- The Information reported on May 5, 2026 that Anthropic has agreed to pay Google $200 billion over five years for cloud servers and chip access.
- The deal extends an earlier announced Google-Anthropic partnership for cloud and TPU access; the $200B figure is the previously undisclosed contracted value.
- Combined cloud commitments from Anthropic and OpenAI now represent a roughly $2 trillion revenue backlog across Amazon, Google, Microsoft, and Oracle, per The Information.
- Server costs in 2026 are projected at $45 billion for OpenAI and $20 billion for Anthropic; the cumulative growth path implies continuing escalation through 2027-2030.
What Happened
Anthropic has agreed to pay Google $200 billion over the next five years for cloud servers and chip access, The Information reported on May 5, 2026. The Engadget summary frames the figure as the previously undisclosed contracted value of an Anthropic-Google partnership announced earlier this month. The deal sits alongside Anthropic’s separate multi-billion-dollar arrangement with Amazon Web Services, which similarly provides cloud and chip capacity.
Why It Matters
The $200 billion contract size is the largest publicly reported single AI infrastructure deal to date. Combined with Anthropic’s Amazon arrangement and OpenAI’s analogous cloud commitments, the AI infrastructure spending pattern has reached a scale that materially shapes the largest cloud providers’ financial trajectories. The Information’s reporting that combined Anthropic and OpenAI deals represent a $2 trillion revenue backlog across Amazon, Google, Microsoft, and Oracle reframes the AI boom: the underlying business is not primarily about consumer or enterprise AI revenue at the labs, but about flow of cloud-infrastructure capital from labs to hyperscalers.
Technical Details
The $200 billion figure covers a five-year period — implying roughly $40 billion per year average spend at Google alone. Earlier projections estimated Anthropic’s 2026 server costs at $20 billion total; the new disclosure suggests substantial year-over-year growth in the contracted commitment. For context, OpenAI’s 2026 server costs are projected at $45 billion. The Information’s $2 trillion revenue-backlog figure across the four hyperscalers (Amazon, Google, Microsoft, Oracle) represents commitments running over multiple years; year-by-year revenue recognition will be substantially smaller than the headline figure.
The deal extends the financing pattern that has driven 2026’s AI capex acceleration. Google, Amazon, Microsoft, and Meta now project a combined $725 billion in 2026 AI capital spending, per Financial Times reporting summarized earlier this month — up 77% from $410 billion in 2025. The revenue from Anthropic and OpenAI commitments backs that capex investment: hyperscalers are spending $725 billion partly because they have $2 trillion in committed multi-year AI customer revenue against which the spending will be amortized.
The “circular” nature of these deals has drawn scrutiny. Nvidia‘s investments in OpenAI, OpenAI’s $1 billion loan to Cerebras (secured by warrants for 33+ million Cerebras shares), and now Anthropic’s $200B Google commitment alongside Google’s earlier $3B+ investment in Anthropic create a pattern where the same capital appears to flow between AI infrastructure providers and AI labs in multiple directions. Engadget’s coverage notes these are “expensive circular deals” that “aren’t exactly a sustainable business practice.”
Who’s Affected
Google’s cloud and TPU business gains the largest single contracted AI revenue line item disclosed in 2026. Amazon, Microsoft, and Oracle remain Anthropic’s other major infrastructure providers; the Google deal positions them in competitive context. OpenAI’s analogous cloud commitments — primarily with Microsoft Azure but increasingly diversified — face a clearer benchmark for what frontier-lab cloud spending looks like. Nvidia, AMD, Broadcom, and other AI silicon providers benefit downstream from increased GPU and accelerator deployment to fulfill the cloud commitments. Investors and analysts watching the AI infrastructure trade gain a critical datapoint for the size of the underlying customer obligations supporting hyperscaler capex.
What’s Next
Watch for SEC disclosures from Google, Amazon, Microsoft, and Oracle that would formally book Anthropic-related revenue and contract values. Whether OpenAI’s analogous Microsoft commitment is similarly disclosed in upcoming earnings will be the natural comparison. The Cerebras IPO covered earlier this week — with banks fielding $10 billion in orders against a $3.5 billion offering — and Anthropic’s reported ongoing $50 billion funding round at $900 billion valuation will indicate how aggressively capital markets continue to underwrite the cloud-spending pattern. The structural risk: if AI revenue growth at the labs decelerates while contracted cloud commitments remain, the financial dynamic between labs and hyperscalers could become a stress point.
