What is new

Amulya Vishwanath spent seven years leading product marketing for NVIDIA's robotics platforms before founding Techable VC, an early-stage firm investing in what she calls physical AI across robotics, materials science and life sciences. She spoke about her investment thesis on TechCrunch's Equity podcast, in an episode hosted by Dominic-Madori Davis.

The headline argument from the conversation is a pushback against a popular idea in robotics circles: that the field is about to have its own "ChatGPT moment," the point at which a single general-purpose model suddenly makes the technology dramatically more capable and accessible to everyone, the way OpenAI's chatbot did for text generation in late 2022. That view has been voiced by other investors in the space, including General Intuition CEO Pim de Witte, who argued earlier this year that one general foundation model could make much of today's specialized robotics work redundant.

Vishwanath's "Lego era" argument

Vishwanath disagrees with the idea that robotics is close to that kind of singular breakthrough. Her description is that the field is currently in a "Lego era," meaning the basic building blocks the industry needs are still being assembled piece by piece: actuators (the motors and mechanisms that let a robot move), robotic hands, the software that controls these systems, and the safety standards that govern how they can be used. In her framing, there is no shortcut to a general-purpose robotics model until these underlying components mature on their own.

This is a useful distinction for anyone following AI hype cycles. A "foundation model" is a large model trained on broad data that can be adapted to many tasks, which is what powered the leap in general-purpose chatbots. Applying that same logic to robotics assumes that once such a model exists for physical tasks, it would instantly generalize across different robot bodies, environments and safety requirements. Vishwanath's position, as described in the podcast, is that physical hardware and regulation do not move at software speed, so a single model breakthrough would not automatically translate into deployable robots the way it did for text and chat.

What counts as "physical AI"

A second theme from the interview is how broadly Vishwanath defines the category she invests in. Rather than equating physical AI with humanoid robots or factory arms, she describes it as covering "a molecule, a material, a medicine," alongside a machine on a factory line. In practice, this means her fund Techable VC looks at robotics, materials science and life sciences as related fields under one investment thesis, rather than treating robotics as a separate, narrower category.

She also used the phrase "conscious automation" during the discussion, though the episode summary does not spell out a precise definition of the term beyond its use in describing her investment philosophy. Readers interested in exactly how she distinguishes conscious automation from ordinary automation would need to listen to the full episode, since the written summary does not elaborate further.

Where she sees opportunity

According to the podcast description, Vishwanath looks for founders tackling large, structural problems rather than narrow technical demos. The two examples given are family-run factories that cannot find enough workers, and surgeons facing burnout — both framed as big, persistent labor and workforce problems that physical AI technologies could address.

  • Vishwanath led product marketing for NVIDIA's robotics platforms for seven years before founding Techable VC.
  • She describes robotics as being in a "Lego era," with core components like actuators, robotic hands, software and safety standards still being built out.
  • This contrasts with the view, voiced by General Intuition CEO Pim de Witte, that a single general foundation model could soon make much specialized robotics work redundant.
  • She defines physical AI broadly, including molecules, materials and medicines, not just robots.
  • She points to labor-shortage problems, such as family-run factories lacking workers and surgeons facing burnout, as examples of the big problems worth solving.
  • She discusses wanting her VC backing to function as a moat for founders, and says her NVIDIA network helps with due diligence and customer introductions.
  • The episode also touches on where she believes the U.S. and China each hold advantages, and her view that humanoid robots should be useful for more than demonstrations such as dancing.

Limits and open questions

Because this is a podcast interview rather than a research paper or product launch, there are no benchmark numbers, product names, or specific portfolio companies mentioned in the published summary. The description does not explain in detail what "conscious automation" means, nor does it specify which countries or companies Vishwanath believes hold the edge in the U.S.-China comparison she discusses, or how humanoid robots might be used beyond demonstrations. Readers who want those specifics, along with the reasoning behind her view that VC backing should act as a founder's competitive moat, would need to listen to the full audio episode rather than rely on the written recap alone.

How to listen

The conversation is available as an episode of TechCrunch's Equity podcast, hosted by Dominic-Madori Davis, and can be found on YouTube, Apple Podcasts, Overcast, Spotify and other podcast platforms, with updates also posted on Equity's accounts on X and Threads.

Sources