Anthropic’s $34.9 billion August compute agreement with Lambda can now be ended by either side. The amended S-1 Anthropic filed with the SEC on 10 November says Lambda missed a deadline to secure the data center lease, so the agreement “is currently terminable by either party”. Neither side has given notice, and Anthropic says it does not intend to terminate.
The August agreement is 84.5% of the $41.3 billion Anthropic has committed to Lambda, and equal to 3.9% of the $900.9 billion of total committed spend in the same filing, by our arithmetic.
Business Pill · Financial Conditions Explained
A short explainer on financial conditions, how the price and availability of money move markets. It teaches the general idea only and does not discuss Anthropic or Lambda.
The key insight: Anthropic’s S-1/A of 10 November says its $34.9 billion August agreement with Lambda is now terminable by either party, because Lambda did not secure the facility lease by the deadline. Neither side has given notice. The agreement is 84.5% of Anthropic’s $41.3 billion of Lambda commitments, by our arithmetic.
The 2 November S-1/A described the termination right as a condition that had not yet been tested: “Under our August 2026 agreement, which represents approximately $34.9 billion of the aggregate commitment, either party may terminate if Lambda does not timely secure the facility lease”.
Eight days later, the 10 November filing says the condition has been triggered: “either party may terminate if Lambda does not secure the facility lease by a specific date. Lambda did not secure the facility lease by such date and, as a result, the August 2026 agreement is currently terminable by either party”.
It then adds: “Neither party has delivered notice of termination as of the date of this prospectus, and we do not currently intend to terminate the agreement”. The filing does not name the facility or give the date that was missed.
The rest of the Lambda wording is unchanged in substance. Anthropic signed the three Lambda agreements in May, June and August 2026 “to access compute powered by NVIDIA GPUs”, with “fees accruing only as capacity is delivered to and accepted by us”, over terms extending into 2034. Apart from the August clause, each agreement is non-cancelable except for Lambda’s uncured material breach or other limited cases, and some can be cut tranche by tranche if Lambda misses delivery milestones.
Anthropic’s August 2026 agreement with Lambda against its other Lambda agreements, its contracts subject to termination rights or contingent conditions, and its total committed spend, from the S-1/A filings of 10 and 2 November 2026. Percentages are our arithmetic.
The Wider Pool of Cancelable Deals
The same section puts a figure on contracts that are not yet firm. Today’s filing says “Approximately $186.2 billion of the aggregate total of our contractual obligations relates to arrangements that are subject to either exercisable termination rights or contingent conditions prior to commencement”, such as a counterparty’s ability to secure financing, facilities or regulatory approvals. On 2 November that figure was $213.8 billion.
That is $27.6 billion, or 12.9%, lower in eight days, by our arithmetic. The filing does not break down the change, so it does not show whether the Lambda agreement sits in the new figure. Total committed spend rose over the same period, from $895.4 billion to $900.9 billion.
What Lambda Has Said
No public statement from Lambda or Anthropic about the missed lease date or the termination right is on record. A search of Google News and of Lambda’s newsroom on 10 November found none. When the August deal was first reported, at $35 billion, Data Center Dynamics summarised the Wall Street Journal’s report as saying it covered “350MW of capacity at a data center under development in Nueces County, Texas”, and that “Nvidia will actually hold the lease on the data center, according to WSJ”. The S-1/A does not confirm either detail.
Lambda’s own recent statements are about building and financing. On 4 November it announced a planned campus in Lowndes County, Alabama, which it calls “an estimated $33.1 billion investment in Alabama”, with construction expected to begin in the first quarter of 2027. That announcement does not mention Anthropic, and the S-1/A does not say where the August capacity was to be housed, so nothing on record links the two.
On financing, Lambda’s investor page says it “has raised over $1.5B in equity” and lists three senior secured debt facilities this year: $1 billion in May, $926 million in August and $1 billion in October. Those add up to about $2.9 billion; the $34.9 billion August agreement is about 11.9 times that, by our arithmetic. Lambda said the October facility funds “three committed customer deployments with two investment-grade offtakers across multiple data centers”; it did not name them.
