DigitalOcean's Q3'26 Earnings Conference Call - DIGITALOCEAN HOLDINGS INC

Explicit Guidance:

• Q4 2026 Revenue: $348–$351 million, representing 44–45% year-over-year growth.
• Full Year 2026 Revenue: $1.208–$1.211 billion, representing ~34% year-over-year growth.
• Full Year 2027 Revenue: Projected growth of 60% or more (increase from prior 50%+ outlook) on a materially higher 2026 base.
• Q4 2026 Adjusted EBITDA Margin: 42–43%.
• Full Year 2026 Adjusted EBITDA Margin: 41–42%.
• Full Year 2026 Adjusted Free Cash Flow Margin: 20–22%.
• Q4 2026 Non-GAAP EPS: $0.30–$0.32 on ~128–129 million weighted average fully diluted shares.
• Full Year 2026 Non-GAAP EPS: $1.60–$1.62 on ~122–123 million weighted average fully diluted shares.
• 2027 Margins/Cash Flow: Detailed guidance for FY2027 margins and cash flow expected in the Q4 2026 earnings call (February 2027).

Business-Segment Outlook:

• AI-Native Cloud Adoption: Customers are shifting from simple token consumption to "agentic" workloads requiring runtime, tools, memory, and learning loops; 90% of AI revenue now derives from customers also consuming core cloud primitives.
• Product Mix & Attach: 60% of revenue comes from customers using four or more products; 90% of AI customer ARR comes from inference services and core cloud (inference services up ~600% YoY).
• Agent Growth: Agent-related traffic represents ~50% of unique website visitors; >37,000 customers deployed via agents in the current year.
• Yield per Megawatt: Targeting sustained ARR of ~$18 million per average active megawatt, driven by software, platform attachment, and token/agentic workloads rather than just capacity additions.
• Core Cloud: Stable, highly profitable ~$900 million ARR business growing at 16%+ (up from 16% last year, accelerating to 40% overall) with AI attachment driving traditional core products.
• Capacity Ramp: Memphis (Q3), Kansas City (Q2), and Richmond (Q1) data centers launched ahead of schedule; 75+ megawatts of incremental capacity expected to come online in 2027.

Market and Competitive Outlook:

• Supply/Demand Dynamics: Demand continues to far outstrip supply; management expects supply-demand normalization to eventually lower spot prices but anticipates use cases (personal agents, media production) will expand the market pie faster than price normalization impacts yields.
• Competitive Positioning: Differentiates from bare-metal focused providers by offering a complete platform (tokens, agents, data, runtime) rather than point solutions; claims "3 to 4 months ahead" on model router technology.
• Pricing Power: Ability to capture rising prices via consumption and short-term contracted revenue models; closed deals well north of $20 million ARR per megawatt.
• Customer Concentration: Minimal risk; top 25 customers represent only 27% of revenue with no single customer in the high single digits.
• Technology Shifts: Expectation that agentic AI will lift global token consumption 24x by 2030; AI "center of gravity" shifting from human chat to autonomous agents performing work.

Macro / Industry Outlook:

• Economic Environment: Management expresses high confidence in the "agentic era," viewing current market nervousness as a differentiation opportunity due to DigitalOcean's balanced financial profile (profitable, low leverage, flexible balance sheet).
• Funding Environment: Notes that the current debt market favors strong financial discipline; DigitalOcean's minimal leverage (1.2x net leverage) and flexible capital structure (new $785M equipment lease, $1B ATM program) position it to access funding efficiently.
• Regulatory/Structural: No specific regulatory changes anticipated; focus remains on structural shifts where "self-build" is not viable for most customers compared to the platform model.

Strategic Initiatives:

• Capacity Expansion: Committed 180 megawatts total; secured 20 megawatts for 2028 in Q3; 75+ megawatts planned for 2027 (ramping H2 faster than H1).
• Technology Upgrades: 2027 rollout expected to shift from B300-generation GPUs to GB300s and potentially Vera Rubin or similar next-gen technologies to improve token efficiency and ARR per megawatt.
• Financing: Proactive $1 billion at-the-market (ATM) equity program established for option value; plan to retire remaining $312 million of 2026 convertible notes by December 2026.
• Product Innovation: Continuous rapid iteration (70 releases in 65 days); launch of "Agent Droplets" to simplify entry for non-AI customers into agentic workflows.

Risks and Headwinds to Outlook:

• Data Center Execution: Dependence on tier-one co-location providers for timely capacity delivery; delays in physical build-outs could impact ramp speed (though management cites strong track record).
• Technology Cost/Capex: Newer GPU generations (GB300, Vera Rubin) involve higher CapEx per megawatt ($35–$40M range), though yields are expected to remain favorable.
• Margin Pressure: Gross margins may face temporary pressure from depreciation and lease costs of new data centers before revenue ramps.
• Market Normalization: Potential downward pressure on spot prices and token costs as supply increases, though management expects platform attachment to mitigate this.
• Model Efficiency: Increased model efficiency could theoretically reduce token consumption volume, though management argues agentic workloads will drive volume growth.

Management Tone on the Future:

• Confidence Level: Extremely high; CEO states, "I have never been more confident in where DigitalOcean is headed."
• Differentiation Narrative: Repeated emphasis on being "different by design" regarding platform completeness, financial discipline, and customer diversification compared to hyperscalers and bare-metal providers.
• Guidance Trajectory: Significant upward revision in expectations; raised FY2026 revenue midpoint by $120 million and FY2027 growth rate from 30-50% to 60%+.
• Strategic Outlook: Views the "agentic era" as a massive opportunity where the current platform structure is ideally suited, with a belief that growth is durable, profitable, and driven by software yields rather than just hardware deployment.