DigitalOcean raises Q2 guidance on AI customer growth

DigitalOcean (DOCN) expects Q2 revenue growth of ~29% (vs ~14% in Q2’25) and projects remaining performance obligations to exceed $800M, growing more than 10x from Q2 FY2025 (up $550M+ in Q2’26). The company also added 20MW of committed data center capacity (total ~155MW) with new capacity coming online in late 2027/early 2028, and expects to be at or above the top end of guidance for adjusted EBITDA margin and non-GAAP EPS. While Nvidia shares slipped on a report about DeepSeek designing its own AI chip, DOCN momentum is reflected in 10 analysts raising earnings estimates upward.

Analysis

DOCN’s signal is less about one-quarter growth and more about a change in workload mix: long-duration inference commitments imply the company is moving from opportunistic developer spend toward stickier production demand. That matters because it can support multiple expansion only if investors believe those contracts convert into repeatable utilization rather than one-off logo wins; otherwise backlog headlines are just forward-loaded revenue with limited terminal value.

The real bottleneck is supply, not demand. The added capacity arrives far too late to drive the next 2-4 quarters, so near-term upside depends on management proving it can monetize existing headroom without sacrificing gross margin to win AI workloads. If the company is forced to buy capacity at unfavorable economics, the market will eventually re-rate this as a volume story with subscale infrastructure economics, not a durable AI platform winner.

Second-order, the custom-chip narrative is a modest negative for NVDA over a 6-18 month horizon because every incremental in-house ASIC project reduces long-run dependency on merchant GPUs at the margin. But near term, that is more about budget reallocation than outright demand destruction: AI spend is still rising, and inference-heavy customers often optimize cost first, which can favor lower-cost clouds like DOCN before it meaningfully hurts frontier GPU demand.

Contrarian view: the stock’s huge run likely leaves little room for another easy leg unless next earnings show not just faster growth but sustained margin discipline and a cleaner capacity path. The consensus may be underestimating how often AI cloud names look strongest exactly when scarcity is at its peak and the economics are least proven.

Key Decisions for Investors

• Buy DOCN only on post-print consolidation or a pullback toward the prior breakout area; target a 3-6 month hold with upside tied to confirmation that AI demand converts into durable RPO and not just a one-quarter beat.
• If already long DOCN, finance it with a partial collar or call spread into earnings to protect against multiple compression if management fails to quantify 2026 capacity conversion and margin durability.
• Pair trade: long DOCN / short NVDA in small size for 6-12 months as a relative-value hedge on the shift from merchant GPU capex to inference/cloud optimization; thesis fails if NVDA re-accelerates capex commentary or if DOCN misses capacity delivery.
• Set a watch item on DOCN gross margin and utilization in the next earnings release; if revenue beats but margin guidance slips, treat the move as sentiment-driven and fade rallies.
• Do not chase the move outright above current levels unless management provides explicit visibility on incremental megawatts and contract duration; without that, risk/reward is unfavorable after a 355% year-long rerating.