CoreWeave has locked in a multi-year supply agreement with Solidigm for solid-state drives, a deal designed to support the company's rapid expansion of AI cloud infrastructure. The partnership underscores how specialized storage hardware is becoming a critical piece in the race to build and scale systems that can handle massive AI workloads.
Why storage matters for AI
Training and running large AI models demands more than just powerful GPUs. The data has to move fast — from storage to compute and back again. Slow or insufficient storage can create bottlenecks that leave expensive processors idle. Solidigm's SSDs, built on technology originally developed by Intel's NAND business, are optimized for the high-throughput, low-latency access patterns that AI training jobs require. The deal gives CoreWeave a guaranteed supply of these drives over several years, helping it plan capacity without worrying about component shortages.
What the deal means for CoreWeave
CoreWeave started as a cryptocurrency mining operation before pivoting hard into AI cloud services. The company now runs one of the larger fleets of Nvidia GPUs available for rent, and it has been signing big contracts — including a reported multi-billion-dollar deal with Microsoft. But hardware supply chains have been a persistent headache for every cloud provider chasing AI demand. By securing a long-term SSD commitment, CoreWeave removes one more variable from its build-out plans. The company can now order servers knowing the storage will be there when they're assembled.
Solidigm's place in the AI supply chain
Solidigm, a subsidiary of SK hynix, has been pushing its SSDs as purpose-built for AI data pipelines. The company's QLC-based drives, for example, offer high density at lower cost, which matters when you're filling racks with petabytes of storage. The CoreWeave deal is a signal that even relatively young cloud providers are thinking about storage architecture as a strategic differentiator, not just a commodity buy. It also shows that Solidigm is winning customers beyond the traditional enterprise data center.
The financial terms of the agreement were not disclosed. Neither company said when deliveries would begin or how much capacity the deal covers. But the multi-year commitment suggests both sides expect AI cloud demand to keep climbing — and that storage will stay a key piece of the puzzle.
