ComputePublished Top storyAll news from that day
Ongoing case: Anthropic's Claude models 20 storiesAnthropic's S-1 lists compute commitments it must generally pay whatever its revenue
Anthropic, the company behind Claude, lists commitments of at least $114.5 billion to Google and $110.0 billion to Amazon, plus up to $120.5 billion with SpaceXAI.
Check our sources · 13 facts from 1 source
Illustration made with AI for ai notis
Key points
- The S-1 says Anthropic committed to spend at least $114.5 billion with Google by July 2033 and at least $110.0 billion with Amazon by April 2036, regardless of actual usage.
- The SpaceXAI arrangements are worth up to $120.5 billion, about $3.1 billion a month once delivered, and either side may cancel on 90 days' notice after January and March 2027.
- AMD's agreement sets no fixed commitment; Anthropic expects over $20.0 billion if it buys full initial capacity. Broadcom leases total about $161.2 billion rent over their life, excluding interest.
What happened
Anthropic's S-1 describes its most significant compute arrangements, which it says generally must be satisfied regardless of its revenue or utilization. It lists commitments to Microsoft, Google, Amazon, SpaceXAI, Nscale and Lambda, a hardware supply agreement with AMD, and five-year master equipment leases with Broadcom. The totals are the filing's own figures for each agreement.
Underlined sentences link to the facts we checked. Dotted words open a short explainer.
Fact 1 of 13 · from U.S. Securities and Exchange Commission, 26 Oct 2026
Our compute commitments are substantial and generally must be satisfied regardless of our revenue or utilization
What it means for you
Our viewIf you depend on Claude for production work, note that the S-1 says these commitments generally must be satisfied regardless of Anthropic's revenue or utilization. The SpaceXAI agreements can be cancelled on 90 days' notice after their initial periods. The AMD figure is Anthropic's expectation, not a commitment.
Ask your Anthropic account team whether committed capacity affects your rate limits or pricing terms, and read the compute section again when the filing is amended.
It adds no new facts.
Your reaction
We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.
We count reactions per story and day, never who reacted. They help choose what goes in the monthly issue.
Check our sources
Every sentence above is checked against this source.
1 Anthropic, PBC Form S-1 registration statement
Open the source archived copy-
The S-1 says Anthropic's compute commitments "are substantial and generally must be satisfied regardless of our revenue or utilization."
citeOur compute commitments are substantial and generally must be satisfied regardless of our revenue or utilization
-
The S-1 says that under a November 2025 agreement with Microsoft, amended in February and October 2026, Anthropic committed to spend at least $44.9 billion between November 2026 and May 2033, regardless of actual usage.
citeIn November 2025, we entered into an agreement with Microsoft, which was subsequently amended in February and October 2026, pursuant to which we committed to spend an aggregate of at least $44.9 billion between November 2026 and May 2033, regardless of actual usage.
-
The S-1 says that under a March 2026 agreement with Google, amended between May and August 2026, Anthropic committed to spend at least approximately $114.5 billion between April 2026 and July 2033, regardless of actual usage.
citeIn March 2026, we entered into an agreement with Google, which was subsequently amended between May and August 2026, and pursuant to which we committed to spend at least approximately $114.5 billion between April 2026 and July 2033, regardless of actual usage.
-
The S-1 says Anthropic estimates aggregate future payments of approximately $125.0 billion to $150.0 billion for additional TPU compute from TPU providers including Google, to be delivered between 2028 and 2031, and that this is not included in the contractual commitments table.
citeWe estimate the aggregate future payments for such additional compute to be approximately $125.0 billion to $150.0 billion, which are not included in the contractual commitments table above.
-
The S-1 says that under an April 2026 agreement with Amazon, Anthropic committed to spend at least $110.0 billion between May 2026 and April 2036, regardless of actual usage.
citeIn April 2026, we entered into an agreement with Amazon, pursuant to which we committed to spend an aggregate of at least $110.0 billion between May 2026 and April 2036, regardless of actual usage.
