DigitalOcean
Usage-based + subscription cloud infrastructure. Self-serve, low-touch, transparent pricing for compute (Droplets), managed Kubernetes/databases, storage, plus AI/ML (GenAI Platform, GPU Droplets via Paperspace) and managed hosting (Cloudways). Land-and-expand across a large SMB/developer base.
Earnings, margins, COGS & capex
Profitable, cash-generative SMB cloud growing mid-teens organically, now inflecting to 20%+ as an AI/GPU-cloud attach (GenAI Platform + Paperspace) compounds off a large developer base. FY2025 gross margin ~60% and adjusted EBITDA margin ~40% fund a step-up in GPU/data-center capex that is compressing gross margin (56% in Q1 2026) and near-term FCF in exchange for a larger AI TAM. Balance sheet is roughly net-cash-neutral after the 2025 convert refinance.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~0¢ is cost of goods and ~86¢ operating expense, leaving ~14¢ of operating profit (~29¢ net).
Revenue trend
Margins
declining on data-center depreciation + colocation costs
guided 37-39% FY2026 (buildout drag)
pressured by capacity start-up costs
normalizing lower ex-tax-benefit
guided down to 9-12% FY2026 on capex/start-up costs
COGS structure
Primary COGS: data-center capacity (power, space, network), server/GPU depreciation, bandwidth, and third-party colocation. GPU/AI compute is more capital- and power-intensive than legacy CPU Droplets, so the mix shift is already pressuring gross margin (61% -> 56% YoY in Q1 2026) even as it expands addressable revenue.
Capex
Step-change up in 2026. Company secured ~60 MW across four new locations (+80% committed capacity), with per-MW capex higher than 2025 gear due to component costs and higher-density AI equipment; ~$100M of one-time capacity start-up costs baked into FY2026, driving the FCF-margin guide-down to 9-12%.
Latest earnings
Beat and raise - revenue above the $249-250M guide and consensus ~$254.7M; non-GAAP EPS $0.44 vs ~$0.27 expected; stock surged ~48% on the print; FY2026 revenue guide $1.130-1.145B (~25-27%)
FY2026: revenue $1.130-1.145B (~25-27% growth); adjusted EBITDA margin 37-39%; adjusted FCF margin 9-12% (incl. ~$100M one-time capacity start-up costs; ~18-21% excluding them)
- AI customer ARR
- $170M, +221% YoY (from ~$53M)
- $1M+ customer ARR
- $183M, +179% YoY
- Net dollar retention
- 101%
- New capacity secured
- ~60 MW across 4 locations (+80% committed)
Growth drivers
- AI/GPU cloud — AI customer ARR reached $170M in Q1 2026, +221% YoY (from ~$53M a year earlier) - fastest-growing segment
- Scaler+ / larger customers: $1M+ ARR customers reached $183M (+179% YoY)
- GenAI Platform (agent/inference), GPU Droplets and Paperspace bringing AI workloads to the SMB/developer base
- Managed hosting via Cloudways and higher-value managed services (Kubernetes, databases) driving expansion
- International and digital-native business growth; product-led, low-cost-to-serve motion
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-02-24. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
DigitalOcean is the profitable, developer-loved on-ramp for the long tail of SMB/AI workloads - reaccelerating from mid-teens to ~25-27% growth as a high-growth AI/GPU attach compounds on a large, sticky, cash-generative base.
- Growth is inflecting: FY2026 revenue guide ~25-27% vs. +16% in 2025, led by AI ARR +221% YoY to $170M
- Real profitability and cash flow (~40% adj EBITDA, ~$168M adj FCF in 2025) - it can self-fund the GPU buildout without diluting like unprofitable AI-cloud peers
- Structural niche: hyperscalers are too complex/expensive for the SMB long tail; DO owns developer experience, docs and transparent pricing
- Land-and-expand improving at the top - $1M+ ARR customers +179% YoY - shows the platform scales upmarket
- Balance sheet is roughly net-cash-neutral after the 2025 convert refinance (~$741M cash vs ~$937M converts), so it enters the buildout from strength
- Optionality: GenAI Platform, GPU Droplets and Paperspace position DO to capture SMB AI inference/agent demand that's still early
A capital-light, mid-teens-growth SMB cloud is spending its way into a capital-intensive GPU business at a rich AI valuation - chasing a segment the hyperscalers and specialist GPU clouds also want, with margins already compressing.
