xAI’s Compute Deal with Anthropic: A Shifting Narrative on Duration and Commitment
In a recent development that has raised questions within the technology sector, xAI, the artificial intelligence venture spearheaded by Elon Musk, appears to be recalibrating public statements regarding its significant compute deal with AI research firm Anthropic. Initially reported as a multi-year commitment, the precise terms and duration of this arrangement are now subject to scrutiny following remarks from Musk himself.
The Initial Agreement and its Implications
Earlier this month, xAI reportedly secured a substantial compute capacity agreement with Anthropic, involving billions of dollars monthly for exclusive access to Anthropic’s “Colossus” cluster. Such a transaction was positioned as mutually beneficial: providing xAI with critical computational resources necessary for its ambitious AI development, and bolstering Anthropic’s revenue stream amidst the intense competition for AI hardware and infrastructure. This deal was widely seen as a strategic move for xAI to accelerate its progress in the AI landscape.
Musk’s Contradictory Statement
However, Elon Musk recently took to the social platform X to offer a different perspective on the agreement’s commitment. He stated that SpaceX “has not committed to leasing Colossus for years,” characterizing the arrangement as a “180 day lease with 90 day notice mutual cancellation thereafter.” Musk emphasized that the short-term nature of the lease was xAI’s preference, not Anthropic’s, and indicated a willingness to provide a “reasonable off-ramp” while reserving the option to reclaim capacity if “compute gets super tight.”
Discrepancy with Regulatory Filings
This account contrasts sharply with disclosures made in SpaceX’s recent S-1 filing. The filing explicitly details a cloud services agreement with Anthropic PBC, stipulating that “the customer has agreed to pay a monthly fee through May 2029.” While acknowledging a standard 90-day termination clause for either party, the filing clearly frames the engagement as a three-year arrangement, with capacity beginning to ramp up in May 2026 at an initially reduced fee. Similar language, including a specific monthly payment figure of “$1.25 billion per month through May 2029,” is reiterated across multiple sections of the document, suggesting a deliberate contractual structure rather than a minor oversight.
Analysis of the Discrepancy
The divergence between Musk’s public statements and the formal regulatory filing raises significant questions about the definitive terms of the compute deal. While contractual definitions of “lease” can sometimes be nuanced, the S-1 filing’s explicit mention of payment obligations extending to May 2029 strongly implies a committed, multi-year arrangement. The implication is that either the description of the lease duration is being strategically downplayed, or the short-term, flexible arrangement Musk described is somehow reconciled with the longer-term payment commitment outlined in the S-1. The lack of immediate clarification from xAI or Anthropic leaves the precise nature of this critical infrastructure agreement open to interpretation.
Business Style Takeaway: The apparent discrepancy between public statements and regulatory filings concerning xAI’s compute deal with Anthropic highlights the critical importance of transparency and consistency in corporate communications, particularly during security offerings. Businesses and investors must scrutinize such announcements for potential misrepresentations that could impact valuation and strategic decision-making within the high-stakes AI infrastructure market.
Details can be found on the website : techcrunch.com