DigitalOcean tops second quarter expectations amid surging AI demand

DigitalOcean Holdings Inc. today posted second quarter earnings that topped analyst expectations across the board.

The company also raised its third quarter guidance.

DigitalOcean launched in 2011 to provide an alternative to the industry’s three leading public clouds. The company says that its platform is easier to use and provides lower pricing across some services. DigitalOcean has more than 680,000 customers, many of whom are developers who use it for small projects.

The company’s revenue jumped 29% year-over-year in the second quarter to $281 million, topping the consensus estimate by about $2.6 million. The company’s growth was driven by a surge in demand for its artificial intelligence services. According to DigitalOcean, annualized recurring revenue from AI customers jumped 212% on a year-over-year basis to $234 million.

“Our growth rate is accelerating, as revenue grew 29% year-over-year, more than double our growth rate a year ago,” said DigitalOcean Chief Executive Officer Paddy Srinivasan. “The acceleration is coming from our highest spending customers and sophisticated AI Natives, and we are now beginning to land nine-figure annual commitments.”

The company introduced more than a dozen AI features in the second quarter. One of the main highlights is Inference Engine, which helps developers build applications that use multiple AI models. The service sends each user prompt to the most suitable model based on considerations such as processing cost and latency. If a neural network experiences technical issues, Neural Engine can reroute requests to a backup algorithm.

DigitalOcean offers multiple infrastructure options for inference workloads. It provides a serverless platform that is priced per token and dedicated servers equipped with 8 graphics cards apiece. Unlike the former offering, the machines don’t run pre-installed virtualization software. That enables users to customize how their AI models use dedicated servers’ hardware resources, which can boost efficiency.

DigitalOcean also offers several other AI-optimized services. The lineup includes a managed version of Weaviate, an open-source database specifically optimized for inference. It stores records as vectors, mathematical structures that large language models can easily understand.

DigitalOcean’s services run on 15 data centers in a dozen cities. The company disclosed today that it plans to buy 20 megawatts of additional computing capacity through 2028. According to DigitalOcean, the purchase brings its total commited capacity to about 115 megawatts.

Despite its investments in AI infrastructure, the company managed to significantly boost its profitability in the second quarter. DigitalOcean’s adjusted EBITDA, or earnings before interest, taxes, depreciation and amortization, rose 27% to $114 million. That amounted to $0.45 per share on an adjusted basis, well above the $0.26 expected by analysts.

DigitalOcean raised its third quarter outlook to earnings of $0.28 to $0.30 per share on $305.5 in revenue at the midpoint, slightly above the consensus estimate. The company expects to end the year with total sales of $1.17 billion to $1.18 billion.

Photo: DigitalOcean