Beyond software, the next frontier is "physical AI," encompassing robotics and autonomous systems. According to analyst Dan Ives, NVIDIA and Tesla are the two clear frontrunners poised to dominate this space, which represents a massive, multi-trillion-dollar market opportunity in everything from cars to humanoid robots.

Today's AI is largely text-based (LLMs). The next phase involves Visual Language Models (VLMs) that interpret and interact with the physical world for robotics and surgery. This transition requires an exponential, 50-1000x increase in compute power, underwriting the long-term AI infrastructure build-out.

While the US outspends China 12-to-1 on compute for LLMs, China invests 42% more in robotics. This focus on "physical AI"—robots that perceive, think, and act—creates a distinct competitive lane where China is building hardware and software advantages over the West.

Unlike cars, which gather data passively, humanoid robots need active training. To solve this, Musk's strategy is to build a physical 'academy' of 10,000-30,000 Optimus robots performing self-play on various tasks, using this real-world data to close the 'sim-to-real' gap from millions of simulated robots.

Ken Goldberg quantifies the challenge: the text data used to train LLMs would take a human 100,000 years to read. Equivalent data for robot manipulation (vision-to-control signals) doesn't exist online and must be generated from scratch, explaining the slower progress in physical AI.

Beyond selling chips, NVIDIA strategically directs the industry's focus. By providing tools, open-source models, and setting the narrative around areas like LLMs and now "physical AI" (robotics, autonomous vehicles), it essentially chooses which technology sectors will receive massive investment and development attention.

AI's computational needs are not just from initial training. They compound exponentially due to post-training (reinforcement learning) and inference (multi-step reasoning), creating a much larger demand profile than previously understood and driving a billion-X increase in compute.

Unlike LLMs trained on vast digital text, humanoid robots need immense amounts of real-world physical data to learn simple tasks. It's estimated that 100 million hours—over 11,000 years' worth—of interaction data is needed to create truly smart, useful humanoids, highlighting the scale of the challenge.

Brett Adcock states that Figure AI's "Helix 2" neural net provides the right technical stack for general robotics. The biggest remaining obstacle is not hardware but the immense data required to train the robot for a wide distribution of tasks. The company plans to spend nine figures on data acquisition in 2026 to solve this.