AivexaNewsSearch
AI news for builders and product teamsChecked every hour

Physical AI is more than robots, says the Nvidia veteran now investing in it

Collected Oct 16, 2026

Amulya Vishwanath, who spent seven years leading product marketing for Nvidia's robotics platforms, has launched an early-stage venture firm called Techable VC, where she serves as founder and managing partner. She discussed her investment thesis on an episode of TechCrunch's Equity podcast with senior venture reporter Dominic-Madori Davis. Techable VC writes first checks into physical AI companies across robotics, materials science, and life sciences, according to the discussion.

Vishwanath's central argument is that physical AI is broader than robotics. She defines it to include "a molecule, a material, a medicine," or a machine on a factory line. That framing widens the addressable space well beyond humanoid robots and warehouse automation. In practice, this means a founder building a new actuator, a simulation stack, or a drug-discovery platform could all sit inside the same thesis. For product teams, the distinction matters because the buyers, sales cycles, and regulatory paths differ sharply across those segments even when the underlying models are similar.

She describes robotics as being in its "Lego era." Her point is that the basic building blocks are still being rebuilt: actuators, robotic hands, software, and safety standards. The metaphor implies that today's components are modular, interchangeable, and not yet settled into stable platforms, much as Lego bricks can be combined in many ways before a dominant design emerges. A likely trade-off is that this openness creates room for new entrants but also makes it hard to build durable moats on any single component, since a better brick can displace an incumbent quickly.

This view runs counter to the expectation that robotics, powered by a general foundation model, is about to have its "ChatGPT moment" — the moment when a shockingly capable chatbot launched in late 2022 and reset expectations across the software industry. General Intuition CEO Pim de Witte told Equity earlier this year that a single general foundation model would make much of today's specialized robotics work redundant and would be that moment. Vishwanath's position is effectively the opposite: the specialized, unglamorous work of rebuilding components and standards is the work that matters right now.

On the podcast she also discussed what she calls "conscious automation." She argues the best opportunities solve large, concrete problems, citing family-run factories that cannot find workers and surgeons facing burnout. Both examples point to labor scarcity and human capacity limits rather than to consumer-facing novelty. For developers and product managers, that suggests the near-term demand for physical AI may come from operators trying to keep existing workflows staffed, not from end users buying robots.

Vishwanath spent seven years at Nvidia and is described as one of the first women executives to spin out of the company and launch her own early-stage firm. She said she wants founders to see their venture capitalist as their moat, and that her years at Nvidia built the network she now uses for due diligence and customer introductions. That network effect is a common pitch among operator-turned-investor funds, and it is worth noting that access to customers is often as valuable to an early-stage hardware or materials company as the capital itself.

The conversation also covered where the United States and China each hold an edge in physical AI, and why humanoids should be good for more than dancing. Neither the specific national advantages nor the alternative humanoid use cases are detailed in the episode description.

Read against the current robotics funding cycle, Vishwanath's "Lego era" framing is a caution against assuming a single foundation model will collapse the stack. If components and safety standards are genuinely unsettled, then integration, reliability, and certification work carry real value, and the companies that own those pieces may outlast those that only own a model. This suggests investors and builders should watch for consolidation around specific actuators, hands, and safety regimes as a signal that the Lego era is ending. It is also a reminder that benchmark progress on generalist robot policies does not automatically translate into deployed systems in factories or operating rooms.

Why it matters: Founders and product teams working on robotics, materials, or life sciences now have a stated early-stage thesis to test against — component-level rebuilding plus labor-scarcity problems — from an investor with seven years inside Nvidia's robotics business. If Vishwanath is right that the building blocks are still being rebuilt, then differentiation for the next few years may come from depth in actuators, hands, software, and safety standards rather than from a single general model. Teams pitching generalist robotics should expect questions about which specific brick they own, and inference from the discussion suggests the strongest near-term traction may sit in industrial and clinical settings where worker shortages are already acute.

Editor's note: The source is a podcast episode announcement, so several of the listed discussion topics are not elaborated in the available description.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

Amulya Vishwanath, who spent seven years leading product marketing for NVIDIA’s robotics platforms, says robotics is currently in its “Lego era,” with the basic building blocks like actuators, robotic hands, software, and safety standards still being rebuilt. This view runs counter to those who believe that robotics, powered by a general foundation model, is about to have […]