
Musk Would Take Anthropic’s Compute Back. Can SpaceX Earn as Much From It?
By Dalton Skach · 2 min read
Elon Musk says SpaceX would be happy to take back the computing capacity Anthropic reportedly plans to stop using after mid-2027.
“We would be happy to take the compute back. No problem,” Musk wrote on X at 12:40 p.m. ET, adding that he thinks the opposite will happen. He was responding to reports based on secondhand notes from an analyst meeting. Anthropic’s reported decision remains unconfirmed.1
My read: the risk for SpaceX is replacing Anthropic’s revenue on equally good terms. Finding a use for the machines is a lower bar. Musk’s willingness to take them back suggests confidence in demand, but demand alone doesn’t establish what the capacity will earn.
The exit terms matter more than the headline
Anthropic’s IPO filing describes agreements worth up to $120.5 billion with SpaceX’s AI business through 2029. That figure is a ceiling, not guaranteed revenue.2
Once the applicable locked-in periods end in January and March 2027, either side can cancel with 90 days’ notice. The relationship therefore has to keep making economic sense well before 2029.
That is where I’d focus. A large contract gets attention when it is signed. The customer’s decision to stay after it can leave tells us more about the durability of the business.

A new use needs to replace the economics
If another customer takes the capacity at comparable prices with little downtime, SpaceX could absorb an Anthropic exit without much disruption. A better-paying customer could make the change attractive.
Using the capacity internally is a different proposition. SpaceX would gain resources for its own AI business, but give up outside customer revenue. Its products would then have to earn enough to justify that trade.
And if SpaceX needs to discount the capacity or wait for a replacement, the machines could remain useful while becoming less profitable. Strong industry demand would not erase that cost.

I’d judge this on replacement pricing, downtime, and the length of the next commitment. Those are the details that would turn Musk’s confidence into a business case.
The next news to watch is Anthropic’s direct account of its plans. But the investment question is already clear: can SpaceX earn as much from this capacity without Anthropic? Until there is evidence that it can, I’d treat an early exit as a meaningful revenue risk, even if Musk welcomes the compute back.
This article reflects Dalton Skach’s views as of November 11, 2026. It provides general research and does not offer personalized investment advice or recommend buying, selling, or holding a security. As of November 8, 2026, Dalton owns shares in Tesla and SpaceX. Editorial Standards & Disclosures · Terms of Use
Sources and calculations
- Elon Musk, X, November 11, 2026. Original post retrieved through X’s syndication endpoint. Timestamp 17:40:21 UTC, or 12:40 p.m. Eastern. The reply is to Aaron Burnett’s discussion of secondhand analyst-meeting notes. It establishes Musk’s response, not confirmation of Anthropic’s intentions. ↩
- Anthropic, PBC, amended Form S-1, November 10, 2026. Accession 0001628280-26-072484, page 180, Infrastructure Service Arrangements, SpaceXAI paragraph. Agreements up to $120.5 billion extending into 2029; initial non-cancelable periods end January and March 2027; either party may cancel with 90 days’ advance written notice. The filing does not confirm the reported analyst-meeting remarks. ↩
Research cutoff: November 11, 2026. Replacement pricing, downtime, internal use and revenue consequences are BTF analysis, not company forecasts or announced replacement contracts. Original AI-generated conceptual photography and sculpted illustrations are not documentary images of an actual facility. Generation prompts and provenance are retained with the working files.


