
AMD will invest up to $50 billion in Anthropic and simultaneously reach a cooperation agreement for computing power supply
AMD announced a strategic partnership with Anthropic, with a maximum investment of $50 billion in the future. Anthropic will deploy 2 gigawatts of AMD Instinct MI450 GPU computing power, with the first gigawatt expected to be completed in the first half of next year. This move aims to expand computing power reserves to meet the surge in demand for the Claude model, as Anthropic has previously signed computing power agreements with SpaceX and others

On Wednesday, AMD announced a strategic partnership with Anthropic and stated that it would invest up to $50 billion in the artificial intelligence company in the future.
According to the agreement, Anthropic will adopt AMD Helios rack-level solutions, deploying a total of 2 gigawatts of computing power using AMD Instinct MI450 series GPUs. The announcement indicated that the first batch of 1 gigawatt of computing power (gigawatts have become a common metric for measuring the scale of AI data centers) will be deployed in the first half of next year.
AMD CEO Lisa Su stated in a statement, "We are very pleased to deepen our collaboration with Anthropic to achieve gigawatt-scale deployment of the AMD Helios computing platform."
AMD is one of the major beneficiaries of the AI industry boom, producing GPUs that can be used for model training and handling large computing loads, directly competing with market leader NVIDIA. AMD has already established partnerships with several leading companies in the industry, including Anthropic's competitor OpenAI.
Previously, AMD reached an agreement with OpenAI to issue up to 160 million shares of AMD common stock warrants to the latter, accounting for about 10% of the company's total equity; the unlocking of the equity is tied to performance metrics related to computing power deployment and AMD's stock price performance.
The signing of the agreement between AMD and Anthropic is the latest development in Anthropic's intensive establishment of computing infrastructure partnerships this year, aimed at continuously expanding its computing power reserves. In April, Anthropic stated that the market demand for the Claude series models and related products surged, putting pressure on its computing infrastructure, affecting the stability and performance of user-facing services during peak periods.
In May, Anthropic signed a contract with SpaceX, owned by Elon Musk, to rent all the computing power of its Colossus No. 1 data center located in Memphis, Tennessee. According to SpaceX's prospectus, under the agreement, Anthropic pays SpaceX $12.5 billion per month, with the contract lasting until May 2029.
In April of this year, Anthropic signed a multi-billion dollar computing power agreement with Amazon, while also finalizing multi-gigawatt scale computing power procurement contracts with Google and Broadcom. Media reports previously indicated that Anthropic is also in preliminary discussions with Meta to lease the latter's computing resources.
Anthropic's business scale has rapidly grown this year, with its main AI code assistant, Claude Code, contributing significant incremental revenue. The company disclosed in May that its annualized revenue exceeded $47 billion, while its total revenue for the entire previous year was about $10 billion In the same month, Anthropic completed a new round of financing, with a post-investment valuation reaching $96.5 billion. The company is preparing for a large-scale IPO, which is expected to be listed as early as this year. Anthropic submitted a confidential filing to the U.S. Securities and Exchange Commission in June, and has not disclosed more details about the offering
