Backblaze's $335M CoreWeave Pact Signals AI Storage Gold Rush for Startups

For startups building the AI infrastructure stack, the line between commodity and differentiated service is blurring fast. Backblaze, a cloud storage upstart turned public company, just locked in a $335 million, five-year deal to feed CoreWeave’s insatiable AI storage appetite. This isn’t just a contract win—it’s a signal that venture-scale returns can come from mastering unglamorous layers like HDD-based storage when paired with the right AI-native partners. With CoreWeave itself a darling of the AI cloud boom (last valued at $19B+), the partnership offers a blueprint for how nimble startups can carve defensible positions even as hyperscalers tighten their grip.

Backblaze, the publicly traded cloud storage provider (Nasdaq: BLZE), and CoreWeave, the AI-focused cloud platform (Nasdaq: CRWV), have jointly announced a substantial five-year, $335 million agreement that sees Backblaze delivering multi-exabyte HDD-based storage capacity to underpin portions of CoreWeave’s managed storage infrastructure. The deal, disclosed on June 23, 2026, represents a landmark commercial validation for Backblaze’s pivot toward serving AI-era workloads and highlights the critical role of cost-efficient, large-scale storage in the AI value chain.

“

The $335 million price tag—while spread over five years—provides a predictable revenue stream for Backblaze, which reported total revenue of $129.4 million in 2025.

The arrangement integrates Backblaze’s technology into CoreWeave’s AI Object Storage, specifically powering the HDD-backed tiers. CoreWeave’s customers—many running the full gamut of AI pipelines from training to inference to retrieval-augmented generation (RAG)—will gain immediate access to additional storage tiers without needing any code modifications. The press release emphasizes that CoreWeave’s patented LOTA distributed cache will work in tandem with Backblaze’s HDD storage to optimize data placement and preserve high-performance flash resources for the most demanding AI compute operations.

Contextually, this agreement lands at a moment when AI infrastructure spending is surging. IDC projects worldwide AI infrastructure revenue to exceed $100 billion by 2026, with storage representing a growing slice. Hyperscalers like AWS, Azure, and Google Cloud dominate, but a second tier of specialized providers—CoreWeave for GPU-accelerated compute and Backblaze for cost-optimized storage—are forging alliances that let them compete on flexibility and economics. CoreWeave, which positioned itself as "The Essential Cloud for AI," has rapidly scaled to a reported valuation exceeding $19 billion in private markets prior to going public, and now needs vast storage substrates to match its GPU clusters. Backblaze, historically a consumer/reseller backup provider, has steadily rebuilt its platform to capture data-intensive enterprise workloads, claiming over 100,000 customers worldwide.

The $335 million price tag—while spread over five years—provides a predictable revenue stream for Backblaze, which reported total revenue of $129.4 million in 2025. Even assuming straight-line recognition, this deal alone could add roughly $67 million per year, representing a 50%+ boost to top-line growth potential. However, the exact revenue recognition will depend on usage milestones and storage consumption, not just a flat fee; the "multi-exabyte" descriptor suggests tiered consumption. For CoreWeave, offloading bulk HDD infrastructure to a specialist allows capital to be concentrated on GPU and networking assets, potentially improving its unit economics.

Gleb Budman, Backblaze’s co-founder and CEO, framed storage as “the foundation every AI workflow is built on,” underscoring how idle compute is wasted without the right data pipelines. Nick Hoover, CoreWeave’s VP, acknowledged Backblaze’s reputation for making complex HDD storage reliable and consumable at scale. These quotes—while promotional—hint at a shared vision: specialized infrastructure vendors bolting together to offer a full-stack AI cloud alternative.