Analysis
Amulya Vishwanath spent seven years at Nvidia leading product marketing for its robotics platforms before leaving to found Techable VC, an early-stage firm investing across robotics, materials science and life sciences under what she calls a 'physical AI' thesis. On TechCrunch's Equity podcast, she laid out an argument that cuts against a lot of the capital currently flowing into robotics: the sector isn't close to its own ChatGPT moment, and waiting for one general foundation model to deliver it is the wrong bet.
The 'Lego era' argument
Vishwanath describes robotics as being in a 'Lego era,' where the basic building blocks -- actuators, robotic hands, software, and safety standards -- are still being rebuilt piece by piece rather than unified. That's a direct challenge to a competing thesis aired on the same show: General Intuition CEO Pim de Witte has argued that a single general foundation model would make much of today's specialized robotics work redundant. Vishwanath disagrees, and her background gives the disagreement some weight -- she spent years watching which parts of Nvidia's robotics stack enterprise customers actually struggled to integrate, rather than theorizing about it from the outside.
This isn't an abstract disagreement for founders raising right now. If Vishwanath is right, the startups worth funding are the ones solving a specific layer of the stack -- better actuators, better perception, better safety certification -- rather than the ones pitching a single model that handles everything. If de Witte is right, those specialized bets get commoditized the moment a general-purpose model catches up, the same way specialized NLP startups got squeezed once large language models matured.
The risk in Vishwanath's own framing is that 'physical AI' is broad enough to cover almost any hardware-adjacent AI startup, which makes it a useful fundraising label and a hard one to diligence against. She hasn't disclosed specific portfolio companies or check sizes from Techable VC, so it isn't yet possible to see which side of her own thesis she's actually betting real capital on. What is clear is that her move -- a technical operator leaving a dominant platform company to start an investing career -- is becoming a more common path into venture than it was even two years ago, as physical AI draws people with hardware backgrounds that pure software VCs typically lack.