The Metzingen based robotics company is targeting $1.4B to fund a Neuraverse data layer and real world training sites. The underlying question is whether a non Silicon Valley player can assemble the data, compute, and deployment infrastructure that physical AI requires.
NEURA Robotics, a six-year-old company based in Metzingen, a small city in southwestern Germany, announced on June 10 that it is raising up to $1.4 billion in Series C funding to build the data and training infrastructure for "physical AI." The category, as NEURA uses it, covers robots that learn to move and manipulate in the real world rather than only in simulation. The round is a target rather than a closed financing, and the company's choice of investors signals exactly which part of that bet it wants to fund.
The named roster is unusually broad for a Series C in industrial robotics. According to the company's Series C announcement and The Robot Report's write-up, the round includes Tether, Qualcomm Technologies, Amazon, NVIDIA, imec.xpand, Bosch, Schaeffler, the European Investment Bank, Lingotto Horizon, InterAlpen Partners, AGC, and Giano Capital. That is a strategic stack rather than a financial one. A stablecoin issuer sits next to a chip designer, a cloud hyperscaler, a foundry-affiliated deep-tech fund, two of Germany's largest industrial suppliers, a multilateral lender, and a deep-tech growth firm. Each is being asked to fund a different layer of the same architecture.
What the architecture is supposed to support is the Neuraverse, a cloud-based platform connecting robots, developers, and partners, paired with NEURA Gyms, purpose-built real-world training environments where robots accumulate sensor data through physical interaction rather than simulation alone. Founder and CEO David Reger has framed the combination as a category bet: physical AI requires a different kind of data layer than language models, and that data has to be collected where robots actually work, on factory floors, in warehouses, and in service settings.
The product line anchoring the bet is broader than the humanoid headlines suggest. NEURA already ships the MAiRA cognitive robot arm, the LARA arm, the MAV mobile platform, the MiPA service robot, and a four-legged Quadruped, with the 4NE1 humanoid in development and an in-house AI brand called AURA running across the line. Industrial partners listed on the company's homepage include Bosch, Schaeffler, NVIDIA, Qualcomm, Kawasaki, Vodafone, AWS, SAP, HD Hyundai Robotics, and Körber, with academic and integration partners such as TU Munich, KENMEC, Beumer Group, tesa, Abicor Binzel, and Drees & Sommer also named.
The interesting question is not whether NEURA can raise $1.4 billion. It is whether a non-Silicon Valley industrial robotics company can assemble the data, compute, and deployment infrastructure that physical AI requires, and what the gap looks like between the rhetoric and the current footprint.
The rhetoric is large. NEURA says its order backlog and strategic deployment pipeline exceed $1 billion and that it is targeting production of several million robots by 2030. It also claims, per the company's Series C page, that this is the largest funding round ever for a full-stack robotics company. Both numbers are company assertions. The backlog is not independently audited, the production target is a 2030 ambition, and "largest ever" claims in a category as small and fast-moving as humanoid robotics are not yet subject to the kind of third-party benchmarking that would let a reader confirm them. They signal ambition rather than milestones.
The footprint is more modest. Tech.eu's coverage and the company's own communications put headcount somewhere between roughly 1,000 and 1,600 employees across eight or more global locations, with the higher figure the more recent one. The company has begun naming operating talent, including COO Jens Fabrowsky, but the public record of deployments, fleet sizes, and customer outcomes at the scale implied by the round is still thin. Independent operator or analyst perspective on what the Neuraverse actually runs in production, and what the NEURA Gyms have produced, is not in the public sources reviewed.
That is the structural question the round is really testing. Physical AI, the category rather than the marketing, does require something language models do not: large, diverse, real-world sensor data tied to physical outcomes, plus the compute and tooling to train on it. NEURA's pitch is that a company with industrial-robotics roots, European manufacturing adjacency, and an unusually broad investor base can build that layer in the open, from outside the U.S. cloud-AI cluster. The $1.4B target is the budget. The Neuraverse and the NEURA Gyms are the instruments. Whether "cognitive robots" trained on a shared physical-AI data layer can ship at scale and earn the category thesis is what the next 24 to 36 months of deployments will have to answer.
For now, the round is open, the structure is set, and the most concrete fact on the record is the investor list. It is a cross-section of the companies whose chips, capital, and factories physical AI will have to run on if the bet works.