Nvidia Veteran Defines Physical AI as More Than Robots

Overview

The narrative surrounding Physical AI is undergoing a significant revision, moving beyond the flashy imagery of humanoid robots to encompass a broader spectrum of tangible industries. Amulya Vishwanath, a veteran who spent seven years leading product marketing for NVIDIA’s robotics platforms, argues that the sector is currently in a foundational phase she describes as its "Lego era."

In this view, the basic building blocks—ranging from actuators and robotic hands to software frameworks and safety standards—are still being constructed and refined. This perspective directly challenges the prevailing optimism that a single general foundation model will instantly replicate the "ChatGPT moment" for physical machinery. Instead, Vishwanath suggests that the transformation will be incremental, deep, and applicable far beyond traditional manufacturing floors.

Redefining Physical AI

Vishwanath, now the founder and managing partner of Techable VC, insists that defining physical AI solely through the lens of robotics is reductive. During a recent discussion on TechCrunch’s Equity podcast, she outlined a much wider scope for what constitutes physical artificial intelligence.

"Physical AI is not just about robots," Vishwanath explained. "It is about a molecule, a material, a medicine, or a machine on a factory line. It is about applying intelligent systems to anything that exists in the physical world."

This expanded definition includes advancements in materials science and life sciences, sectors that are often overlooked in conversations dominated by hardware and automation. By viewing AI as a tool for manipulating matter at the molecular or structural level, the investment landscape shifts from merely building better bots to solving complex problems in drug discovery, sustainable materials, and precision manufacturing.

The "Lego era" analogy underscores the current state of the industry. While early building blocks are available, they are not yet standardized or universally compatible. Actuators, sensors, and control software are still being rebuilt to meet the rigorous demands of real-world deployment. This implies that the path forward is not a sudden leap to general-purpose robot workers, but a steady accumulation of specialized, functional components.

The "ChatGPT Moment" Debate

A significant point of contention in the current tech discourse is whether a universal foundation model will render specialized robotics obsolete. Pim de Witte, CEO of General Intuition, previously suggested to Equity that a single general foundation model would make much of today’s specialized robotics work redundant. The assumption is that if an AI can understand the world conceptually, it can control any physical system without task-specific programming.

Vishwanath’s tenure at NVIDIA and her current investment thesis offer a counter-narrative. While she acknowledges the power of foundational models, her focus remains on the operational and engineering realities of deploying AI in physical environments. The challenge is not just intelligence, but the ability to interact safely and efficiently with unstructured, messy real-world conditions.

This distinction is crucial for investors and founders. It suggests that the "moment" for physical AI will not be a singular launch event, but a gradual integration into diverse industries. The value lies in solving specific, high-friction problems rather than waiting for a magical generalist agent to solve everything.

Investment Strategy and "Conscious Automation"

As a venture capitalist, Vishwanath is writing first checks where she sees the potential for conscious automation—a term she uses to describe AI systems that are not only autonomous but also context-aware and purpose-driven. She believes the best opportunities lie in solving critical labor shortages and operational bottlenecks.

Key sectors she highlights include:

• Family-run factories that struggle to find workers. For these businesses, AI-driven automation is not a luxury but a survival mechanism.
• Surgeons facing burnout. Here, physical AI can assist in precision tasks, reducing the cognitive and physical load on medical professionals.
• Materials and pharmaceutical development, where AI can accelerate the discovery process by simulating molecular interactions.

Vishwanath emphasizes that humanoids should be good for more than dancing. Their true value will be realized when they can navigate dangerous, dirty, or difficult jobs that humans are increasingly unwilling or unable to perform.

The Role of VC and Global Dynamics

Drawing on her experience as one of the first women executives to spin out of NVIDIA, Vishwanath advises founders to view their venture capital firm as part of their competitive moat. Her network, built over years at NVIDIA, is instrumental in her due diligence process and in providing customer introductions to portfolio companies.

She also offered insights into the geopolitical landscape of AI development. Vishwanath noted that both the United States and China hold distinct advantages in the physical AI race. The U.S. tends to lead in foundational software models and deep-tech research, while China excels in manufacturing scale and rapid deployment of hardware solutions.

For investors and entrepreneurs, the takeaway is that success in physical AI requires a dual focus: robust software intelligence and practical, scalable hardware implementation. The era of "Lego" construction is still underway, but the structures being built will define the next decade of industrial and scientific progress.