Physical AI financing strain as Integral AI fails to secure follow-on funding

Physical AI startups are facing a funding crunch, highlighted by Integral AI’s reported downfall. The robotics company, founded by former Google researchers Jad Tarifi and Nima Asgharbeygi, said it is building “foundational world models” for robotics and self-driving systems, but capital needs have outpaced results. Integral AI raised about $4.7M–$5.5M in seed funding, with backers including SoftBank’s Deepcore and Samsung Next. However, the company—around 15 employees—has been seeking roughly $10M for a next round as of March 2026, underscoring the physical AI financing gap. The article notes a “gap between hype and hardware.” In December 2025, Integral AI announced what it called an “AGI-capable model” that could let robots learn new skills without labeled data. The firm worked with industrial partners such as Denso Corp., with engagement also mentioned with Toyota and Sony. The key issue is data and deployment cost. Training software models is expensive but relies on abundant internet text. Training robots for tasks like warehouse navigation requires costly physical interaction data from real-world trials or complex simulations. Failures also carry higher safety and operational stakes. Unlike pure software AI bets that can progress via demos or benchmarks, physical AI financing for robotics depends on working hardware, real-world deployments, and safety records. Even Integral AI’s strong founder credentials and industrial partnerships were not enough to “smoothly” secure follow-on capital, suggesting tougher conditions for future funding in the physical AI sector.