This AI Infrastructure Stock Is Up 177% in 2026, and It Is Significantly Cheaper Than Nebius

DigitalOcean's Q2 revenue rose 29% year over year to $281 million, while remaining performance obligations surged 12x to $894 million, signaling a sharply expanding AI-cloud revenue pipeline. ARR from customers spending at least $1 million increased 214% to $259 million, AI-customer ARR grew 212% to $234 million, and inference-service adoption climbed 800%. Management expects at least 50% revenue growth in 2027 versus projected 30.5% growth in 2026; the article argues its 16x trailing sales multiple, roughly one-third of Nebius' valuation, supports further upside.

Analysis

DOCN’s investable question is not whether AI demand exists, but whether contracted demand converts into durable, high-margin revenue rather than capital-intensive GPU resale. The sharp mix shift toward larger accounts can improve net retention and reduce SMB churn sensitivity, but it also changes the company’s historical economics: customer concentration, implementation obligations, and GPU procurement exposure rise materially. Over the next 1-3 months, the stock will trade on evidence that AI bookings are being recognized without a corresponding deterioration in gross margin, free-cash-flow conversion, or debt leverage.

The relative valuation discount to NBIS is directionally justified by different business models. NBIS has scarcity value in dedicated AI capacity and hyperscaler/lab exposure, whereas DOCN must prove that its developer distribution can support a premium platform multiple as workloads scale; its lower price/sales multiple should not be read as a free arbitrage absent comparable growth durability. A more relevant competitive risk is that AWS, Azure and GCP can use credits, bundled inference, and model ecosystems to cap DOCN’s pricing on successful customer cohorts, while CoreWeave and NBIS compete for the same constrained GPU supply.

Contrarian view: the market may be extrapolating a backlog/RPO step-up that includes multi-year, capacity-reservation commitments with materially different cancellation, utilization, and margin characteristics than legacy cloud subscriptions. If AI inference adoption is genuinely sticky, DOCN’s installed base gives it an underappreciated cross-sell path; if it is experimentation-led, revenue recognition and utilization will lag bookings. The thesis is falsified by two consecutive quarters of AI-led growth accompanied by declining gross margin, falling dollar-based net retention, or management reducing 2027 revenue/EBITDA targets.

Near-term upside remains possible after momentum-driven gains, but risk/reward is better on post-earnings confirmation or a pullback than chasing. The key catalyst is the next quarterly disclosure of AI revenue mix, RPO duration/current-versus-noncurrent split, GPU utilization, and incremental gross margin; without these, the claimed valuation discount is not independently actionable.

Key Decisions for Investors

• Place DOCN on a long watchlist for the next earnings release; initiate only if AI revenue conversion and gross margin are stable-to-up sequentially, with a 6-12 month target of 20-30% upside versus 15% downside on a break below the post-earnings low.
• For existing DOCN exposure, trim momentum-driven gains and retain a core position; use a 3-6 month protective put spread around the next results because a bookings-to-revenue conversion miss could compress the multiple quickly despite continued topline growth.
• Avoid a simple long DOCN/short NBIS pair: the names have different capex intensity, customer bases, and supply scarcity exposures. Revisit only after comparing forward revenue growth, gross-margin trajectory, net debt, and contracted-capacity duration following both companies’ next reports.
• Monitor NVDA GPU availability/pricing and hyperscaler inference pricing as external alerts. A decline in accelerator costs and stable inference pricing supports DOCN margins; GPU scarcity or aggressive bundled pricing from AWS/Azure/GCP would undermine the long thesis within 1-2 quarters.