-
The S-1 says that between May and October 2026 Anthropic entered into agreements with SpaceX.AI LLC (formerly X.AI LLC) to access compute powered by NVIDIA GPUs, under which it agreed to spend an aggregate of up to $120.5 billion over terms extending into 2029.
citeUnder the SpaceXAI Arrangements, we agreed to spend an aggregate of up to $120.5 billion over terms extending into 2029
-
The S-1 says the SpaceXAI arrangements amount to approximately $3.1 billion per month once all capacity has been delivered, which Anthropic expects in December 2026, and that after initial non-cancelable periods ending in January 2027 and March 2027 each agreement may be cancelled by either party on 90 days' written notice.
citeconsisting of approximately $3.1 billion in the aggregate per month once all capacity has been delivered, which we expect in December 2026 ... Following initial non- cancelable periods ending in January 2027 and March 2027 under the respective agreements, each agreement may be cancelled by either party upon 90 days advance written notice.
-
The S-1 says that in August 2026 Anthropic agreed with subsidiaries of Nscale to spend up to approximately $44.6 billion over terms extending into 2033 for compute powered by NVIDIA GPUs.
citeUnder the Nscale Services Agreements, we agreed to spend an aggregate of up to approximately $44.6 billion over terms extending into 2033, with fees accruing only as capacity is delivered to and accepted by us.
-
The S-1 says the Lambda Agreements, entered into in May, June and August 2026, commit Anthropic to spend approximately $41.3 billion over terms extending into 2034 for compute powered by NVIDIA GPUs.
citeUnder the Lambda Agreements, we agreed to spend an aggregate of approximately $41.3 billion over terms extending into 2034, with fees accruing only as capacity is delivered to and accepted by us.
-
The S-1 says that under its Hardware Supply Agreement with AMD, Anthropic committed to obtain an initial tranche of capacity in calendar year 2027 using AMD-based Helios rack-scale systems, provided conditions on deployment proposals are met; the agreement sets no fixed dollar commitment, and Anthropic expects expenditures to exceed $20.0 billion if it procures the full initial capacity.
Toned down to what the source says
citewe committed to obtain an initial tranche of compute capacity in calendar year 2027 by purchasing AMD-based Helios rack-scale systems ... provided certain conditions related to deployment proposals are met. The agreement does not specify a fixed dollar commitment. If we procure the full amount of this initial committed capacity, we expect aggregate expenditures to exceed $20.0 billion
-
The S-1 says Anthropic has entered into five-year master equipment leases with Broadcom, accounted for as finance leases, with rent over the life of the leases of approximately $161.2 billion excluding interest.
citeWe have entered into five-year master equipment leases with Broadcom ... These Lease Arrangements are accounted for as finance leases. Our rent payments over the life of these leases are equal to an aggregate of approximately $161.2 billion excluding interest
-
The S-1 says the Broadcom leases provide access to TPU-based compute capacity, a portion of which began to come online in 2026, with additional capacity expected from 2027.
citeThe Lease Arrangements provide us with access to TPU-based compute capacity, a portion of which began to come online in 2026, with additional capacity expected to come online starting in 2027.
-
The S-1 says: "The training of Claude models has required, and we expect will continue to require, substantial compute."
citeOur ability to develop, train, improve, and deploy frontier models depends on our access to enormous and increasing quantities of specialized compute. The training of Claude models has required, and we expect will continue to require, substantial compute.
Every sentence above is checked against this source. Each fact has its own address you can cite.
Source 1 · U.S. Securities and Exchange Commission · 26 Oct 2026
Anthropic, PBC Form S-1 registration statement
The S-1 says Anthropic's compute commitments "are substantial and generally must be satisfied regardless of our revenue or utilization."
Our compute commitments are substantial and generally must be satisfied regardless of our revenue or utilization
The S-1 says that under a November 2025 agreement with Microsoft, amended in February and October 2026, Anthropic committed to spend at least $44.9 billion between November 2026 and May 2033, regardless of actual usage.
In November 2025, we entered into an agreement with Microsoft, which was subsequently amended in February and October 2026, pursuant to which we committed to spend an aggregate of at least $44.9 billion between November 2026 and May 2033, regardless of actual usage.
The S-1 says that under a March 2026 agreement with Google, amended between May and August 2026, Anthropic committed to spend at least approximately $114.5 billion between April 2026 and July 2033, regardless of actual usage.
In March 2026, we entered into an agreement with Google, which was subsequently amended between May and August 2026, and pursuant to which we committed to spend at least approximately $114.5 billion between April 2026 and July 2033, regardless of actual usage.
