SIAgentNews.com
← All news

Anthropic enters $120.5 billion compute deal with SpaceX

The AI chatbot maker is partnering with Elon Musk's SpaceX to secure massive computing power for its models, despite previous public friction between the two.

· First reported by Moneywise on MSN

$120.5 billionMaximum value of the SpaceX compute deal

The short version

  • Anthropic's IPO raised $81.3 billion, valuing the company at approximately $1.5 trillion.
  • The deal with SpaceXAI covers compute costs through 2029, reaching a monthly tab of about $3.1 billion.
  • The agreement includes a 90-day exit clause for both companies starting in early 2027.
  • SpaceX reported that 34.5% of its third-quarter revenue came from an unnamed customer, identified by reports as Anthropic.

Why it matters

This deal shows that despite public rivalry, the massive demand for AI hardware is forcing competitors to become each other's primary suppliers. It also signals a shift toward long-term, multi-billion dollar infrastructure commitments to sustain the growth of large language models.

A massive infrastructure partnership

Anthropic, the company responsible for the Claude AI chatbot, has officially begun trading on the Nasdaq. The company's initial public offering raised roughly $81.3 billion, a figure that surpassed the $75 billion raised by SpaceX during its own IPO in June. This valuation places Anthropic at approximately $1.5 trillion according to reports from NPR and the Associated Press.

As part of its growth strategy, Anthropic has agreed to spend up to $120.5 billion with SpaceXAI, the AI unit formerly known as xAI. This massive financial commitment is intended to secure the necessary compute power required to train and operate the company's large-scale AI models. The agreement is structured to run into the year 2029, providing a long-term roadmap for the company's technical expansion.

The scale of this deal is underscored by the monthly costs involved. Once the full promised capacity is delivered, which Anthropic expects to happen in December, the monthly expenditure is projected to reach about $3.1 billion. This highlights the intense capital requirements currently facing the leading developers of generative artificial intelligence.

The mechanics of AI compute

In the world of artificial intelligence, compute refers to the specialized hardware, power, and data center space required to run complex models. For Anthropic, this means securing a steady supply of high-end chips and the electricity to run them. The deal with SpaceX provides access to massive facilities, including the Colossus 1 data center in Memphis, Tennessee.

This specific facility offers more than 300 megawatts of power and over 220,000 Nvidia graphics processing units. These chips are the primary engines for AI development, performing the heavy lifting for model training. By leveraging SpaceX's existing infrastructure, Anthropic can scale its operations without having to build every piece of hardware from scratch.

The partnership has already begun to show significant results in the financial records of both companies. Between May and October, the two sides signed and amended several agreements for Nvidia chips owned by SpaceX. The current monthly figures for these arrangements are now close to 2.5 times the rate that was originally listed in SpaceX's initial IPO prospectus.

From public rivalry to private profit

The business relationship between Anthropic and Elon Musk is particularly notable given their recent public history. In February, Musk accused the Claude chatbot of being biased against certain groups, calling the behavior misanthropic and evil. He even mocked the company's name, which is derived from the Greek word for human, suggesting it was ironic for the company to be perceived as anti-human.

However, the S-1 registration statement filed with the SEC reveals that Musk has a significant financial stake in Anthropic's success. Despite his public criticisms, his company is becoming a primary service provider for his competitor. This transition from public hostility to a multi-billion dollar commercial agreement illustrates the pragmatic realities of the AI hardware market.

The financial impact of this partnership is already visible in SpaceX's quarterly filings. The company attributed 34.5% of its third-quarter revenue to an unnamed 'Customer B.' Reports from Bloomberg have since identified this customer as Anthropic. This suggests that SpaceX is becoming a major player in the infrastructure side of the AI boom, serving its biggest rivals.

Contractual safeguards and growth

While the $120.5 billion figure is a significant ceiling, the contract includes specific protections for both parties. The agreements include an initial non-cancelable period that ends in early 2027. After that time, either Anthropic or SpaceX can terminate the deal with 90 days' written notice, allowing for flexibility if market conditions change.

This flexibility is important because the cost of compute is subject to volatility. If the price of specialized chips and data center space falls, Anthropic may have the option to walk away to find cheaper alternatives. Conversely, if prices rise significantly, SpaceX could use the exit clause to seek higher rates from other customers or renegotiate terms.

SpaceX leadership remains optimistic about the growing demand for these resources. During a recent earnings call, the company's CFO stated that new compute deployments are paying for themselves in less than a year. He noted that many customers are currently asking for compute, and the company is actively building more supply to meet both internal and external needs.

The broader infrastructure landscape

Anthropic is not relying solely on SpaceX for its long-term infrastructure goals. The company is also developing a program called Anthropic Direct. Under this initiative, partners will build their own data centers and then lease the capacity back to Anthropic. The projected rent for these five-year leases is estimated at $125.2 billion, excluding interest.

When combining these various commitments, Anthropic's total infrastructure obligations are substantial. According to an amended filing from November 10, the company's total commitments sit at about $900.9 billion. Nearly all of this spending is dedicated to the compute and infrastructure necessary to maintain its position in the AI market.

This massive spending indicates that the AI industry is entering a phase of heavy industrialization. The focus is shifting from just developing software to securing the physical backbone of the internet. For the reader, this means that the future of AI will be defined as much by who owns the power plants and chips as by who writes the best code.

What to watch

  • Whether Anthropic exercises its 90-day exit clause if compute prices drop.
  • How many other 'Customer B' style deals SpaceX will sign with AI rivals.
  • The progress of the Anthropic Direct data center leasing program.

Try it

Check the SEC's EDGAR database for Anthropic's S-1 filing to see the full breakdown of their infrastructure commitments.

Sources and links

First reported by Moneywise on MSN. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news