Key Highlights

● DigitalOcean Holdings (NYSE:DOCN) surged 12.62% to close at $120.23 on volume of about 4.59 million shares, roughly 1.37 times normal turnover, ending a five-session losing streak.

● The company repurchased approximately $472 million of its 0.00% convertible senior notes due 2030 on July 23 to 24, funding the buyback with a registered direct offering of about 12.5 million shares.

● Shares are up roughly 150% since the start of the year, yet closed about 36% below their 52-week high of $187.50, reflecting one of 2026's most violently traded names.

● Second-quarter results are due before the open on August 4, 2026, with consensus near $0.26 per share on roughly $276 million of revenue.

A Violent Bounce After a Violent Sell-Off

Shares of DigitalOcean Holdings, Inc. (NYSE: DOCN) surged 12.62% in the latest session to close at USD 120.23, on volume of about 4.59 million shares, roughly 1.37 times the stock's normal turnover. The move ended a five-session losing streak and came four trading days before the cloud infrastructure provider reports second-quarter results.

No company announcement was issued on the day of the rally. The likeliest explanations are positioning-driven: a rebound from a steep drawdown and short covering into a binary event. Short interest stood near 11.9% of the available float heading into the session, a crowded bet that becomes expensive when a stock turns.

That context matters, because DigitalOcean has become one of 2026's most violently traded names. The shares are up roughly 150% since the start of the year, yet they closed about 36% below their 52-week high of USD 187.50. The 52-week low is USD 25.56, and market capitalisation is around USD 14 billion.

From Developer Cloud to "AI-Native Cloud"

DigitalOcean built its business selling simple, predictably priced virtual servers and managed databases to developers, startups and small technology firms, customers underserved by the complexity of Amazon Web Services, Microsoft Azure and Google Cloud. It now serves more than 650,000 customers.

Over the past year it has repositioned around inference, the stage where a trained AI model actually answers queries in production, as opposed to the training stage that consumes the largest GPU clusters. At its Deploy 2026 conference in April, DigitalOcean introduced a five-layer "AI-Native Cloud" spanning roughly 20 data centres with Nvidia and AMD accelerators, core compute and Kubernetes, an Inference Engine with serverless endpoints, data and retrieval services including managed Weaviate, and managed agent tooling.

The strategic claim is that DigitalOcean is not a GPU landlord. Management argues that renting bare-metal accelerators is a commoditising business, and that owning the software layer above the silicon is where durable margin sits. Company-published benchmarks claim faster response times and lower cost than comparable hyperscaler agent services, though those comparisons rest on DigitalOcean's own workload assumptions and are not independent testing.

The Numbers Behind the Re-Rating

First-quarter 2026 results, published on 4 May, were the inflection point. Revenue reached USD 258 million, up 22% year over year, with non-GAAP diluted earnings per share of USD 0.44 against consensus near USD 0.26. GAAP net income was USD 16 million, or USD 0.15 per diluted share. Adjusted EBITDA, earnings before interest, tax, depreciation and amortisation, adjusted for items such as stock compensation, was USD 105 million, a 41% margin.

What moved the stock was the mix. Total annual run-rate revenue reached USD 1.03 billion, up 22%; within that, AI customer ARR hit USD 170 million, up 221%, and ARR from customers spending over USD 1 million a year reached USD 183 million, up 179%. Management noted that more than 80% of AI ARR came from inference and core cloud rather than bare-metal GPU rental.

Guidance was raised sharply. Full-year 2026 revenue was lifted to USD 1.130–1.145 billion, implying 25–27% growth, and the 2027 outlook was raised to more than 50% growth from roughly 30% previously, a rare magnitude of upward revision for an infrastructure business.

A Pre-Announcement That Changed the Story Again

On 7 July, DigitalOcean took the unusual step of pre-announcing. Second-quarter revenue growth had accelerated to approximately 29%, against 14% a year earlier, and adjusted EBITDA margin and non-GAAP EPS would land at or above the high end of prior guidance (revenue of USD 272–274 million, EBITDA margin of 37–38%, EPS of USD 0.20–0.23).

The headline figure was contracted backlog. Remaining performance obligations were expected to exceed USD 800 million, more than ten times the year-earlier level, after an increase of over USD 550 million in the quarter alone. The weighted average life of that backlog stretched from 1.6 years to more than three years, reflecting multiple nine-figure annual customer commitments. The company also secured a further 20 megawatts of data centre capacity for late 2027 and early 2028, taking committed capacity to roughly 155 megawatts.

Chief Executive Paddy Srinivasan said the company keeps winning sophisticated AI customers and that demand continues to accelerate. Chief Financial Officer Matt Steinfort has been blunter, stating that demand for the platform exceeds supply.

Balance Sheet Housekeeping, and the Overhang It Created

On 23–24 July, DigitalOcean repurchased approximately USD 472 million of its 0.00% convertible senior notes due 2030, funding the buyback with a registered direct offering of about 12.5 million shares priced at USD 117.54. Roughly USD 153 million of the notes remain outstanding. The company says 96% of the issued shares retired debt and 4% covered the premium paid to noteholders, and it intends to buy back around 500,000 shares to neutralise that residual dilution.

Economically the deal is defensible: it removes a conversion overhang and frees borrowing capacity for capital expenditure. Mechanically, placing 12.5 million shares days before earnings is a plausible contributor to the slide from roughly USD 141 on 7 July to about USD 107 before the latest session's bounce.

Bull Case, Bear Case

The bull case is that DigitalOcean has found genuine product-market fit in inference at a moment when hyperscaler capacity is tight and enterprises are cost-sensitive. Backlog up more than tenfold, lengthening contract durations and 41% EBITDA margins are not typical of a capacity reseller.

The bear case starts with valuation and concentration. The stock trades near 54 times trailing earnings on trailing revenue of about USD 949 million. Nine-figure commitments from a handful of AI customers introduce concentration risk, and such buyers are often venture-funded and unproven. Capital intensity is rising, management has flagged higher capex per megawatt, squeezing cash generation; adjusted free cash flow was just USD 2 million in the first quarter.

Competition is formidable: AWS, Azure and Google Cloud above, Akamai's Linode, Vultr, CoreWeave and Nebius alongside.

What to Watch

Second-quarter results are due before the open on 4 August 2026, with consensus near USD 0.26 per share on roughly USD 276 million of revenue. Given the July pre-announcement, the bar on the quarter itself is largely set; attention will fall on the 2026 exit growth rate, updated 2027 commentary, capex guidance and the composition of that backlog.

Sixteen analysts carry an average 12-month target near USD 177, far above the current price, a spread reflecting genuine disagreement rather than consensus comfort. This is a high-beta position in which execution and financing decisions can each move the shares double digits in a single session.

This article is general financial information and market commentary only. It does not constitute personal financial advice, an offer, a solicitation or a recommendation to buy, sell or hold any security. Investors should conduct their own research and consider obtaining professional advice before making investment decisions.