Anthropic inks a $35B cloud deal with Nvidia-backed Lambda
Two of the frontier's biggest spenders put fresh numbers on the board on the same afternoon. Anthropic is locking in tens of billions of dollars of Nvidia capacity through a smaller cloud partner it has never bet this big on before, and OpenAI's ad business — barely six months old — passed a billion-dollar run rate and is now in 40-plus countries.
Anthropic has signed a $35 billion cloud deal with Lambda, the Nvidia-backed cloud provider, the Wall Street Journal reported Monday. A person familiar with the matter told Reuters the agreement will bring Nvidia's chips online to handle demand for Claude, the company's flagship model. The deal is among the largest single cloud commitments any AI lab has signed and confirms Lambda, once a niche GPU-rental outfit, as a serious tier-two player in the foundation-model supply chain.
The shape of the agreement matters more than the headline number. Anthropic is paying for capacity it does not yet run; in exchange, it gets priority access to the Nvidia hardware that the rest of the industry is also scrambling for. Lambda, meanwhile, takes on the build-out risk that would normally sit with a hyperscaler. It is a clear signal that the big three — AWS, Microsoft, and Google — are no longer the only path to frontier-scale compute. We wrote last month about Anthropic's separate $45B commitment to Nscale — Anthropic commits $45B to rent 460MW of Nscale AI compute — and the new Lambda deal points the same way: Anthropic is hedging its infrastructure across multiple smaller providers, not concentrating on a single anchor tenant.
The catch is concentration on the chip side. Nvidia sits on both ends of every one of these deals — supplying Lambda, taking a stake, and now helping finance the underlying silicon itself. As labs diversify their cloud partners, they are also deepening the dependency on the one company that controls the silicon.
OpenAI's ChatGPT Ads business has reached $1 billion in annualized revenue run rate less than 200 days after launch, the company said in a blog post Monday, and is now live in more than 40 countries. Self-service access through Ads Manager is rolling out across India, Europe, the Middle East, and North Africa, OpenAI said, and small and medium-sized businesses now represent a material share of the business — a milestone for a sales channel that started as a managed-service pilot.
The headline number is striking, but the structure behind it matters more. OpenAI added 50 technology and measurement partners to the platform, and the majority of campaigns now run on CPC or outcome-optimized bidding rather than flat-rate placements. One e-commerce advertiser reported a 3x return on ad spend across campaigns in 28 days; a technology partner said more than 80% of ad-driven ChatGPT traffic came from new customers. OpenAI reiterated that ads are clearly labeled, do not influence model output, and that advertisers cannot see private conversations.
The 200-day ramp lands the ads business at roughly a third of what OpenAI said the ad line is targeting by year-end — $2.5 billion in ad revenue, on top of a broader annualized revenue figure the company says will exceed $40 billion. If that trajectory holds, ChatGPT Ads becomes one of the fastest ad businesses to scale past the billion-dollar mark in the platform era, and the clearest evidence yet that advertising, not subscriptions, is the real growth frontier in consumer AI.
What to watch: whether the $2.5 billion year-end target holds, and whether Anthropic follows with a similar ad product of its own as Claude's consumer usage continues to climb.
Is Nvidia the de facto central bank of the AI build-out, or just the supplier everyone keeps choosing? Tell us in the comments.
