The chipmaker will supply high-bandwidth memory, DRAM, and SSDs for Claude’s data centres, run Anthropic’s models internally, and take a stake in its latest funding round.
Micron Technology and Anthropic have signed a strategic agreement that ties the memory maker to the AI company across three fronts at once.
A multi-year supply deal for the hardware that runs Anthropic’s models, a joint effort to design memory and storage architecture around AI workloads, and a strategic investment by Micron in Anthropic’s latest funding round.
The companies announced the agreement in a joint statement, framing it as a way to scale what they called next-generation AI infrastructure.
Under the supply portion, Micron said it will provide high-bandwidth memory (HBM), DRAM, and solid-state drives (SSDs) for Anthropic’s data-centre infrastructure.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!
Those three product lines map neatly onto the bottlenecks that large AI systems run into: HBM feeds data to accelerators fast enough to keep them busy, DRAM holds the working set, and SSDs store the rest.
For a company training and serving models at the scale Anthropic now operates, securing that supply is less a procurement detail than a strategic dependency.
The two firms also said they will jointly analyse how memory and storage subsystems perform across different workloads, with the stated aim of improving performance, energy efficiency, and what they described as “token economics” in Anthropic’s infrastructure.
The phrase is doing some work there. Every token a model generates carries a cost in power and silicon, and shaving that cost is one of the few levers an AI company has that does not involve simply buying more chips.
The arrangement runs in both directions. Micron said it has deployed Anthropic’s Claude models internally to accelerate coding and to enable agentic use cases across its engineering, manufacturing, and enterprise functions.
In other words, the supplier is also a customer, using the models it helps power to run its own operations. It is a tidy demonstration of the loop the AI industry keeps reaching for, in which the technology is sold partly on the strength of the seller using it themselves.
Financial terms were not disclosed. Micron did not put a figure on the supply agreement, and it did not say how much it is investing in Anthropic’s Series H round.
Both companies left those numbers out of the announcement, so any attempt to size the deal, in dollars committed or capacity reserved, would be guesswork.
What is on the record is the scale of the company Micron is buying into. Anthropic said on 1 June 2026 that it had confidentially filed for a US IPO, having raised $65bn in its Series H at a valuation of $965bn.
That round, which earlier in its life had been reported as a $30bn raise at a $900bn valuation, has drawn in a roster of strategic backers, with Micron now among them.
The pattern is by now familiar. Chip and cloud giants have spent the past year writing cheques to the AI labs that buy their hardware, blurring the line between supplier and shareholder.
Google has committed to invest up to $40bn in Anthropic, and NVIDIA has built a sprawling book of AI equity bets across the companies that consume its accelerators.
Micron’s move places it in the same camp, hedging a supply relationship with an ownership stake.
Neither company set out a timeline for when the joint architecture work will yield results, nor did they say how much HBM or storage capacity the supply deal covers.
Anthropic’s confidential filing means its detailed financials remain out of public view for now. The next concrete read on the partnership is likely to come when Micron reports earnings, where the company will have to account for both the investment and any revenue the supply deal begins to generate.
Get the TNW newsletter
Get the most important tech news in your inbox each week.