DigitalOcean: Growth Acceleration Is Real, but the Stock Has Outrun What It Can Prove Next

DigitalOcean (NYSE: DOCN) reported second-quarter 2026 results on August 4 that beat on revenue, EPS, and almost every operating metric. The problem for investors is the same one that has been building for months: the stock has climbed 321% over the past year and 174% year-to-date, and the valuation now demands flawless execution through 2027 or more to justify where it sits. The growth story is genuine. The risk/reward at this price is not as clean as it was three quarters ago. My posture is Hold - wait for a pullback or two consecutive quarters of 50% revenue growth before buying.

What Changed in Q2

Revenue came in at $281 million, up 29% year-over-year, well above the roughly 14% growth rate a year ago and accelerating from 22% in Q1 and 18% in Q4 2025. Non-GAAP EPS of $0.45 beat the consensus estimate of $0.26 by a wide margin. Adjusted EBITDA (earnings before interest, taxes, depreciation, and amortization - a rough cash-earnings proxy) was $114 million at a 40% margin. The stock opened at $119 and climbed to around $137, adding about 4% on the day as the beat landed.

The acceleration is the headline, but the details matter more than the headline. Revenue growth from customers spending $500K+ per year jumped 160% year-over-year, and revenue from million-dollar customers surged 214% to $259 million in annual run-rate revenue. AI customer ARR (annual recurring revenue from customers using AI-related services) grew 212% to $234 million. The company signed its first nine-figure annual customer commitments, extending the weighted average contract life from 1.6 years to over 3 years. That kind of deal-size and duration shift tells you the customer base is maturing beyond the small-and-mid-market developer base that built this company.

Remaining performance obligation - the pool of contracted revenue not yet recognized on the income statement - hit $894 million, up 12 times from $71 million a year ago. Of that, $366 million is expected to be recognized in the next 12 months. This is the single most forward-looking metric DigitalOcean reports, and the move from $71 million to $894 million is not a rounding error. It is a structural change in the revenue visibility profile.

The Cash Flow Tension

Here is where the story gets more complicated. Q2 adjusted free cash flow was $61 million at a 22% margin, a strong quarter. But trailing twelve-month free cash flow was only $41 million, down 57% year-over-year. The reason: capex hit $268.5 million over the past year. Management has committed to approximately 155 MW of data center capacity, with 20 MW of additional capacity signed for late 2027 and early 2028.

The company funded part of this buildout with an $888 million equity raise in Q1 and a separate registered direct offering that financed the repurchase of $472 million in convertible notes. Total debt stands at $1.68 billion and equity at $887 million. The debt-to-equity ratio is 103.6%.

The operating model is simple: DigitalOcean is spending heavily today on GPU capacity and data center infrastructure so it can deliver on the contracted revenue sitting in its $894 million RPO pool. If that capacity comes online on schedule and the nine-figure commitments convert to actual usage at projected run rates, the capex converts to high-margin revenue. If adoption stalls, pricing compresses, or capacity buildout slips, that $268 million in capex becomes a drag on returns.

GAAP net income was $35 million in Q2, flat-to-down from a year ago, and the net income margin fell to 13% from a higher base. That is a function of stock-based compensation, depreciation, and the accounting treatment of the convertible notes - not a sign of deteriorating economics, but a reminder that GAAP profitability is thin relative to the market cap.

Valuation: The Bridge Has Stretched

DigitalOcean has a $15.5 billion market cap and a 16.5 times trailing EV/sales multiple. On a run-rate basis from Q2 revenue, that compresses to roughly 14 times. The stock trades at 103 times forward earnings, 50 times EV/EBITDA, and 17 times book value. The PEG ratio - price-to-earnings divided by growth rate - prints around 0.55, which looks cheap, but the PEG is backward-looking and assumes growth will sustain at current acceleration.

Management now expects roughly 30% revenue growth for fiscal 2026 and more than 50% for 2027. If 2026 revenue hits approximately $1.17 billion, the market cap implies about 13 times forward sales. If 2027 revenue reaches the $1.75 billion range at 50% growth, it drops to roughly 9 times forward sales. Those are not outrageously expensive multiples for a cloud infrastructure company. But they are only defensible if the 50% growth actually happens.

The stock has already priced in that 50% growth and more. A 321% annual return means the market is paying for a perfect outcome - accelerated capacity buildout, conversion of those nine-figure commitments, AI inference demand staying hot, and no margin erosion from competitive GPU pricing pressure. The valuation bridge is there, but it has only enough room for one major stumble.

What Would Break the Thesis

Three risks dominate:

• AI demand concentration. Eighty-five percent of AI customer ARR comes from inference and core cloud products rather than bare metal GPU rental, which is a good product mix. But 234 million in AI customer ARR out of a total $1.125 billion ARR means roughly 21% of the revenue base now depends on AI demand holding. If inference spending decelerates or customers shift to in-house infrastructure, that growth engine slows.

• Capex execution risk. The company is adding data center capacity that will not come online until 2027 and 2028. Construction delays, cost overruns, or capacity that sits idle would pressure margins and returns. The $888 million equity raise reduced leverage but diluted shareholders; additional funding is possible if demand outpaces the current buildout plan.

• Competitive compression. The hyperscalers (AWS, Azure, Google Cloud) have far more GPU inventory and can undercut on pricing if they choose. DigitalOcean's positioning around cost transparency and ease of use for AI builders is a real advantage for mid-market customers, but the total addressable market for inference workloads is attractive enough that the big players will compete aggressively.

The Verdict

DigitalOcean is not a broken business wearing an AI narrative. Revenue growth is genuinely accelerating, the RPO pipeline is real and growing at structural pace, adjusted margins are holding around 40%, and the large-customer base is expanding at triple-digit rates. The AI-Native Cloud platform is not vaporware - early customers like Higgsfield AI and Hippocratic AI are running production workloads, and early Inference Engine users drove token consumption up approximately 30 times in the last 60 days.

But the stock has done the hard work for you already. A 321% annual gain at $132 a share means the risk/reward reset has not occurred. At 103 times forward earnings and 16.5 times sales, you are paying for the 50% growth scenario with minimal margin for error. The Cheap-Enough Bridge - where valuation resets faster than the business deteriorates - does not apply here because there is no reset.

I am rating DigitalOcean a Hold. The growth story is intact, the Q2 results are strong, and the pipeline suggests 2027 could be the explosive year management describes. But the entry point matters, and this stock has outrun what it can prove in the next two quarters. Wait for either a meaningful pullback that brings the forward multiple closer to the 8-to-10 times sales range where the growth story becomes compelling, or wait for Q3 and Q4 to deliver consecutive quarters that validate the 35%+ exit-2026 growth rate. Until one of those conditions clears, patience is the better trade.

Rating: Hold