DigitalOcean secures $725M for AI cloud expansion

DigitalOcean has arranged a $725 million equipment financing facility to support its expansion into AI cloud services, with an additional $300 million available if further lender commitments are secured. The funding is earmarked for GPUs, CPUs, and related hardware as the company prepares for anticipated demand growth in 2027 and 2028. This move reflects a broader shift in its business model, where capital-intensive AI infrastructure requires upfront investment before revenue materializes from customer workloads.

Financial and operational shifts

The facility matures in September 2030, aligning capital expenditure with the revenue generated by the equipment over time. This structure contrasts with DigitalOcean’s earlier approach, where compute resources were provisioned from a more flexible pool of conventional hardware. AI infrastructure, however, demands pre-financed capacity to meet customer commitments, introducing new operational and financial complexities.

The company’s AI customer annual run-rate revenue reached $234 million in the second quarter of 2026, a 212% year-over-year increase, while total quarterly revenue grew 29% to $281 million. Customers spending over $1 million annually now contribute $259 million in annual run-rate revenue, indicating rapid adoption among larger clients. These customers often sign longer contracts and expect guaranteed capacity, making hardware procurement, data center power allocation, and deployment schedules critical to sales execution.

DigitalOcean had 155 megawatts of committed data center capacity by the end of the second quarter, with an additional 20 megawatts expected in 2027 and 2028. Remaining performance obligations rose to $894 million from $71 million a year earlier, reflecting a growing backlog of contracted business that must be supported by pre-installed infrastructure.

Risks and strategic trade-offs

Debt financing allows DigitalOcean to spread infrastructure costs over the revenue-generating lifespan of the equipment, but it also introduces fixed obligations that future cash flows must cover. The company reported $61 million in adjusted free cash flow for the second quarter, with a margin of 22%, but its full-year 2026 forecast projects margins between 11% and 13% as investment increases. Revenue guidance for 2026 is $1.17 billion to $1.18 billion, representing 30% to 31% growth.

The core risk lies in GPU utilization. Accelerated computing hardware incurs costs regardless of customer demand, and newer generations can render older equipment economically obsolete before its technical lifespan ends. DigitalOcean must align procurement, deployment, and customer onboarding with precision to avoid idle capacity or growth constraints. Early indicators suggest demand is accelerating: customers using its Inference Engine increased token consumption by roughly 30 times over a 60-day period, and 85% of AI customer revenue came from inference and core cloud services rather than bare metal.

The company’s strategy hinges on making GPUs an entry point for broader platform adoption. GPU Droplets integrate with Kubernetes clusters, vector databases, and Spaces Object Storage, while the wider ecosystem includes managed databases, networking, block storage, and developer tools. The goal is to convert AI workloads into multi-service consumption, increasing revenue per customer.

Customer concentration and platform evolution

DigitalOcean’s customer base is shifting toward larger clients. In the second quarter, customers spending at least $100,000 annually accounted for 35% of revenue, while those spending over $1 million represented 23%, up 214% year-over-year. This concentration brings scale benefits but also introduces new challenges: larger AI customers have more volatile workloads, higher support demands, and longer procurement negotiations tied to capacity commitments.

Despite this shift, DigitalOcean retains over 680,000 customers, many of whom are smaller developers. The company continues to emphasize per-second billing, managed Kubernetes, and one-click software deployments to serve this segment, while expanding GPU infrastructure and inference services for enterprise AI workloads. Balancing these two markets requires maintaining reliability and simplicity for smaller users while meeting the demands of larger clients without overcomplicating the platform.