Freeform’s AI laser factory is a bet that software will finally eat hardware
Silicon Valley has spent a decade promising that making atoms will become as fast and flexible as shipping code. Freeform’s new $67 million Series B is one of the clearest attempts yet to make that cliché real. If the startup succeeds, a big chunk of the world’s metal parts could be designed, simulated and printed under the control of GPUs rather than machinists. In this piece, we’ll look at what Freeform is actually building, why Nvidia is leaning into it, how it fits into the new wave of “manufacturing-as-a-service” — and what it means for European industry trying to stay relevant in advanced manufacturing.
The news in brief
According to TechCrunch, Los Angeles–based Freeform has raised a $67 million Series B round to scale its metal 3D-printing platform. The round includes Apandion, AE Ventures, Founders Fund, Linse Capital, Nvidia’s NVentures, Threshold Ventures and Two Sigma Ventures. PitchBook reportedly pegs the company’s valuation at around $179 million after the financing.
Freeform currently runs a system called GoldenEye that uses 18 lasers to fuse metal powder into complex parts. The fresh capital should fund Skyfall, a next‑generation platform that upgrades from tens to hundreds of lasers, with an ambition to churn out thousands of kilograms of metal components per day.
The company, founded in 2018 by former SpaceX engineers Erik Palitsch and Thomas Ronacher, positions itself as “AI‑native”: it pairs dense sensor data from the printers with Nvidia H200 GPU clusters on‑site to run real‑time physics simulations and optimize the full manufacturing workflow. Freeform says it is already shipping hundreds of mission‑critical parts, plans to hire around 100 people, and expand its facility to clear a growing backlog of orders.
Why this matters
Freeform is not “just another” metal 3D printing startup. The core bet is that the bottleneck in advanced manufacturing has shifted from hardware to computation.
Traditional metal printing hardware is extremely expensive, temperamental, and optimized for prototyping or low volumes. That’s why aerospace and medical implants use it, but your car chassis or wind turbine does not. Freeform is effectively saying: if you over‑invest in sensing, simulation and control software — and throw serious GPU power at the problem — you can push throughput and reliability high enough to compete with conventional machining for a much wider set of parts.
The winners if this works:
- High‑mix, high‑performance industries: space, defense, energy, robotics and EVs that need complex geometries and fast iteration. They get shorter lead times and less dependence on fragile supply chains.
- Nvidia and the AI hardware stack: this is another proof point that GPUs are not just for LLMs and images, but for industrial physics workloads that directly drive revenue in the real world.
- Software‑first manufacturing startups: Hadrian (CNC), Divergent and VulcanForms (metal printing) all benefit if investors are convinced that “AI+factories” is a real asset class, not a fad.
The potential losers:
- Traditional machine shops that compete mainly on labor cost or legacy relationships.
- Legacy industrial OEMs whose business models depend on selling closed machines rather than networked, data‑rich platforms.
In the near term, the main implication is psychological: this is a vote of confidence that capital is once again willing to fund expensive hardware — as long as the core moat is software and data.
The bigger picture
Freeform sits at the intersection of three big trends: AI‑accelerated engineering, the unbundling of factories into cloud‑like services, and geopolitical pressure to reshore production.
Over the past five years, we’ve seen simulations and digital twins move from niche tools to mainstream. Automotive OEMs, energy majors and even construction firms now rely on high‑fidelity physics models to de‑risk designs. What Freeform is attempting is the next step: keep the simulation loop running during manufacturing itself, fed by sensor data from every layer of metal being fused. That’s a shift from offline design tools to online process control.
At the same time, investors have quietly rediscovered the factory. Companies like Hadrian (aerospace machining), Flex and Foxconn (electronics), VulcanForms and Divergent (additive metal) are building what you could call “AWS for parts”: you send a CAD file; they return certified hardware. Freeform is essentially pitching itself as an API for high‑end metal components, with AI doing the hard work of tuning parameters, managing quality and squeezing more parts per hour from the same hardware.
