Digital Ocean’s Q4 2025 Earnings Report
On February 24, 2026, DigitalOcean (DOCN) shared a strong fourth-quarter earnings report for 2025. The numbers came in well above Wall Street estimates. The cloud provider posted a fourth-quarter revenue of $242.39 million. This is an 18% increase from the same time last year, stepping past the consensus estimate of $237.7 million. Non-GAAP earnings per share reached $0.44, beating the expected $0.38.
Company leadership also shared positive guidance for the future. CEO Paddy Srinivasan mapped out a faster pace of growth. He projected that full-year 2026 revenue will land between $1.075 billion and $1.105 billion, which equals a growth rate of roughly 21%. DigitalOcean expects to finish 2026 with a growth rate of 25% or more and aims to reach 30% growth by 2027. The company is tracking toward a weighted Rule of 50 profitability metric.
Growth Driven by the Agentic Inference Cloud
Artificial intelligence is the main factor behind this recent growth. Annual Recurring Revenue (ARR) from AI customers reached $120 million. This is a 150% increase compared to the previous year. The company is also seeing strong movement with larger clients. ARR from customers spending over a million dollars reached $133 million, which is a 123% increase year over year. In total, DigitalOcean added a record $51 million in organic incremental ARR during the quarter.
The main driver of this success is the Agentic Inference Cloud. This platform connects high-performance AI infrastructure with the company’s traditional general purpose cloud products. DigitalOcean offers a complete software stack for AI native companies. These companies use the platform to move from training models to running them in actual production. Developers can easily manage complex tasks. For example, they can use CPU droplets to prepare data before sending the work to GPU clusters. The platform is seeing heavy use. Customers have created over 19,000 AI agents and put more than 7,000 of them into production.
DigitalOcean is adding physical space to keep up with customer needs. The company secured 30 megawatts of new data center capacity. This new space will become available throughout 2026 and 2027. This growth is tied to solid business agreements. Shortly after the quarter closed, the company signed multiple 8-figure, multiyear committed contracts with AI native businesses.
Financial Strategy and the Neocloud Market
Some analysts have pointed out concerns about capital efficiency. During the fourth quarter, operating cash flow dropped 19.7% to $57.28 million. Cash flow margins also shrank from 35% to 24%. To balance its investments with incoming revenue, DigitalOcean is choosing to finance and lease the equipment for its data centers and GPU capacity.
This approach is very different from heavily funded neocloud competitors like CoreWeave. Analysts often note that these neocloud companies are making massive direct deals with Nvidia, relying on complex financial engineering. CoreWeave has secured billions in debt backed by Nvidia hardware. The company also receives direct equity investments from the chipmaker. People closely watch this pattern and often call it circular finance. In this setup, Nvidia invests in a neocloud, and the neocloud then uses that money to buy Nvidia chips. Analysts worry this can make demand and valuations look artificially high.
Competitors take on tens of billions in debt to own their hardware outright and build multi-gigawatt facilities. Because DigitalOcean leases its capacity instead, some investors worry the company might miss out on bulk vendor pricing or economies of scale.
The Strategic Benefits of Leasing Hardware
Leasing equipment might actually be DigitalOcean’s biggest advantage. The AI infrastructure market requires a lot of money, and technology becomes outdated fast. Computing hardware has a short useful life. Nvidia pushes a tight one-year release schedule for new designs. This makes older equipment lose value quickly.
By leasing instead of taking on billions of dollars in debt to own hardware, DigitalOcean stays flexible. Competitors who used debt to buy fleets of last-generation Nvidia H100 or early Blackwell GPUs might soon be stuck with outdated equipment. If the market moves toward new, highly efficient chips built specifically for inference, neoclouds with heavy debt will be holding legacy hardware.
Cerebras is a great example of this new technology. The company builds hardware specifically for AI inference. Its wafer-scale architecture, like the WSE-3, stores entire models directly on the chip. This completely removes the bandwidth delays found in traditional GPUs. Recent tests show that Cerebras can run large open-source models like OpenAI’s GPT-OSS 120B at over 3,000 tokens per second. Nvidia’s newest Blackwell GB200 systems reach 650 tokens per second. For open-source models like Llama 3 70B, the Cerebras CS-3 system is reportedly up to 21 times faster than Nvidia’s Blackwell B200. It also uses less energy and costs less. The price-performance ratio is clear. Cerebras delivers 4,000 tokens per dollar, while Blackwell delivers 1,300.
As open-source models improve, the most efficient hardware will win the inference market. DigitalOcean’s leasing strategy puts it in a great position. The company can smoothly switch to next-generation chips like Cerebras without losing money on past purchases. They can offer customers the fastest and most affordable hardware, keeping a strong edge over debt-heavy neoclouds.
Focusing on Inference Over Training
DigitalOcean has made a very profitable choice by focusing on AI inference instead of AI training. The wider neocloud market mostly focuses on training AI models. Training requires massive groups of bare-metal GPUs. It is fast becoming a crowded space where companies compete to offer the lowest prices. Training work is also temporary. Once a model finishes training, the massive computing need disappears.
Inference is different. It is the ongoing process of running a model to serve real users. It requires a much deeper software ecosystem. This includes storage, networking, security, and agentic workflows. DigitalOcean targets this specific area. More than 70% of its AI customer ARR comes from inference services and core cloud products, rather than basic bare-metal GPU rentals.
This choice turns DigitalOcean into a reliable platform with high profit margins, rather than a basic provider of commodity infrastructure. Inference workloads are stable and easy to predict. They grow as the customer’s own user base grows, working much like traditional SaaS recurring revenue. DigitalOcean supports the whole development cycle. The platform offers everything from a virtual private cloud to managed databases and generative media model hosting. Because of this, customers have no reason to leave as they scale.
Looking Ahead
DigitalOcean’s earnings report for February 24, 2026, shows a practical approach to the growing AI industry. Analysts may have questions about leasing and short-term profit margins, but the company has a clear long-term plan. DigitalOcean is avoiding the massive technical debt that affects highly leveraged neoclouds. They achieve this by focusing on reliable inference workloads and using flexible leasing agreements to grow. As highly efficient chips like Cerebras enter the market, DigitalOcean has the flexibility to adopt them. This practical strategy helps the company support open-source AI and maintain a steady, profitable role in cloud infrastructure.
Disclaimer:
All views expressed are my own and are provided solely for informational and educational purposes. This is not investment, legal, tax, or accounting advice, nor a recommendation to buy or sell any security. While I aim for accuracy, I cannot guarantee completeness or timeliness of information. The strategies and securities discussed may not suit every investor; past performance does not predict future results, and all investments carry risk, including loss of principal.
I may hold, or have held, positions in any mentioned securities. Opinions herein are subject to change without notice. This material reflects my personal views and does not represent those of any employer or affiliated organization. Please conduct your own research and consult a licensed professional before making any investment decisions.