Jonas
Schneider

Most CNC machine shops in Europe run their equipment roughly 15-20% of the time. The rest is scheduling friction, human bottlenecks, and the slow churn of tacit knowledge that lives in a machinist's hands and disappears when he retires. Jonas Schneider, Founder and CEO of Daedalus, has a precise plan to fix both problems - and $41.1 million to fund it.

Daedalus operates a 50,000-square-foot precision manufacturing facility in Karlsruhe, Germany, where AI software orchestrates every step on the shop floor. The output: milled and turned aluminum, steel, and stainless steel parts at tolerances under one micrometer. The clients: defense contractors, medical device makers, aerospace companies, semiconductor manufacturers. The pitch Schneider uses: AWS for precision manufacturing.

Before there was a factory, there was a problem. While Schneider was leading the software side of OpenAI's robotics team - one of the first engineers the company ever hired - he found himself waiting months for precision-machined replacement parts. Standard industrial reality, universally accepted, almost universally ignored. He did not ignore it.

In November 2019, Schneider left OpenAI - at a moment when every serious AI talent was angling to stay or join - and returned to Germany. He went through Y Combinator's Winter 2020 batch, raised $2.6 million from Khosla Ventures, then an $11.5 million seed round led by Lee Fixel's Addition. In February 2024, Nokia-backed NGP Capital led a $21 million Series A.

What separates Daedalus from typical AI-first manufacturing plays is that Schneider actually runs a factory. The company doesn't just sell software to machine shops - it is the machine shop. That distinction matters enormously: it forces the software to work in the real world, where chips fly, coolant temperatures shift, and a single worn cutting tool changes the geometry of the next thousand parts. Daedalus's AI monitors all of it in real time.

The bottleneck Schneider talks about most is not mechanical - it's cognitive. The typical CNC shop runs low not because machines are slow but because human operators cannot hold the full complexity of a job in their heads: optimal cutting parameters, tool wear compensation, inspection sequencing, rework decisions. His software takes those decisions off the floor and encodes them. More critically: it preserves them. When an expert machinist leaves, the knowledge stays.

Schneider studied Computer Science at the Karlsruhe Institute of Technology - one of Germany's engineering flagships - from 2012 to 2016, interning at Stripe along the way. His research record from the OpenAI years covers multi-goal reinforcement learning, hindsight experience replay, domain randomization, one-shot imitation learning, and simulation-to-reality transfer. These are not adjacent interests. They form a single obsession: making machine systems learn from physical feedback.

Manufacturing is a physical feedback problem at industrial scale. Every chip, every surface finish, every tool path is a data point. Schneider's bet is that those data points, properly captured and modeled, can let a relatively small software layer make decisions that used to require decades of craft. At Daedalus, the machines are already running at twice the industry average utilization rate. The company's 150 employees manage a throughput that conventional shops would staff very differently.

His broader argument - increasingly central to the European tech conversation - is that the continent's manufacturing heritage is not a liability in the AI era. It is an asset. Germany has the machine inventory, the supply chains, the quality standards, and the engineering culture. What it has lacked is the software layer that can unlock all of it. Daedalus is building that layer inside a real factory, not on a slide deck.

Schneider is also, quietly, a very early backer of Weights & Biases - the experiment-tracking platform used by research teams worldwide. He invested as an angel in 2017, when OpenAI was still young and W&B was barely a product. The investment reflects the same intuition that runs through his career: the infrastructure problem is always the most important one.

Daedalus serves defense, aerospace, medical devices, and semiconductor customers - sectors where tolerances aren't just a quality preference but a regulatory requirement. The company machines aluminum, steel, and stainless steel to sub-micrometer specs, using 3-5 axis milling and mill-turn operations. It is not trying to make commodity parts cheaper. It is trying to make high-complexity precision parts faster and more reliably than any human-managed shop can.

The Series A in early 2024 brought NGP Capital alongside existing backers Addition and Khosla Ventures. The round set the company's trajectory toward scaling the factory footprint and deepening the AI platform. Schneider's stated goal is not to replace machinists - it is to make manufacturing facilities capable of running with far fewer of them, at a time when the demographic math of European industry makes finding experienced machinists increasingly difficult.

There is a longer arc here that Schneider returns to: the accumulated knowledge in a machinist's head represents decades of trial, error, and refinement. When it is gone, it is gone. The software layer Daedalus is building is, among other things, a preservation project. A way to not lose what took a generation to build.