Backblaze: The Backup Drag Is Fading

Backblaze said B2 Cloud Storage is now 62% of revenue and grew 34% YoY in Q2’26, accelerating top-line growth. The company attributes demand to AI model training and robotics “foundational models,” supporting broader customer growth and higher spend. A $313M net CoreWeave contract over five years is expected to ramp significantly by mid-2027, implying substantial incremental revenue.

Analysis

The important signal is not simply that a smaller storage vendor is growing faster; it is that AI workloads are changing the mix of storage demand toward high-retention, high-availability capacity where utilization matters more than raw capacity shipped. That tends to improve operating leverage for the best-positioned niche providers, but only if bandwidth and support costs do not rise faster than bookings. The biggest second-order effect is competitive: low-cost cloud storage vendors and bundled hyperscaler storage offers may have to defend share with price, which can limit how much of this demand converts into durable margin expansion.

The near-term market will likely overcapitalize the far-dated contract visibility. Because meaningful ramp is not immediate, the next 1-2 quarters matter more for confirming whether the AI-driven mix shift is lifting gross margin and free cash flow, rather than whether headline revenue continues to accelerate. If customer concentration rises or the growth rate normalizes back toward the prior baseline, the multiple can compress quickly since small-cap infrastructure names are usually priced off narrative continuity, not just backlog.

The contrarian view is that investors may be underestimating how commoditized the storage layer can become once AI demand broadens; if the workload shifts from training to inference or if model builders optimize data retention, storage intensity could disappoint. The clearest falsifier is any sign that growth decelerates below the recent run-rate or that margin expansion stalls despite the better mix. CRWV is mainly a read-through on end-market spending, not a direct beneficiary from this contract economics, so I would not extrapolate this into a full AI-infra basket thesis.

Key Decisions for Investors

• Long BLZE on a pullback over the next 1-3 weeks, sized small, with a 1-3 month horizon into the next earnings print; target a re-rating if growth/margin commentary stays above recent run-rate, and cut if sequential growth or gross margin rolls over.
• If options liquidity is adequate, use a BLZE call spread 3-6 months out rather than outright stock to define risk; the far-dated revenue visibility is real, but the market may be paying too much today for a mid-2027 ramp.
• Watch, do not chase, CRWV as a sentiment read-through on AI infrastructure spend; it is more exposed to compute-capex and funding risk than to this storage demand, so this is not a clean bullish pair.
• Relative-value idea: long BLZE versus a legacy storage incumbent like NTAP on a 3-6 month horizon if the market starts rewarding AI-native storage growth over slower secular growth; thesis fails if BLZE margin leverage does not materialize.