital on normal terms (as investors expect growth matching the valuation), exposes itself to painful corrections at the first sign of disappointment, and - perhaps most importantly - distorts the perception of the entire industry.

Institutional investors burned by one overvalued AM company tend to become cautious toward the entire sector.

Desktop Metal: a lesson the market seems to forget

You don’t have to look far for a perfect illustration of these mechanisms. Desktop Metal - a company once backed by Google Ventures, BMW, and Ford - went public in December 2020 via a SPAC merger at a $2.5 billion valuation, briefly becoming the most highly valued 3D printing company in the world.

At the time, in the first three quarters of 2020, its revenue was $13 million. Thirteen. Not billions - millions.

The company never reached profitability. In 2022, Desktop Metal wrote down $499 million in goodwill, followed by another $113 million in 2023, citing sustained valuation declines due to market conditions.

On July 28, 2025, Desktop Metal filed for Chapter 11 bankruptcy protection.

A company that raised $580 million through SPACs and maintained multi-billion-dollar valuations for years ended its journey selling assets for a total of $17 million.

Fifteen years of building, a billion dollars raised, thousands of pages of pitch decks about a manufacturing revolution - and $17 million at the end.

Yet the investment carousel in the AM industry did not stop after this case. Money continues to flow into companies in ways disconnected from their fundamental value - because the narrative of 3D printing’s future is too compelling for the market to resist.

Tesla and the lesson AM should learn

The history of capital markets offers an analogy that is even more spectacular - and, unlike Desktop Metal, still unfolding. Tesla spent years as a company whose valuation triggered allergic reactions among analysts committed to rational thinking.

For over a decade, its shares were valued many times higher than those of Toyota, Volkswagen, or General Motors - companies generating tens of times more revenue and profit, with global manufacturing, service, and distribution networks.

The justification? Tesla wasn’t just a car manufacturer. It was a technology company, the future of transport, a platform for autonomous driving, an exponent of the S-curve of energy transition.

And for a time, that narrative worked. Let’s take on this example: Elon Musk announced a $25,000 electric car - the Model 2. Investors were thrilled. Then he said it would be ready in 2023. Then he pushed it to 2025.

Then he canceled the project entirely, declaring that the future belonged to robotaxis, not cheap cars.

He invested billions in autonomous driving and humanoid robots to transform Tesla into an AI company. And then reality started knocking in a way that became hard to ignore.

Today, Tesla is facing a downturn difficult to imagine for a company that recently set the pace for the entire EV market.

Global car sales fell by 8% in 2025 compared to 2024, to 1.64 million units - while 2022 projections assumed Tesla would sell that number in the first quarter of 2026 alone. Net profit shrank by 46%.

Robotaxis - the cornerstone of new Musk’s strategy - are operating in Austin, Texas, under human supervision with still relatively high accident rate compared to the average driver.

The humanoid robot Optimus, expected to reach mass production in 2025, practically does not exist as a commercial product.

Cybertruck, the flagship product meant to prove Tesla could build big, rugged things, sold around 50,000 units - just 20% of planned capacity and a microscopic fraction of the one million reservations Tesla once boasted.

JP Morgan analysts estimate that, under a fair valuation, Tesla shares are worth less than half their current market price. Some models suggest that once the speculative premium disappears, the decline could exceed 90%.

And here lies the lesson for the AM industry

For years, Tesla served as proof that narrative-driven valuation works - that the market can and should price the future, not the present, and that technological ambition is worth more than profitability.

But a narrative without fundamentals is a delayed sentence, not an alternative to profit.

It seemed that the additive manufacturing industry should have learned from this - but the carousel keeps spinning.

The question is: who falls off next?

Belgian Amnovis has just set foot on the American market - and in a very strategic location.

Amnovis - a contract manufacturer specializing in metal 3D printing for regulated industries - has acquired the additive manufacturing operations of Westconn Precision Technologies and opened a new production facility in North Webster, Indiana.

This is no random location. North Webster sits right next to Warsaw - a town known in the medical world as the Silicon Valley of orthopedics.

Amnovis was founded in mid-2020 by Ruben Wauthle - a man who has spent years immersed in metal 3D printing for medical applications. Between 2010 and 2015, he worked at the legendary Belgian company LayerWise, a pioneer in metal 3D printing for implants, which was later acquired by 3D Systems. He then served for several years as Business Development Director for Healthcare.

When he founded Amnovis, he knew exactly what he was doing - and for whom. From the outset, the company has focused on regulated sectors, primarily orthopedics and implants, offering contract metal 3D printing services using LPBF, along with CNC machining and post-processing.

