Key Takeaways
- DigitalOcean (DOCN) shares surged 12.6% to $126.67 following a strategic presentation at Goldman Sachs Communacopia + Technology Conference
- Company unveiled AI-native cloud platform emphasizing inference over training workloads
- Inference services now generate approximately 85% of AI-related revenue with superior margins
- Company acquired 20 megawatts of additional capacity and saw spot instance offerings sell out within minutes of launch
- Executives upgraded 2024 exit growth outlook to above 35% and 2027 annual growth projection to above 50%
Shares of DigitalOcean (DOCN) climbed 12.6% to finish at $126.67 on Tuesday following a strategic presentation by CEO Paddy Srinivasan at the Goldman Sachs Communacopia + Technology Conference.
DigitalOcean Holdings, Inc., DOCN
Year-to-date, the stock has surged 160%, although it remains approximately 30% below its 52-week peak of $181.29 reached in June 2026.
The central theme from leadership emphasized DigitalOcean’s transformation into an AI-native cloud infrastructure designed for autonomous agents rather than human operators.
Srinivasan articulated the shift directly: “Cloud 1.0 was built to support applications that were mostly created, deployed, and managed by humans. But now, the cloud we need to build supports applications created by agents.”
The platform’s strategy prioritizes inference operations over training, with management asserting that inference workloads offer greater sustainability and monetization potential.
Inference Services Driving Revenue Composition
Currently, approximately 85% of DOCN’s artificial intelligence revenue originates from inference-related offerings, encompassing token savings programs, reserved instances, and spot instances. Bare metal AI services account for the remaining 15%.
The company’s core cloud infrastructure maintains gross margins around 70%, representing its highest-margin segment. Leadership disclosed that GPU list pricing has increased roughly 30%, driven by market dynamics and proprietary software enhancements.
CFO Matt Biilmann emphasized the company’s competitive advantage stems from flexible contract terms that enable rapid pricing optimization.
“When you enter the token economy, the question isn’t about supply and demand, how many GPUs do you have? It’s about how many tokens you can provide me and at what quality level?”
The token-based service, launched approximately 120 days prior to the conference, has already attracted between 6,000 and 7,000 customers.
Infrastructure Expansion and New Product Rollouts
From an infrastructure perspective, DigitalOcean commissioned three additional data centers this year, each ahead of projected timelines. The organization secured 20 megawatts of additional capacity following its previous guidance announcement.
Spot instances, introduced weeks before the conference, experienced immediate sell-out within minutes of availability. The company also unveiled Agent Harness and Open Harness Runtime, enabling customers to incorporate frameworks such as Hermes, Codex, and OpenClaw directly into their workflows.
The platform’s sandbox infrastructure can initialize an agent in hundreds of milliseconds and execute restarts in under 100 milliseconds, dramatically outpacing the multi-minute timelines associated with traditional virtual machine environments.
From a commercial perspective, DigitalOcean appointed Kevin, previously with Vercel, to serve as Chief Revenue Officer.
Leadership elevated its 2024 exit revenue growth projection to above 35% and its 2027 annual growth target to above 50%, with comprehensive details anticipated alongside November financial results.
Management noted that H100 pricing has maintained upward momentum, including through recent weekends, while next-generation GPU architectures are delivering improved token output efficiency per megawatt of power consumption.


