Anthropic wants control. The AI company founded by former OpenAI executives has struck preliminary deals for more than a dozen U.S. data centers. Those sites could deliver over one gigawatt of power-hungry capacity. Yet the price tag runs into tens of billions. So the startup turned to an old partner for help.
Google stepped in. Alphabet’s cloud unit agreed to backstop lease payments at multiple facilities. That financial guarantee let Anthropic secure what amounts to a $35 billion loan package, according to a Bloomberg report from June 9, 2026. The arrangement reveals how even well-funded AI labs lean on big tech balance sheets to move fast.
Google’s Guarantees Unlock Massive Compute
Details paint a picture of tight integration. Google doesn’t just provide guarantees. It co-designs some server chips destined for those centers. The setup builds on earlier pacts. In April 2026 Anthropic announced an expanded partnership with Google and Broadcom. The deal secured multiple gigawatts of next-generation tensor processing units expected online from 2027. “This groundbreaking partnership with Google and Broadcom is a continuation of our disciplined approach to scaling infrastructure,” said Krishna Rao, Anthropic’s chief financial officer, in the company’s official announcement on April 6, 2026.
But the latest moves go further. Bloomberg noted Google now backstops leases at five data centers. Those commitments helped Anthropic obtain the $35 billion in financing. The numbers add up quickly. Power, cooling, land. All of it costs more as models grow larger and inference demands explode. And the race doesn’t wait.
Earlier this month Anthropic disclosed a confidential filing for a U.S. initial public offering. The timing feels deliberate. Massive infrastructure bets signal confidence to public investors. They also highlight dependence. Even with a soaring valuation, pure-play AI firms struggle to shoulder every dollar alone.
Consider the funding backdrop. In late May 2026 Anthropic closed a $65 billion Series H round. Altimeter Capital, Dragoneer, Greenoaks and Sequoia Capital led the investment. Post-money valuation hit $965 billion. That figure topped OpenAI and turned heads across Silicon Valley. “This funding will help us serve the historic demand we are experiencing, stay at the research frontier, and bring Claude to more of the places where work happens,” Rao said in the company’s May 28, 2026 announcement.
Google’s role predates this frenzy. The search giant invested early in Anthropic. It poured in billions more this spring. In April Alphabet committed up to $40 billion. Ten billion arrived immediately at a roughly $350 billion valuation. Another $30 billion hinges on performance targets. The CNBC story published April 24, 2026 framed the move as Google spreading its AI bets while locking in a key partner.
Those dollars buy more than equity. They buy priority access to compute. Anthropic committed to spend $200 billion on Google Cloud services and chips over five years, Reuters reported in May citing The Information. The figure dwarfs previous deals. It also cements Google’s position as a primary supplier even as Anthropic trains on multiple clouds.
AWS remains the main training partner. Microsoft Azure hosts Claude too. Yet the Google relationship runs deeper on custom silicon. TPUs power much of Anthropic’s recent growth. Last October the companies expanded TPU usage to as many as one million units. That pact alone carried a price tag in the tens of billions and more than a gigawatt of capacity. Google Cloud CEO Thomas Kurian praised the choice at the time. “Anthropic’s choice to significantly expand its usage of TPUs reflects the strong price-performance and efficiency its teams have seen with TPUs for several years,” he said in the Anthropic blog post from October 23, 2025.
The current data center push takes that cooperation to another level. Anthropic now operates its own facilities rather than renting raw cloud instances. Control brings advantages. Custom configurations. Faster iteration. Lower long-term costs if utilization stays high. But it demands capital. Huge capital. Google’s guarantees lower the risk for lenders. They also tie the two companies closer together at a moment when both chase leadership in frontier AI.
Competition sets the pace. OpenAI pursues its own Stargate supercomputer project with SoftBank and others. Meta builds massive GPU clusters in-house. Every player hunts for power. Data center construction now faces delays from grid constraints, permitting and component shortages. Securing more than a gigawatt in one stroke counts as rare progress.
Yet questions linger. How much of the $35 billion facility financing flows back to Google through chip and cloud purchases? The economics blur. Google supplies the TPUs that populate those centers. Broadcom designs aspects of the custom silicon. The circle feels complete. Anthropic gains scale. Google gains a high-volume, long-term customer committed to its hardware roadmap.
Revenue tells part of the story. Anthropic’s run rate recently topped $30 billion. That marks a sharp jump from roughly $9 billion at the end of 2025. Demand for Claude models across enterprise, coding and consumer use cases keeps climbing. But training larger models still burns cash faster than revenue arrives. Hence the infrastructure scramble.
Executives at Anthropic speak of staying at the research frontier. That requires ever-larger clusters. Current deals with Google, Amazon and even SpaceX for Colossus GPUs reflect a multi-cloud strategy. No single provider can meet all needs. Dependence on any one risks outages or pricing pressure. Spreading bets reduces that exposure. Still, Google’s financial backstops set it apart.
The Yahoo Finance article that broke aspects of the latest data center story on June 14, 2026 highlighted more than a dozen preliminary agreements. It noted Google’s potential guarantees on lease obligations. The piece also referenced the IPO filing and valuation surge. Those elements match reporting from Bloomberg and Reuters in the days that followed.
Broader market reaction stayed muted. Alphabet shares trade on expectations of cloud growth and AI monetization. A massive committed customer helps. Yet investors watch margins. Guaranteeing leases carries contingent liability even for a company with Google’s cash reserves. So far the bet looks strategic. It secures a leading AI lab as a cloud and silicon partner while hedging against its own Gemini models falling behind.
Anthropic, meanwhile, prepares for public scrutiny. An IPO would expose its burn rate, compute costs and reliance on big tech. The $965 billion valuation leaves little room for disappointment. Meeting performance milestones for the extra $30 billion from Google becomes critical. So does translating hype into sustained revenue.
Power remains the ultimate constraint. One gigawatt today. Multiple gigawatts tomorrow. Nuclear restarts, renewable projects and grid upgrades all take years. Companies that lock in sites and financing now gain years of advantage. Google’s willingness to stand behind Anthropic’s leases signals confidence in that timeline.
The partnership carries risks too. Regulators eye concentration in AI infrastructure. Antitrust concerns swirl around big tech investments in startups. Google and Anthropic already face questions about their closeness. Expanded financial ties could draw more attention. Yet both sides appear undeterred.
So the deals keep coming. More capacity. More guarantees. More chips. Anthropic chases scale to train the next generation of models. Google supplies the foundation while reaping returns on its early stake and ongoing cloud business. The arrangement isn’t charity. It’s calculated interdependence.
And the bill keeps growing. Thirty-five billion here. Two hundred billion committed over five years there. These figures once seemed absurd. Now they represent table stakes for those who intend to lead. The companies that secure compute today will likely set the agenda tomorrow. Google just made sure it holds a prominent seat at Anthropic’s table.
WebProNews is an iEntry Publication