Former Nvidia Robotics Executive Says the Industry Remains Stuck in Its 'Lego Era'
Key Takeaways
- •Former Nvidia robotics executive Amulya Vishwanath described the industry as being in a 'Lego era' because foundational components like actuators, hands, software, and safety protocols are still being developed.
- •Crunchbase tallied $47.4 billion in physical AI investments in the first half of 2026, nearly quadruple the previous half-year's $12 billion, with Waymo, Anduril, Shield AI, and Saronic together receiving more than half of the total.
- •Gartner predicts that fewer than 100 companies will use humanoid robots beyond testing by 2028 and fewer than 20 will move them into production, citing limited dexterity, high integration costs, and short battery life as key hindrances.
- •IDC reported that global humanoid shipments exceeded 18,000 units in 2025, roughly 95% from Chinese manufacturers, but 85% went to performances, education, data collection, and tours rather than factories, and it forecasts more than 510,000 shipments by 2030.
- •Specialized physical AI is already commercially viable, as shown by Waymo's 15 million rides in 2025 and roughly $16 billion raised in February 2026, indicating that near-term opportunities lie in proven automation, sensors, computing, and software.
Amulya Vishwanath, a former robotics executive at Nvidia and founder of venture firm Techable VC, says the robotics industry remains in its "Lego era," even as billions of dollars flow into physical AI. The comment frames the field's central tension: capital is accelerating while the foundational components for real-world deployment are still under construction.
Why she calls it a 'Lego era'
Speaking on an October 16 episode of TechCrunch's Equity podcast, Vishwanath said the sector's fundamental building blocks—actuators, robotic hands, software, and safety protocols—are still being developed. She spent seven years in Nvidia's robotics product marketing before moving into early-stage investing in physical AI.
Expert opinion on the timeline diverges. Pim de Witte, CEO of General Intuition, predicts that a single general-purpose robotics model will displace much of today's specialized robotics work. Moore Threads chairman Zhang Jianzhong offered a more conservative view: "It is certainly still some way from its 'GPT' moment."
Gartner says fewer than 20 will reach production
Gartner predicts that fewer than 100 companies will be using humanoid robots beyond testing by 2028, while fewer than 20 will actually carry them into production across manufacturing and supply chains. The gap between the two thresholds means that, even on Gartner's timeline, testing and pilots will far outnumber production deployments.
"The promise of humanoid robots is compelling, but the reality is that the technology remains immature," Gartner analyst Abdil Tunca said. Limited dexterity, high integration costs, and short battery life remain major hindrances.
18,000 humanoids versus 4.7 million factory robots
Despite growing interest, very few humanoid robots have been deployed in factories. According to IDC, global shipments exceeded 18,000 units in 2025, with Chinese manufacturers accounting for roughly 95% of the total. Even so, 85% of those shipments went to performances, education, data collection, and guided tours rather than industrial production. IDC forecasts that shipments will surpass 510,000 by 2030.
Traditional industrial robots have progressed much further. The International Federation of Robotics reported 542,000 installations in 2024, and the number of industrial robots in operation worldwide has reached 4.7 million. While shipments and installations are not the same measure, the figures illustrate how deeply established factory automation already is.
Physical AI is also advancing outside the factory. Self-driving vehicles are a standout example: Waymo announced it completed 15 million rides in 2025, and the company raised roughly $16 billion in February 2026. Its success shows that specialized AI technology can be commercially viable, even as general-purpose human have yet to find widespread applications.
Where the $47.4 billion is actually going
Crunchbase counted $47.4 billion invested across 5 physical AI deals in the first half of 2026, nearly four times the previous half-year's $12 billion. The category spans vehicles, defense, drones, and sensors—not just humanoids.
Waymo led with $16 billion, followed by Anduril at $5 billion, Shield AI at $2 billion, and Saronic at $1.75 billion. Together, those four rounds total roughly $24.75 billion—more than half of the half-year total—showing how concentrated physical AI capital remains in a small number of large checks. Investors continue to back robot intelligence more broadly: Cryptopolitan reported on Generalist's roughly $600 million funding round and $3 billion valuation.
Techable VC's portfolio, meanwhile, includes autonomy startup Alpha Z, industrial robotics developer Trener Robotics, soft-materials company morph., and Andromeda Surgical.
The near-term winners may be chips and software
Gartner's September research identifies physical AI agents among four trends reshaping warehouses. Rather than betting everything on human-shaped robots, companies can deploy specialized systems today.
For the global AI market, the more immediate opportunity lies in proven automation, sensors, computing, and software. The general-purpose humanoid remains a longer-term wager. The concrete markers from here are already on the record: whether deployments move beyond the 85% of 2025 shipments that went to performances, education, data collection, and guided tours, and whether shipments track toward IDC's forecast of more than 510,000 units by 2030.
Source: Cryptopolitan
