Economy

Anthropic reportedly in preliminary talks to lease computing power from Meta

Illustration for: Anthropic reportedly in preliminary talks to lease computing power from Meta
AI-generated illustration. Visual interpretation does not represent real individuals or scenes.

Jul 17, 2026

THE BARE STORY

Artificial intelligence firm Anthropic is in very preliminary discussions to lease computing power from Meta, according to reports and a person familiar with the matter. The potential agreement is reportedly valued at approximately $10 billion, with one report specifying the deal could span over two years. Following reports of the potential agreement on Friday, shares of Meta rose off their daily lows.

The discussions come as Anthropic seeks to secure access to Nvidia-made artificial intelligence chips amid ongoing supply constraints. To manage these constraints, Anthropic currently places usage limits on its most advanced models, such as Fable. The firm, led by co-founder and chief executive officer Dario Amodei, recently established a similar agreement with Elon Musk’s SpaceX to utilize computing capacity at its Colossus 1 data center.

For Meta, the potential deal aligns with its goal to expand into the cloud computing business. Meta CEO Mark Zuckerberg stated in May that the company was considering selling computing power to external customers as a new revenue stream to monetize its AI investments. Zuckerberg also noted last October that companies regularly offer to purchase compute power from Meta at a premium. To support its technology infrastructure, Meta's capital expenditures could reach up to $145 billion in 2026.

In a related development, Dave Brown, a longtime executive at Amazon Web Services (AWS), is leaving AWS after 19 years and is reportedly in talks to join Meta. Brown previously led AWS's compute and machine learning services division, where he managed custom Trainium AI chips and key AI infrastructure. Meta declined to comment on the report of the potential leasing deal.

Same Facts. Different Perspectives.

Two AI models. Two viewpoints. One factual foundation.

Right Perspective

• Monetizing the Capital Engine Systemic stability and sustained technological progress rely on corporate efficiency and the successful monetization of heavy capital investments. Meta’s potential $10 billion leasing agreement validates its massive infrastructure spending, showing that building high-capacity data centers yields direct commercial value. By converting excess compute capacity into a premium revenue stream, as envisioned by CEO Mark Zuckerberg, the market rewards massive capital risk and creates a self-sustaining cycle of technological reinvestment.

• Optimizing Scarce Market Resources Market efficiency is achieved when high-demand assets flow to where they can be most productively utilized through private contracting. Amid acute shortages of Nvidia-made AI chips, private B2B leasing agreements allow companies with surplus capacity, like Meta or SpaceX, to support specialized developers like Anthropic without government intervention. This flexible, market-driven allocation of compute power ensures that valuable hardware does not sit idle and that the most advanced models continue to be trained.

• Securing Infrastructure Scale Dominance Maintaining global technological leadership requires immense scale, massive capital expenditure, and the acquisition of world-class operational talent. Meta’s projected capital expenditures of up to $145 billion by 2026, paired with the potential recruitment of AWS's infrastructure veteran Dave Brown, represents the necessary concentration of talent and capital required to build next-generation hardware pipelines. Only organizations with this level of scale can absorb the immense financial risks required to pioneer and secure the foundational infrastructure of the future global economy.

How it may affect me

As a U.S. reader:

• In the short term, you may see a reduction in current usage limits on Anthropic's most advanced artificial intelligence models, such as Fable, if the company successfully secures more computing power.

• Over the longer term, you may experience a more competitive and open AI market with diverse supply chains if smaller developers can successfully bypass traditional cloud monopolies to access hardware.

• You may face a future digital landscape where AI development and raw computing power are concentrated within a few massive tech conglomerates, potentially reducing public leverage and regulatory oversight.

Read the story at

Note: All TheBareNews content is AI-generated. For additional context, reporting, and updates, you are invited to explore the news outlets linked above.