IPO Approaches: Anthropic Spends $517 Billion on Computing Power, Still Trails OpenAI
Original Title: "Arms Race Before IPO! Anthropic Locks in $517 Billion Computing Power in 11 Months, Yet Still Lags Behind OpenAI"
Original Author: Long Yue, Wall Street Insights
Anthropic has signed power agreements worth a staggering $517 billion over the past 11 months. The direct catalyst for this procurement frenzy is the explosive demand generated by the growth of its products, Claude Code and Cowork, this year.
On September 6, according to a comprehensive analysis by The Information of public statements and previous reports, since October last year, Anthropic has signed agreements securing at least 14.8 gigawatts of computing capacity for use over the coming years. This is in addition to the 1 to 2 gigawatts already secured, further expanding its total scale.
This figure far exceeds the plans disclosed to investors by Anthropic in December 2025, when the company stated it would spend $180 billion on server rentals by 2029. Based on the currently announced agreements, total expenditures over the next decade could reach as high as $517 billion, covering cloud capacity leasing, chip procurement, and data center leases.
It is worth noting that Anthropic may have more undisclosed agreements, and the capacity secured by signed agreements may not all be utilized.
Additionally, although Anthropic's annual revenue has surpassed $65 billion, exceeding OpenAI's $40 billion, its computing power reserves still lag behind OpenAI's planned target of 30 gigawatts.
Core Suppliers: Amazon and Google Dominate
Anthropic's early partners and investors, Amazon and Google, remain its largest computing power suppliers, collectively providing 11 gigawatts of capacity, with costs exceeding $300 billion over the next decade.
In the past year, Anthropic has also extended its reach to partners that were previously more closely associated with OpenAI:
Microsoft: In November last year, Anthropic announced an agreement with this long-time supporter of OpenAI, committing to spend at least $30 billion to lease approximately 1 gigawatt of Microsoft Azure servers (equipped with Nvidia chips).
SpaceX: In May this year, Anthropic announced a contract with Elon Musk's SpaceX, paying $1.25 billion monthly to lease AI servers, with the agreement lasting until May 2029, totaling a potential cost of up to $45 billion.
Lambda and Nscale: Just last month, Anthropic signed a cloud contract worth $80 billion with startup Lambda and UK-based Nscale, securing at least 460 megawatts of capacity for six years.
Building Own Data Centers: Betting on Long-term Control
In addition to leasing, Anthropic is also accelerating the layout of its own data centers.
In November last year, Anthropic announced plans to jointly invest $50 billion with cloud computing startup Fluidstack to build data centers in Texas and New York, with some facilities already in operation this year.
Additionally, Anthropic has signed a lease with former Bitcoin miner TeraWulf, which is developing an AI data center in Kentucky that will ultimately provide 401 megawatts of computing power.
On the chip front, Anthropic signed an agreement with Google in April this year to procure multi-gigawatt scale tensor processing units (designed by Broadcom); in July, it added a 2-gigawatt order for AMD chips.
Gap with OpenAI: Revenue Ahead, Computing Power Behind
Although Anthropic has surpassed OpenAI in revenue—according to Bloomberg, its annual revenue has exceeded $65 billion, surpassing OpenAI's revenue of over $40 billion as of July this year—there remains a significant gap in computing power scale between the two.
OpenAI indicated to investors in April this year that it plans to have 30 gigawatts of computing power by 2030. According to insiders, OpenAI projected last month that its computing power expenditure before 2030 would be approximately $750 billion, higher than the previous estimate of $665 billion.
OpenAI has signed numerous agreements with companies such as Oracle, Microsoft, and CoreWeave. OpenAI executives stated that these commitments have established a competitive advantage over rivals.
In contrast, the total computing power plans disclosed by Anthropic still seem far below OpenAI's target level.
-- Price
Key Variables Before IPO
Whether computing power can keep pace with growth directly relates to Anthropic's profit margins and customer retention.
Anthropic and its CFO Krishna Rao have prioritized computing power agreements over the past year. With the IPO prospectus expected to be made public in the coming weeks, investors are most eager to see the company's latest disclosures on computing capacity and expenditures.
However, these agreements are not set in stone. The construction of data centers faces multiple obstacles such as site selection, grid access, and government approvals, along with resistance from local governments and component shortages.
There is also uncertainty at the contract level. For example, in the agreement between SpaceX and Anthropic, either party can cancel the contract with 90 days' notice.
Original link
This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.
You may also like

Kraken Owner Invests in Altcoin, Considers Further Purchases. You Should Recognize Him

Nubank Suspends Crypto Withdrawals: What We Know So Far

Banking Outage in Japan: 128 Banks 'Frozen' at Once

Beyond APY: Saying Goodbye to Black Box Trust and Reassessing the Security Logic of ETH Staking

Democratic Party's AI Regulation Push Revealed: Plans for Intensive Legislation and Strict Scrutiny of Tech Giants

AI Begins to Shake Google's Search Advertising Business Behind $500 Billion Market Value Erosion

What is the real challenge for South Korean exchanges now that the regulatory framework is in place?

BlackRock Loses $1 Billion Through Bitcoin ETF

U.S. Stocks Plunge $2.8 Trillion Overnight! Top Silicon Valley VC Reveals: 90% of Open Source Model Spending Goes to China

Cassius Cuvee on Market Euphoria, High Leverage, and How to Stay in the Game

Why Did HYPE Soar and Then Fizzle Out After Wall Street Extended an Olive Branch?

What Did Crypto Investors Get for Their $189 Million Bet?

The Coming Crisis in AI Infrastructure: What About USD.AI?

JP Morgan Kinexys: How Should Banks Use Blockchain?

Current Chief Prosecutor: 'Public Disclosure of Actual Controllers Needed to Prevent Predatory Coin Issuance'

Midterm Elections Raise $3 Billion, Prediction Markets Emerge as Big Winners

Tether Earns $1.5 Billion in Q1: Where Does USDT's Revenue Come From?

Unpatched Bitcoin Lightning nodes face channel-draining bug

Claude's Evolution Speed Doubles, 80% of Increment Comes from AI

BCRA has already purchased over $16 billion this year, but reserves fell due to a payment to the IMF

Circle Reports US$ 226 Million Profit, But Shares Drop 9%: What Happened

Brazil's Economy Minister stated that the Selic rate should be reduced, but in a "sustainable" manner

A Counterintuitive Logic: The Anti-AI Movement Won't Eliminate Demand, But It Gives the Industry More Time to Absorb the Bubble

How to Determine Loss Attribution When DeFi Issues Arise?

Bank of England Raises Interest Rates: What Changes for Investors

Does Anyone Really Need Blockchain? Reflections Triggered by Kalshi

Post-Quantum Era Consensus Layer Divergence: Ethereum Maintains Decentralization, Monad Prioritizes Speed

Hunter Biden Discusses Meme Coin 'LAPTOP': Design for Transparency and Recommendations for the Democratic Party

Why Did Institutions Bet $40 Million on PhotonPay When Stablecoin Transfers Are Cheap Enough?









