Anthropic’s $35 billion Lambda Deal: How Nvidia is expanding its reach across AI

According to the WSJ, Nvidia itself holds the lease for the data centre where Anthropic’s computing will take place, meaning the facility is not directly controlled by either Lambda or Anthropic. 

The deal shows that Nvidia’s role in the AI race now goes well beyond selling chips. The company is also getting involved in financing, data-centre capacity and partnerships that help AI firms build the computing infrastructure they need. 

The development comes as Anthropic prepares for a possible stock market listing as early as October. 

Nvidia sits at the centre of the deal 

The data centre involved in the Lambda deal is being built by Hut 8, a bitcoin mining company that has expanded into data centres. The facility is located in Nueces County, Texas.

A few weeks before Anthropic and Lambda finalised their $35 billion agreement, Nvidia reached its own deal with Hut 8 to secure the data-centre capacity.

Lambda then signed its agreement with Anthropic and will provide the computing capacity to the AI company.

In simple terms, three companies are involved between Anthropic and the computers it needs. Nvidia controls the lease, Hut 8 is building and operating the physical data centre, while Lambda is the company selling the computing capacity to Anthropic. 

Lambda will use Nvidia chips at the site. It bought those chips from Nvidia, which is also an investor in Lambda. This puts Nvidia on several sides of the same deal. It is the chip supplier, the company holding the lease for the facility and an investor in the cloud provider. 

It is not yet clear how much Lambda will pay Nvidia to use the space. It is also unclear whether Nvidia will receive any share of the money generated by the Texas facility. 

Hut 8 said in July that an unnamed “high-investment-grade company” had signed 15-year leases covering the entire planned 700-megawatt Texas campus. The company did not name the tenant at the time. 

Hut 8 said the facility, which is expected to cost about $20 billion, would be built to support Nvidia chips. 

Nvidia also announced in July a programme aimed at helping smaller cloud companies such as Lambda. Under the programme, Nvidia would provide credit support to these so-called “neocloud” providers in return for a share of their cloud revenue. 

However, Nvidia has reportedly paused parts of the programme. It is also not clear whether Lambda’s deal is covered by the arrangement. 

Why this matters for Anthropic and Nvidia 

There is another side to the AI spending boom: many of these companies are financially tied to one another through investments, loans, chip purchases and computing contracts. Nvidia, Anthropic and their cloud partners have hundreds of such overlapping deals. In some cases, money from one deal helps fund another, meaning the headline value of these commitments does not always represent entirely separate spending.

The Lambda deal shows how Nvidia is becoming involved in almost every part of the AI infrastructure race.

Instead of simply selling chips to AI companies, Nvidia is helping companies such as Anthropic get access to the computing power they need. At the same time, it is helping smaller cloud providers such as Lambda find major customers. These companies often do not have enough money or resources to secure large data centres on their own. 

The broader Hut 8–Anthropic relationship also involves Google

Another Hut 8 facility being built nearby is designed to house Google’s competing tensor processing units, or TPUs, for Anthropic. Google has provided financial guarantees to help Hut 8 raise debt for that project. 

As a result, Hut 8 is involved in building infrastructure that will support Anthropic using both Nvidia chips and Google’s competing TPU technology. 

Anthropic is spending billions on computing 

The Lambda deal is just one part of a much bigger spending spree by Anthropic. The company is racing to add computing capacity as demand for its Claude AI models continues to rise. It is also trying to keep up with rival OpenAI, which is spending heavily on AI infrastructure.

Earlier this month, Anthropic signed another major deal worth $45 billion over six years with Nscale, another Nvidia-backed neocloud provider. 

Under that agreement, Anthropic will rent about 460 megawatts of computing capacity at a data centre in West Virginia. The facility will use Nvidia’s next-generation Vera Rubin chips from late 2027. 

The project, called the Monarch Compute Campus, will produce its own electricity using gas turbines instead of relying on the regional power grid. 

Anthropic’s contract is also important for Nscale because it can use the deal as collateral to raise the money needed to build the facility. In other words, Anthropic’s promise to pay helps Nscale secure financing for a data centre that does not yet exist. 

The Lambda and Nscale deals alone are worth about $80 billion. They come on top of several other major computing commitments Anthropic has made this year.

The company has a $50 billion computing agreement with Fluidstack and a $10 billion deal with Norway-based cloud startup Volta for data-centre capacity. In July, Anthropic signed a major deal with AMD under which it will deploy up to 2 gigawatts of AMD’s Instinct MI450-series chips (starting in the first half of 2027), representing tens of billions of dollars in hardware. As part of the same agreement, AMD committed to invest up to $5 billion in Anthropic, tied to deployment milestones.

In May, it reached an agreement with SpaceX that provides about $1.25 billion worth of computing capacity every month from two SpaceX data centres.

Anthropic also expanded its partnership with Amazon in April, adding multi-gigawatt computing capacity (including Trainium chips). Amazon’s total committed capital in Anthropic stands at roughly $33 billion (including milestone-tied amounts), and the fair value of its stake was marked at around $190 billion as of mid-2026. In the same period, Anthropic expanded its partnership with Google.

Microsoft and Nvidia have also agreed to invest up to $15 billion in Anthropic. In return, Anthropic has committed to spend $30 billion on Microsoft’s Azure cloud.

Taken together, Anthropic’s disclosed computing commitments now exceed $150 billion. That is an enormous amount for a company that, unlike Microsoft, Google and Amazon, does not own the data centres, chips or power plants it relies on.

Anthropic has previously acknowledged that the rapid increase in Claude usage put pressure on its systems earlier this year. The company faced reliability problems when demand jumped.

It has argued that the huge infrastructure expansion is needed to handle growing demand and compete with other major AI companies.

The big spending comes before a possible IPO 

Anthropic’s aggressive infrastructure spending comes as the company prepares for a possible stock market debut. The company confidentially filed IPO paperwork with the US Securities and Exchange Commission on June 1, 2026. It is reportedly targeting a Nasdaq listing as early as October. 

Goldman Sachs, JPMorgan and Morgan Stanley are expected to lead the offering. Anthropic’s valuation has also jumped sharply.

Its Series H funding round in May valued the company at about $965 billion. That was up from $380 billion just three months earlier, in February.

Some reports have suggested that Anthropic could seek a public-market valuation of as much as $2 trillion. The company’s annualised revenue run rate reportedly reached about $65 billion by the end of July, compared with around $47 billion in the spring. 

That rapid growth is a key part of Anthropic’s pitch to potential investors.

The company needs investors to believe that Claude’s revenue will continue growing fast enough to support the tens of billions of dollars in computing commitments it has made, many of which stretch years into the future.