Historically, we’ve been here before in softer form. In the 1980s and 1990s, Japanese and then German factories pulled ahead with robotics and statistical process control. What’s different now is the scale and granularity of data. Where a Six Sigma program monitored a handful of variables, an AI‑native printer can instrument thousands, from laser power and melt‑pool temperature to micro‑vibrations in the build chamber.
Compared with competitors, Freeform’s edge is its overt embrace of AI hardware and real‑time simulation as the heart of the product, not an add‑on. VulcanForms has taken a similarly data‑heavy approach, while Divergent integrates printing into a full “digital factory” for automotive structures. The race is on to prove who can translate terabytes of melt‑pool telemetry into a defensible margin.
The European angle
For Europe, this kind of platform is not an abstract US venture story. It goes straight to the heart of industrial competitiveness.
European manufacturing — especially in Germany, Italy, France and the Nordics — is built on precision metalworking for automotive, aerospace, machinery and energy. These sectors are under simultaneous pressure from electrification, stricter climate regulation and cheap Asian competitors. An AI‑driven, high‑throughput metal printing layer could be a lever to keep advanced design and small‑series, high‑value production in Europe even if commodity volumes move elsewhere.
Regulation will matter. The EU AI Act will treat AI systems used in safety‑critical components (think aerospace brackets, medical implants, parts for nuclear or grid infrastructure) as high‑risk. Any Freeform‑like platform selling into Europe will need transparent quality controls, auditability of models and clear human oversight. Add GDPR to the mix if worker performance data, video feeds or biometric information from factory floors end up in the training logs.
There is also a strategic autonomy angle. Brussels is already funding microelectronics and batteries (Chips Act, IPCEIs). Advanced manufacturing — including metal additive and AI‑enhanced machining — is the missing pillar. If Europe doesn’t nurture its own “AI factories”, it will end up buying capacity from US or Chinese platforms, with all the usual headaches: data export, extraterritorial sanctions, dependence on foreign defense supply chains.
European players like EOS, SLM Solutions (now part of Nikon), Trumpf or Materialise already dominate important parts of the metal 3D‑printing stack. The question is whether they can evolve from selling machines and software licenses to operating data‑rich manufacturing clouds that look more like Freeform.
Looking ahead
The next 24–36 months will tell us whether Freeform is a curiosity or a template.
Technically, the big milestone is Skyfall: scaling from 18 to “hundreds” of lasers while maintaining consistency is non‑trivial. More lasers multiply not just throughput, but also the number of things that can go wrong — from thermal distortion to complex interactions in the powder bed. If the company can demonstrate stable yields at thousands of kilograms per day, it becomes a serious alternative to traditional machining for a slice of parts.
Commercially, the key signal will be customer mix and contract length. Are orders mostly venture‑funded space and defense startups, or do Tier‑1 automotive, industrial and energy customers start committing to multi‑year capacity? The latter would validate manufacturing‑as‑a‑service as part of the core supply chain, not just a backup.
There are open questions. How defensible is the “we have the most physics data” claim in a world where every serious player is wiring up sensors? Will regulators decide that AI‑controlled manufacturing workflows need their own category of oversight, especially when used for weapons or critical infrastructure? And can Freeform attract the kind of hybrid talent — part ML engineer, part metallurgist, part operations geek — that this model demands?
For European readers, the thing to watch is whether similar ventures emerge locally, perhaps around existing champions like EOS or in newer hubs from Munich to Barcelona. The alternative is a future where EU factories are reduced to low‑margin execution while the AI brains of manufacturing sit in Californian data centers.
The bottom line
Freeform’s raise is less about another shiny 3D printer and more about a power shift: from hardware vendors and job shops to data platforms that treat manufacturing as a continuously optimized, GPU‑accelerated workflow. If AI‑native factories win, countries and companies that control them will gain real leverage over how quickly physical innovation happens — and who captures the value. The open question for Europe is simple: will it build its own Freeforms, or rent them from abroad?