Show all 13 factsShow fewer facts
The S-1 says Anthropic estimates aggregate future payments of approximately $125.0 billion to $150.0 billion for additional TPU compute from TPU providers including Google, to be delivered between 2028 and 2031, and that this is not included in the contractual commitments table.
We estimate the aggregate future payments for such additional compute to be approximately $125.0 billion to $150.0 billion, which are not included in the contractual commitments table above.
The S-1 says that under an April 2026 agreement with Amazon, Anthropic committed to spend at least $110.0 billion between May 2026 and April 2036, regardless of actual usage.
In April 2026, we entered into an agreement with Amazon, pursuant to which we committed to spend an aggregate of at least $110.0 billion between May 2026 and April 2036, regardless of actual usage.
The S-1 says that between May and October 2026 Anthropic entered into agreements with SpaceX.AI LLC (formerly X.AI LLC) to access compute powered by NVIDIA GPUs, under which it agreed to spend an aggregate of up to $120.5 billion over terms extending into 2029.
Under the SpaceXAI Arrangements, we agreed to spend an aggregate of up to $120.5 billion over terms extending into 2029
The S-1 says the SpaceXAI arrangements amount to approximately $3.1 billion per month once all capacity has been delivered, which Anthropic expects in December 2026, and that after initial non-cancelable periods ending in January 2027 and March 2027 each agreement may be cancelled by either party on 90 days' written notice.
consisting of approximately $3.1 billion in the aggregate per month once all capacity has been delivered, which we expect in December 2026 ... Following initial non- cancelable periods ending in January 2027 and March 2027 under the respective agreements, each agreement may be cancelled by either party upon 90 days advance written notice.
The S-1 says that in August 2026 Anthropic agreed with subsidiaries of Nscale to spend up to approximately $44.6 billion over terms extending into 2033 for compute powered by NVIDIA GPUs.
Under the Nscale Services Agreements, we agreed to spend an aggregate of up to approximately $44.6 billion over terms extending into 2033, with fees accruing only as capacity is delivered to and accepted by us.
The S-1 says the Lambda Agreements, entered into in May, June and August 2026, commit Anthropic to spend approximately $41.3 billion over terms extending into 2034 for compute powered by NVIDIA GPUs.
Under the Lambda Agreements, we agreed to spend an aggregate of approximately $41.3 billion over terms extending into 2034, with fees accruing only as capacity is delivered to and accepted by us.
The S-1 says that under its Hardware Supply Agreement with AMD, Anthropic committed to obtain an initial tranche of capacity in calendar year 2027 using AMD-based Helios rack-scale systems, provided conditions on deployment proposals are met; the agreement sets no fixed dollar commitment, and Anthropic expects expenditures to exceed $20.0 billion if it procures the full initial capacity.
Toned down to what the source sayswe committed to obtain an initial tranche of compute capacity in calendar year 2027 by purchasing AMD-based Helios rack-scale systems ... provided certain conditions related to deployment proposals are met. The agreement does not specify a fixed dollar commitment. If we procure the full amount of this initial committed capacity, we expect aggregate expenditures to exceed $20.0 billion
The S-1 says Anthropic has entered into five-year master equipment leases with Broadcom, accounted for as finance leases, with rent over the life of the leases of approximately $161.2 billion excluding interest.
We have entered into five-year master equipment leases with Broadcom ... These Lease Arrangements are accounted for as finance leases. Our rent payments over the life of these leases are equal to an aggregate of approximately $161.2 billion excluding interest
The S-1 says the Broadcom leases provide access to TPU-based compute capacity, a portion of which began to come online in 2026, with additional capacity expected from 2027.
The Lease Arrangements provide us with access to TPU-based compute capacity, a portion of which began to come online in 2026, with additional capacity expected to come online starting in 2027.
The S-1 says: "The training of Claude models has required, and we expect will continue to require, substantial compute."
Our ability to develop, train, improve, and deploy frontier models depends on our access to enormous and increasing quantities of specialized compute. The training of Claude models has required, and we expect will continue to require, substantial compute.
Topics
The morning email
On the mornings we publish: the three top stories and up to four short ones. Free.
We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep
