Elon Musk praises Mythos/Fable, promises not to ‘cut off’ Anthropic

Anthropic is paying SpaceX’s xAI unit $1.25B per month for 300MW of compute through May 2029 (~$40B revenue), while Google rents SpaceX infrastructure for $920M per month through June 2029. Elon Musk publicly reversed course from earlier anti-Anthropic comments, saying he would not “cut them off” and praising Anthropic’s current AI leadership (with Mythos/Fable and potential Mythos 2). The article flags contractual risk if infrastructure were ever shut down and suggests possible intelligence/visibility benefits from hosting compute, leaving the long-term competitive impact uncertain.

Analysis

The market read is less about Musk’s tone shift and more about asset economics: long-duration compute leases turn idle frontier capacity into annuity-like cash flow. That matters because it rewards operators that can serve multiple rival labs without obvious degradation, and it gives them a second source of edge beyond raw GPU supply: operational telemetry on workload patterns, latency requirements, and customer behavior.

For listed names, the first-order beneficiaries are indirect. AMZN and GOOGL can get a modest multiple tailwind if investors infer that AI demand is still outrunning supply and that large customers want redundancy across vendors, but this is mostly a 1-3 month narrative catalyst rather than an EPS event. TSLA is not economically exposed here; any move in TSLA from the Musk angle is likely a sentiment trade and should fade unless it spills into actual capital allocation or governance risk.

The contrarian risk is overestimating the permanence of compute scarcity. If model efficiency improves or enterprise demand pauses, these contracts can go from “scarcity premium” to fixed-cost overhang quickly, especially before the 2029 endpoint. The falsifiers are straightforward: weaker AI capex commentary, lower utilization, or any evidence that lease economics are being repriced downward over the next 1-3 quarters.

Key Decisions for Investors

• No immediate TSLA position; treat any Musk-driven gap up as fadeable unless there is a Tesla-specific earnings or margin catalyst.
• Buy AMZN/GOOGL on 2-4 week pullbacks as a small basket long; upside is mainly multiple support from persistent AI capacity scarcity, not near-term revenue revision.