Amulya Vishwanath on the Evolving Landscape of Physical AI

Robotics in the Lego Era

Amulya Vishwanath, a former product marketing lead for robotics at Nvidia, argues that the robotics sector is currently in its foundational stage rather than on the verge of a sudden breakthrough. While some observers suggest that robotics is approaching a singular, transformative moment similar to the launch of ChatGPT, Vishwanath suggests that the industry is still in a phase analogous to the construction of Lego sets. In her view, essential components such as robotic hands, safety standards, actuators, and software are still being actively redeveloped and refined.

Contrasting Industry Perspectives

This assessment stands in direct opposition to the claims made by other industry figures, such as General Intuition CEO Pim de Witte. De Witte has previously suggested that a single general foundation model could render much of the current specialized robotics work obsolete, effectively triggering an immediate industry-wide evolution. The Scale Report notes that these conflicting narratives highlight a deeper debate regarding whether physical AI will be driven by monolithic models or modular hardware improvements.

The Shift to Early Stage Investing

Following her seven-year tenure at Nvidia, Vishwanath has transitioned into the venture capital space as the founder and managing partner of Techable VC. She is currently deploying capital into the broader field of physical AI, which she defines as encompassing robotics, materials science, and life sciences. The move positions her as one of the few former Nvidia executives to pivot directly into early-stage investment within these highly technical domains.

Context for the Physical AI Market

Why this matters: As AI moves from digital interfaces into the physical world, the tension between software-centric breakthroughs and hardware reliability becomes the central constraint for investors. Vishwanath's focus on materials and life sciences suggests a belief that physical AI is a multi-disciplinary challenge that cannot be solved by language models alone. This approach represents a growing skepticism among hardware-native investors toward the hype surrounding pure foundation models.