An Nvidia insider-turned-VC on why robotics hasn’t hit its ChatGPT moment, yet

On TechCrunch’s Equity podcast, Amulya Vishwanath — seven years leading product marketing for NVIDIA’s robotics platforms, now founder and managing partner of Techable VC — argues robotics is in its “Lego era,” still rebuilding basic blocks like actuators, robotic hands, software, and safety standards. That view pushes against claims that a general foundation model will deliver robotics’ ChatGPT moment; Equity previously aired General Intuition CEO Pim de Witte arguing a single foundation model could make much specialized robotics work redundant.

Dominic-Madori Davis talked with Vishwanath about first checks at Techable and lessons as one of the first women executives to spin out of NVIDIA into an early-stage firm investing across physical AI — robotics, materials science, and life sciences.

We’re with Vishwanath on the hardware humility. ChatGPT was software on existing phones and laptops; robots still need hands that don’t break and safety cases that regulators trust.

Foundation models will matter — perception and planning are getting better fast — but calling a ChatGPT moment before the Lego bricks fit is how you get demo reels and stranded capex. Useful physical AI is boring reliability plus model smarts.

I’m watching where Techable’s first checks land, and whether 2027 robot demos still hand-wave the actuator and safety stack.

Physical AI here spans robotics plus adjacent materials and life-sciences bets, not only humanoids.