Where It Fits Among the GPU Clouds
The Lambda clause is one of the exits we tallied when the first S-1 came out: Anthropic’s three GPU cloud deals, with SpaceXAI, Nscale and Lambda, add up to as much as about $206.4 billion, and each carries termination rights. Those contracts are with the providers, not with Nvidia, as our placement piece notes; Nvidia is buying $10.0 billion of Anthropic stock in the concurrent placement.
The 10 November filing also changes its vocabulary. Its glossary now defines super intelligence, or SI, and says that “to avoid confusion” it uses super intelligence, SI, artificial intelligence and AI “interchangeably”. It then cites Executive Order 14434, “which directs federal executive departments and agencies to use the terms” Super Intelligence and SI in place of Artificial Intelligence and AI.
The order is real and on the record. Executive Order 14434, “Inaugurating the Era of Super Intelligence”, was signed on 29 September 2026 and published in the Federal Register on 2 October. Its text tells agencies to use the new terms “in official correspondence, public communications, websites, reports, policy documents” and other non-statutory documents within the executive branch. For now it defines Super Intelligence as the technologies covered by the existing statutory definition of artificial intelligence, in section 9401(3) of title 15 of the US Code.
In our count of the text of both filings, “frontier AI” falls from 113 mentions on 2 November to zero; “frontier intelligence” rises from 2 to 65, “super intelligence” from 1 to 27, and SI from 0 to 42. The new definition of super intelligence uses, word for word, the same wording the 2 November filing used to define AI. Lambda already describes itself as “the Superintelligence Cloud”.
The Structural Read
As we read the two filings, the August clause moved from a condition to a live option in eight days: on 2 November either party could terminate if the lease was not secured in time; on 10 November the filing says it was not, and the option is open to both.
As we read it, the exit runs both ways: Anthropic says it does not intend to terminate, but the filing gives Lambda the same right, and it says nothing of Lambda’s intention.
Anthropic S-1/A, 10 November 2026
“Neither party has delivered notice of termination as of the date of this prospectus, and we do not currently intend to terminate the agreement”
Three Implications
FOR ANTHROPIC The filing puts contracts subject to exercisable termination rights or contingent conditions at about $186.2 billion, down from $213.8 billion on 2 November; it does not break down the change.
FOR LAMBDA Its recent public statements cover a planned $33.1 billion Alabama campus and three debt facilities totalling about $2.9 billion this year, by our arithmetic; none mentions Anthropic.
FOR THE TERMINOLOGY The filing now uses super intelligence and AI interchangeably and cites Executive Order 14434, which the Federal Register published on 2 October 2026.
The framework argues: “The $1T capex cycle is being underwritten by a $1T concentrated demand commitment from two counterparties whose combined free cash flow is deeply negative”.
As we read it, the Lambda clause shows the other side of that counterparty question: here the condition sat with the supplier, which had to secure a facility lease, and the filing says it did not. The framework predates this filing and does not discuss it.
What Is Not Established
The filing does not say which facility Lambda failed to lease, why, or when the deadline fell. It does not say whether the two companies are renegotiating, or whether Lambda intends to terminate; only Anthropic’s own intention is stated. It does not explain the $27.6 billion fall in contracts subject to termination rights or contingent conditions. The executive order’s text addresses executive departments and agencies; the filing does not say why Anthropic adopted the terms in its own prospectus.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
A $34.9 billion compute agreement that on 2 November could be ended if Lambda failed to secure a lease can now be ended by either party, because that lease was not secured by the deadline. Neither side has acted, Anthropic says it does not intend to, and the deal is 84.5% of Anthropic’s Lambda commitments and equal to 3.9% of its $900.9 billion total, by our arithmetic.
A note on sourcing. Contract terms, commitments and quotations are from Anthropic’s S-1/A filed with the SEC on 10 November 2026; earlier wording and figures are from its S-1/A of 2 November 2026. The executive order is quoted from the Federal Register; Lambda’s statements are from its own website. Sums, ratios, percentages and term counts are our arithmetic. Neither Anthropic nor Lambda was contacted for this piece.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