- The re-rating prices in sustained 25%+ AI-led growth; any AI-ARR deceleration de-rates it violently (it moved ~48% on a single print; 52-wk low $25.56 shows the downside)
- FCF margin guided down to 9-12% (from ~19%) as capex and ~$100M one-time start-up costs hit - the capital-light thesis is eroding
- Gross margin already fell 61% -> 56% YoY in Q1 2026; GPU/data-center depreciation structurally pressures unit economics
- AI/GPU cloud is a brutal, commoditizing, depreciation-heavy business; DO is sub-scale vs. CoreWeave/Nebius and the hyperscalers on capex and supply access
- Core (non-AI) growth is still only mid-teens with NDR just 101% - the legacy business isn't accelerating on its own
- GAAP net income flattered by a one-time ~$70M tax benefit + convert-extinguishment gain in 2025 (net income +207%); underlying GAAP earnings power is far lower (Q1 2026 net margin ~6%)
- A ~$312M 2026 convert tranche matures in Dec 2026 - covered by cash today, but a reminder the converts still need managing
What it is worth
Relative multiples (EV/Sales, P/E) cross-checked vs. growth; the stock re-rated sharply over the past year on AI-ARR momentum (52-wk range $25.56-$187.50).
If AI-ARR growth decelerates, GPU economics/utilization disappoint, or FCF/gross margin stay depressed, the ~14-15x-sales multiple compresses sharply toward historical (low-to-mid single-digit sales) levels - large drawdown risk (52-wk low $25.56 is the memory of the pre-AI multiple).
~25-27% FY2026 growth as guided
EBITDA margin 37-39%, FCF margin dipping to 9-12% then recovering. Fairly valued-to-slightly-rich near current levels; return tracks execution on the AI attach and margin normalization.
If AI ARR keeps compounding (200%+) and total growth holds mid-to-high-20s% with EBITDA margins back toward 40% and FCF recovering post-buildout, a premium growth multiple is defensible and revenue/FCF scale into the valuation - upside toward the prior ~$187 high / analyst targets (~$179 average).
At ~$128 / ~$13.6B market cap on ~$949M TTM revenue, DOCN trades ~14-15x TTM sales and ~57x trailing GAAP P/E (P/E flattered by a 2025 one-time tax benefit + convert-extinguishment gain). The multiple only makes sense if the AI-led reacceleration to ~25-27% is durable and FCF margins recover after the 2026 buildout. Net debt is only ~$196M (cash ~$741M vs ~$937M converts), so EV (~$13.8B) is close to market cap.
SWOT
Strengths
- Large, sticky base of SMBs, startups and developers with a low-touch, self-serve, price-transparent motion (low cost-to-serve)
- Genuinely profitable and cash-generative at scale — ~60% FY2025 gross margin, ~40% adjusted EBITDA margin - unlike many AI-cloud peers
- Fast-growing AI/GPU attach (AI ARR +221% YoY to $170M) that lengthens runway beyond legacy IaaS
- Simplicity/developer-experience brand and extensive documentation/community as a moat vs. hyperscaler complexity
Weaknesses
- Sub-scale vs. hyperscalers (AWS/Azure/GCP) on capex, breadth, geographic footprint and enterprise features
- Net dollar retention only 101% - expansion is modest outside the top AI/large-customer cohorts
- Gross margin already compressing (61% -> 56% YoY in Q1 2026) as GPU/data-center depreciation hits COGS
- Rising capital intensity and one-time buildout costs compress FCF margin (to 9-12% in 2026), a departure from its capital-light history
Opportunities
- Ride AI inference/agent workloads to SMBs priced out of or overwhelmed by hyperscaler AI stacks
- Up-sell managed services (Kubernetes, databases, Cloudways hosting, GenAI Platform) to lift NDR and ARPU
- International expansion and higher-value 'Scaler+' customers moving the mix upmarket
- GPU capacity as a differentiated, developer-friendly alternative to CoreWeave/Nebius for smaller AI builders
Threats
- Hyperscaler pricing power and free-tier/credits pressure at the low end; Cloudflare/Vercel encroaching on developer mindshare
- GPU supply, pricing and depreciation risk - AI-cloud economics can invert if utilization or rental rates fall
- Valuation re-rating risk — after a large run the stock trades at a rich multiple; any AI-ARR deceleration is punished hard (it swung ~48% on one print)
- Customer concentration building in the AI/$1M+ cohorts, and macro sensitivity of the SMB/startup base
Moats, dependencies & bottlenecks
Moats
Simplicity, docs, tutorials and community drive low-CAC, product-led adoption among SMBs/developers - hard to replicate at hyperscaler scale but not a hard lock-in.
