DigitalOcean (DOCN) investor relations material

Strategic evolution and AI-native platform development

• Focus shifted from AI training to inference, leveraging core strengths in developer platforms and global infrastructure.

• Built a deeply integrated platform, prioritizing software development before scaling hardware, to serve agent-driven applications in the emerging AI-native cloud.

• Customer base of 680,000 enabled broad feedback and rapid iteration, with most platform components built in-house for seamless integration.

• Emphasis on agentic workflows and token flow, with new products like Agent Harness enabling efficient, persistent, and secure agent operations.

• Platform designed to maximize value creation through both token and agent flows, moving economics from hardware-centric to software-centric models.

Market trends and customer adoption

• Coding remains a dominant inference use case, but generative media, personal productivity, and go-to-market automation are rapidly growing verticals.

• Customers initially use closed-source models for product-market fit, then shift to open-weight models for cost and control as they scale.

• The gap between frontier and near-frontier models is narrowing, driving more companies to advanced open-weight models and agentic workflows.

• Durable value lies in token and agent flows, not in temporary GPU scarcity pricing.

Economic model and pricing strategy

• Only 15% of AI revenue comes from bare metal GPU; 85% is from higher-margin inference and managed services.

• Software-driven services enable daily price optimization and higher ARR per megawatt, with flexible, short-term contracts allowing rapid repricing.

• Focus on maximizing fleet utilization through elastic compute, spot instances, and global traffic management.

• Software innovation increases the value of each megawatt, with core cloud and AI services driving additional revenue streams.