Who sees the future clearly, builds it: NVIDIA and the technologies we’re not talking about

Source: Nate’s Newsletter Date: 2025-02-05 URL: https://natesnewsletter.substack.com/p/who-sees-the-future-clearly-builds

Summary

The real Nvidia story isn’t GPU sales for LLMs or CapEx charts — it’s Jensen Huang’s multi-decade bet on robotics, autonomous vehicles, and physical-world AI that most financial analysts miss by focusing on short-term metrics. The piece treats this as a founder story: “I for one am not going to count him out yet,” positioning Nvidia’s competitive advantage as long-term ecosystem building for physical AI rather than current infrastructure dominance.

Implications

Capital thread. The 5-7+ year Nvidia thesis is a different investment thesis than the “GPU scarcity premium” trade: it’s a bet on Nvidia owning the compute stack for embodied AI (robots, autonomous vehicles) the way it owns the stack for LLMs. If physical AI becomes the next major compute demand wave, the current GPU-for-LLMs period looks like the on-ramp.

AI economics thread. Jensen Huang’s long-term vision includes CUDA and the Nvidia ecosystem as the enduring platform layer across AI generations. If that vision is right, the current “Nvidia monopoly” narrative understates the durability of their position — and the DeepSeek efficiency story (less compute per model) is a feature, not a threat, because it expands deployment without reducing total compute demand.

Watch: Whether the physical AI / robotics wave actually materializes as a significant Nvidia revenue driver within the 5-7 year window, and whether any challenger (AMD, custom silicon, Arm-based) makes meaningful inroads in the AI training and inference market before that wave arrives.