Westconn Precision Technologies is a family-owned company from Connecticut specializing in precision CNC machining and EDM. Its additive business will now be absorbed into Amnovis’ new Indiana operation.

In practice, this means that metal 3D printing, traditional machining, and electrical discharge machining will come together in one place - a complete package for customers whose components require multiple production stages.

A key element of the entire strategy is mirroring.

The new facility in Indiana is designed to operate as a twin of the Belgian headquarters - with identical quality management systems, procedures, digital workflows, validation frameworks, and hardware strategy.

For customers, this means something very tangible: they can split production across two continents without needing to requalify a supplier twice. One quality system, two production locations - in the EU and the U.S. simultaneously.

This is far from a minor detail. FDA and MDR requirements represent two different regulatory worlds, and most contract manufacturers operate in only one of them. Amnovis is positioning itself to serve customers in both jurisdictions with a single partner.

Considering that Warsaw, Indiana is literally home to Zimmer Biomet and DePuy - a place where global orthopedic purchasing decisions are made — it’s hard to imagine a better neighborhood.

The U.S. medical device market accounts for roughly one-third of global demand.

Companies like 3D Systems and Lincotek have been building bridges between Europe and the U.S. for years, precisely because physical proximity to customers is no longer a competitive advantage - it has become a requirement for market entry.

Amnovis enters this game with a very specific strength: it offers not just manufacturing capacity, but a unified operational system across two continents - one that can genuinely simplify life for customers developing applications in both markets simultaneously.

Atomic Layer from the Past:

9 years ago, Mosaic Manufacturing released the beta version of CANVAS, the first cloud-based platform dedicated to multi-material FFF 3D printing. Integrated with Palette devices, CANVAS solved major infrastructure gaps: existing slicers couldn’t handle complex multi-filament prints, forcing users into inefficient manual workflows.

CANVAS introduced automatic alignment, drag-and-drop color assignment, cloud slicing, and Chroma post-processing. Running entirely in a browser, it removed the need for powerful computers. It also popularized adding color to single-color STL files.

Through intelligent slicing and continuous updates, CANVAS transformed a cumbersome technical process into a user-friendly tool, shaping education and semi-professional 3D printing.

Read all:

News & Gossip:

#1

Eplus3D, Rosswag Engineering, and Qualloy have partnered to build a validated metal AM ecosystem. Rosswag will install an Eplus3D EP-M550 eight-laser system, operational by June 2026. Qualloy powders will be qualified at Rosswag’s German innovation center and then made available online. The project will be presented at Rapid.Tech3D in May.

#2

3D Systems has received EU Class IIa MDR certification for its NextDent Jet Base and Jet Teeth materials, enabling immediate commercialization of the complete NextDent Jetted Denture Solution across Europe.

#3

Scrap Labs has debuted the Scrap 1, a compact metal LPBF 3D printer priced around $9,600 as a kit or $17,990 fully assembled. The 100×100×100 mm system features a 200W laser, open-source Klipper firmware, and 99% part density in steels, copper, nickel alloys, and cobalt chrome. Aimed at small shops and labs, it offers an affordable alternative to six-figure industrial machines. Still in development, the printer is expected to begin shipping in June 2027 if validation progresses as planned.

#4

Modix has launched the MAMA-1000, a large-format 3D printer with a 1,000 x 1,000 x 1,000 mm build volume. A more compact version of the MAMA-1700, it supports both pellet and filament feedstock via interchangeable print heads - filament for finer details, pellets for lower cost and higher throughput.

#5

Farsoon Technologies reported 2025 revenue of 715 million yuan (~$104.5 million). Cumulative global system sales exceeded 1,400 units, including over 800 metal LPBF machines. Those are quite interesting numbers given the fact, the company’s valuation on stock exchange is $4,67 billion.

#6

This is quite interesting… JD.com is one of China’s largest e-commerce platforms, often compared to Amazon. Founded in 1998, it specializes in direct sales - unlike purely marketplace-based competitors, JD controls the entire logistics chain.

Now, the company is accelerating 3D printing’s move from a niche tool into ordinary homes. After seeing 176% GMV growth during June 2024’s “618” festival, JD launched direct sales for brands like Anycubic, Bambu Lab, and Creality.

By 2025, JD began organizing printing farms, offering on-site installation in 31 cities, and opening offline experience centers at JD Home locations. In March 2026, JD unveiled a 3D & UV printing customization platform integrating design, production, and delivery.

Users can upload ideas or IP and turn them into physical products.

This shift from pure e-commerce toward distributed manufacturing positions JD as bridge connecting consumers with production on demand.

Read more: www.3dzyk.com