SpaceX AI Compute: The Predictable Revenue Driver?!
Overall Thesis
SpaceX's AI compute revenue is highly predictable and scalable compared to Tesla's longer-term robotaxi and Optimus ventures, making it a consistent and reliable growth driver.
Narratives
The speaker argues SpaceX's AI compute business, alongside Starlink, is a highly predictable revenue and profit driver, with the company on pace to hit a $100 billion ARR by December 2026 as it scales GPU deployment, power, and monetization. He contrasts this predictability favorably against Tesla's harder-to-forecast robotaxi and Optimus businesses.
Key Arguments
- AI compute ARR is expected to hit $100 billion by December 2026, requiring a little over $8 billion of revenue that month.
- AI compute revenue is a predictable function of GPU acquisition, power/cooling buildout, and monetization per gigawatt.
- Starlink revenue is predictable based on ARPU, subscriber growth, bandwidth per launch, and self-service vs prime broadband mix.
- Many more large compute deals are likely in the works to reach the stated ARR target.
The speaker contrasts Tesla's robotaxi and Optimus businesses with SpaceX's predictable AI compute revenue, noting that while the long-term trend for robotaxis and Optimus could be very profitable, the exact timing and trajectory over the next five to ten years are much harder to predict due to regulatory hurdles and unknowns.
Key Arguments
- Robotaxi rollout is cautious with regulatory hurdles and unknown unknowns, making timing hard to predict over a 5-10 year horizon.
- Optimus could be phenomenally profitable at scale, but its growth trajectory and autonomy progress are difficult to forecast.
- Overall trend and shape of these businesses is easier to predict than exact timing.
Hedges & Caveats
- ARR figures represent annualized run rate, not full-year revenue projections
- AI compute revenue is 'not a hundred percent predictable'
- Many deals are still 'in the works' and 'coming soon' with no guaranteed timeline
- Robotaxi and Optimus have 'unknown unknowns' and regulatory uncertainties
- Exact timing of growth trajectories is 'harder to predict' over 5-10 year horizons