CoreWeave and Solidigm Co-Engineer Liquid-Cooled 122TB SSDs for AI Clusters

CoreWeave has entered into a multi-year strategic agreement with Solidigm to secure priority access to ultra-high-capacity solid-state drives (SSDs). The partnership aims to eliminate storage bottlenecks that frequently hinder the performance of large-scale AI compute clusters.

Under the terms of the deal, the two companies are co-engineering specialized drives optimized for AI workloads. A central feature of this collaboration is the implementation of native direct-to-chip liquid cooling for 122TB SSDs. This cooling technology is essential to manage the extreme heat density generated within dense GPU clusters. CoreWeave is specifically utilizing quad-level cell (QLC) SSDs, which provide the high density and lower cost per terabyte required for the read-heavy pipelines used in AI training.

The Infrastructure Race

CoreWeave, an Nvidia-backed AI cloud provider, is scaling its infrastructure at a rapid pace to meet the surging demand for GPU services. To support this growth, CoreWeave is aggressively expanding its physical footprint. This includes a definitive agreement to acquire Core Scientific for $9 billion in an all-stock deal, which is expected to close in the fourth quarter of 2025.

Removing the Storage Bottleneck

While GPUs provide the raw compute power for artificial intelligence, the ability to feed data to those processors at scale remains a critical physical constraint. Storage is frequently a "quiet bottleneck" in AI architecture; if data cannot be retrieved fast enough, expensive GPUs sit idle. By securing a priority supply chain and developing liquid-cooled, ultra-high-capacity drives, CoreWeave is removing a primary barrier to its operational scaling. Jacob Yundt, a CoreWeave executive, described the partnership as "incredible" for improving data center effectiveness.

Path to Public Markets

This infrastructure push comes as CoreWeave prepares for a significant financial transition. The company is targeting a valuation of over $35 billion in an upcoming U.S. initial public offering (IPO). Strengthening its hardware stack and securing long-term component availability positions the company as a leading AI-native cloud provider ahead of its market debut.

Investors and industry analysts will now be watching to see how the deployment of these 122TB drives impacts the efficiency of CoreWeave's GPU clusters and whether the company can maintain its aggressive expansion timeline as it moves toward its IPO.