Infineon, NVIDIA and Microsoft back European physical AI startups ... - eeNews Europe

Infineon, NVIDIA, and Microsoft aren't investing in European physical AI—they're executing an ecosystem lock-in. Infineon owns the power-management and edge-sensing layer; NVIDIA anchors inference compute and the CUDA moat; Microsoft harvests the application tier via Azure. A vertical silicon-to-cloud closed loop is forming. The real bottleneck in physical AI isn't model scale—it's the millisecond-level perception-decision-actuation loop. Infineon's AI MCUs sit precisely at that chokepoint. Once European startups clear the production threshold in robotics and industrial control, order structures for EDA, CoWoS-L packaging, and HBM will be restructured. On compliance, the EU AI Act classifies autonomous physical-AI systems as high-risk, adding 30-40% to operating costs. Paradoxically, this becomes Europe's structural moat: non-EU entrants face extended certification timelines, widening the local window. Strategically, NVIDIA is embedding CUDA into European industrial DNA while reducing customer dependence on purely US-stack solutions. Microsoft is carving a third pole beyond AWS and GCP. If AMD and Renesas don't respond within 12 months, Europe's edge-AI share will be effectively locked. 12-24 month call: two to three European physical-AI hardware unicorns emerge; Infineon's AI SoC revenue share crosses 15%; edge-inference chip CAGR exceeds 40%. The US-EU-China tripartite structure in AI hardware hardens into a fixed architecture.