Nvidia's Success Boosts 'Physical AI' Theme

Nvidia's record Q4 revenue of $68.1 billion was overwhelmingly driven by its Data Center business, which posted $62.3 billion, a 75% year-over-year increase. CFO Colette Kress directly stated on the earnings call that "physical AI applications built on increasingly smarter and multimodal models are beginning to drive our financial performance." This trend represents a strategic shift from Large Language Models (LLMs) to Large Action Models (LAMs), where AI exits the browser and gets a physical body. The current infrastructure phase involves training robots in massive "Digital Twin" simulations, allowing them to learn complex tasks at 10,000x normal speed before their AI brain is downloaded into a physical chassis. Nvidia's technology, such as the Jetson and IGX Orin platforms, provides the server-grade edge computing power necessary for this transition. This enables real-time, low-latency processing directly on devices like autonomous vehicles and factory robots, a critical requirement for operating in the physical world without constant cloud dependency. Tesla is a key Nvidia customer, utilizing its infrastructure to develop systems like the Optimus robot and its robotaxi fleet. Training these physical systems requires substantially more computing power than training AI chatbots. Elon Musk has framed robotics as a multi-trillion-dollar opportunity, with projections suggesting Optimus alone could generate $25 billion in annual revenue if production reaches one million units. The investment landscape is reflecting this shift, with venture capital funding for robotics hitting $8.8 billion in a single quarter of 2025, a 15-fold increase since 2017. This marks a new wave of AI investment that places equal emphasis on hardware and the physical supply chain, not just the software models. Analysts are modeling significant, though long-term, revenue from these initiatives. One bullish forecast projects Tesla's robotaxi revenue could climb to $250 billion by 2035, assuming 30% AV penetration and a 50% market share for Tesla. However, these ventures are expected to incur near-term losses, with an estimated $500 million gross loss in 2026 as Tesla begins to scale its fleet.