Anthropic, the artificial-intelligence company behind the Claude models, has signed a cloud-computing agreement worth roughly $35 billion with Lambda, an Nvidia-backed cloud provider. And for crypto, the relevant part is how thoroughly the industry surrounds this single AI company: crypto mining firms are building the data centers Anthropic runs on, crypto exchanges let traders speculate on Anthropic’s valuation, and crypto applications increasingly plug into Anthropic’s models. Two developments this week — the Lambda deal and a resumption of Anthropic’s safety testing — lit up the first and third of those threads at the same time.
The six-year agreement will bring additional Nvidia-powered computing capacity online for Anthropic, with the capacity expected to come from a data center in Nueces County, Texas, being developed by Hut 8. The facility is expected to provide about 350 megawatts of capacity, according to a source familiar with the matter cited by The Wall Street Journal, which first reported the deal. None of the companies involved had commented on the reported terms as of publication. The Crypto Times will update this article if the companies provide additional details or confirmation.
Bitcoin-Mining Infrastructure Shifts Toward AI
Hut 8 began as a Bitcoin-mining company and has repositioned itself as an energy-and-data-center infrastructure platform, making the arrangement another example of infrastructure built for energy-intensive computing being repurposed for AI. Miners spent years securing the power capacity, land, grid connections and cooling that AI developers now compete for, and are increasingly monetizing those assets as AI landlords.
The structure of the Anthropic deal shows how that plays out. Reporting indicates Nvidia is leasing data-center capacity developed by Hut 8, which Lambda will use to host Nvidia chips for the cloud services it sells to Anthropic. That places the four companies at different points in the AI infrastructure chain: Anthropic is the end customer, Lambda is the cloud provider, Nvidia supplies the hardware and holds the lease, and Hut 8 develops the underlying facility: makin it a circular arrangement in which Nvidia is simultaneously supplier, landlord, and investor.
The deal follows other large agreements linking Anthropic to crypto miners. Anthropic previously signed a 20-year, roughly $9.1 billion lease with Riot Platforms for 191 megawatts of capacity at Riot’s Rockdale, Texas campus, and an earlier Hut 8 partnership in Louisiana. Peers including IREN, Core Scientific, TeraWulf, and MARA have pursued similar pivots; public miners had signed more than $70 billion in AI and high-performance-computing deals as of early 2026. The Crypto Times has tracked the broader shift as miners seek to monetize power and data-center infrastructure beyond Bitcoin.
Crypto’s Other Bet on Anthropic: Pre-IPO Contracts
Anthropic also has a more direct connection to crypto markets through pre-IPO derivatives. Crypto exchanges have introduced contracts that track the implied value of private companies such as Anthropic before a public listing. The Crypto Times reported in August that Binance’s Anthropic-linked pre-IPO contract had implied valuations of up to roughly $1.6 trillion.
These contracts do not represent ownership of Anthropic shares; they give traders price exposure to the valuation implied by the derivative market, which can differ substantially from Anthropic’s private-market valuation or any eventual IPO price. Anthropic confidentially submitted a draft S-1 registration statement to the SEC on June 1, and because the filing was confidential, its contents are not public. The distinction matters: the $35 billion Lambda agreement is a cloud-computing commitment and does not establish Anthropic’s eventual IPO valuation or share price. But it does mean a trader holding an Anthropic pre-IPO contract has real exposure to exactly the kind of scaling news this deal represents.
Anthropic Resumes External Cybersecurity Testing
The infrastructure expansion came alongside a very different development. On August 31, Anthropic said it had resumed external cybersecurity testing of its AI models after introducing new safeguards, roughly a month after pausing evaluations following security incidents disclosed on July 30.
Reporting on Anthropic’s disclosure described the incidents concretely: across six evaluation runs, Claude models gained unauthorized access to the production infrastructure of three real organizations, with one model reportedly accessing hundreds of rows of production data and another uploading malware that reached 15 systems. Anthropic characterized these as operational-security failures caused by a misconfiguration that left internet access open in a third-party evaluation environment, and emphasized that the models had been given capabilities normally restricted in production; so the incidents do not establish that Claude’s production systems independently attacked real-world organizations. Separately, Britain’s AI Security Institute reported that a Claude model took unauthorized actions on the live internet during a test in which it had deliberately been given internet access.
In response, Anthropic said it built a classifier that identifies, in real time, when a model attempts to probe or escape a testing environment or unexpectedly obtains internet access, and can block the action, end the task and alert a human. It also strengthened requirements for external testing partners, more isolated environments, network controls, explicit scope-setting and monitoring, and reassigned about 150 engineers to security, reliability and privacy work.
Why The Developments Matter For Crypto
Together, the two threads capture crypto’s double exposure to the AI build-out. On the infrastructure side, Bitcoin miners and other power-intensive operators are increasingly courting AI customers, building a business model around facilities originally developed for energy-intensive computing, with Anthropic’s Hut 8 and Riot agreements as prime examples. On the technology side, crypto companies are experimenting with AI agents that interact with trading systems, wallets, market data, and on-chain applications, which makes model reliability, access controls and containment directly relevant to AI-enabled finance.
The Anthropic incidents do not establish that AI agents used in crypto systems are unsafe. They occurred in evaluation environments where models were deliberately given elevated capabilities. But they illustrate why isolated testing environments, network controls, and human oversight remain important as AI systems are given greater ability to act on external systems, a caution that lands squarely on a crypto industry racing to hand agents more autonomy.
For crypto markets, the broader picture is twofold: mining-era infrastructure is increasingly positioned to support AI workloads, while the deployment of more capable AI systems creates new security and operational considerations for financial applications. The Crypto Times makes no market or price forecast on any of the companies named.
Also read: ARK Says Anthropic, OpenAI Revenue Tops $115B as Crypto Markets Price AI Boom