Managed databases, Kubernetes, storage and app deployments create workload gravity, but NDR of 101% shows switching costs are real yet modest for the base.
Efficient, self-serve, low-touch operations yield ~60% gross and ~40% EBITDA margins that undercut hyperscalers for small workloads - but scale disadvantage vs. AWS/Azure/GCP on capex.
Early GPU footprint (Paperspace + ~60MW new capacity) differentiates for SMB AI, but GPU supply and pricing are contested by far-better-capitalized players.
Dependencies
AI-cloud growth hinges on GPU supply, allocation and pricing; DO competes for the same scarce chips as vastly larger buyers.
~60MW of new committed capacity; power availability, lease terms and buildout timing gate AI growth and drive the capex step-up.
Base is macro-sensitive and price-elastic; a startup-funding or SMB downturn pressures usage.
~$937M converts (incl. a ~$312M tranche due Dec 2026); refinancing/dilution risk if AI economics or the equity multiple weaken, though cash covers the near maturity today.
Kubernetes, databases, ML frameworks (Hugging Face models etc.) underpin the managed-services and GenAI offerings.
Advantages
- Profitable and FCF-positive at scale - rare among AI-cloud growth stories
- Low-CAC, product-led, self-serve motion with strong developer brand
- Transparent, predictable pricing that undercuts hyperscaler complexity for SMBs
- Fast-growing AI/GPU attach off an existing installed base (distribution advantage)
- Full-stack simplicity (compute + managed DB/K8s + hosting + AI) in one console
Weaknesses
- Sub-scale vs. hyperscalers on capex, breadth and geography
- Modest overall NDR (101%) - limited organic expansion outside AI/large cohorts
- Gross margin compressing (61% -> 56% YoY) and rising capital intensity eroding the historically capital-light FCF profile
- ~$937M convertible notes (a ~$312M tranche due Dec 2026) to manage, though largely offset by ~$741M cash
- GAAP earnings power overstated by a 2025 one-time tax benefit + convert-extinguishment gain
Bottlenecks
- GPU supply and allocation vs. far-larger buyers (hyperscalers, CoreWeave, Nebius)
- Data-center power and capacity buildout timing/cost (per-MW capex rising)
- Gross-margin compression from GPU/data-center depreciation as the AI mix grows
- NDR / expansion outside the top AI cohorts limiting organic acceleration
- Engineering/enterprise feature depth vs. hyperscalers for larger workloads
Top signals & trends
Top signals
AI attach is compounding far faster than the core - the reacceleration engine.
Management confidence in durable reacceleration; the Q1 beat lifted the outlook.
Capital-light thesis eroding; GPU/data-center depreciation + ~$100M one-time buildout costs bite.
Expansion concentrated in top cohorts; the broad base is roughly flat-to-slightly-expanding.
Rich, high-beta multiple leaves no room for an AI-ARR stumble.
Demand visibility strong enough to commit to a large buildout.
Trends
DO's core demographic increasingly needs GPU/inference - directly addressable by GenAI Platform + Paperspace.
Opens a large TAM but invites price competition and depreciation risk from far-better-capitalized players.
Repatriation and cost-consciousness favor DO's transparent, simple pricing.
Per-MW capex up; pressures unit economics and FCF during buildout.
Edge/frontend platforms compete for the same developer wallet and mindshare.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
GPUs for AI/ML compute (GPU Droplets, Paperspace, GenAI Platform) - the critical AI input.
Alternative CPU/GPU/accelerator supply for compute capacity.
Colocation/data-center capacity and interconnection.
Data-center/lease capacity for the buildout.
Supermicro, Arista) Compute, storage and networking hardware for the fleet.
Core base - developers, indie builders, SaaS startups, agencies; low-touch self-serve.
Fast-growing $1M+ ARR and AI cohorts driving expansion; AI ARR $170M (+221% YoY).
Managed-hosting customers acquired via Cloudways.
Hyperscale leader; overwhelming breadth/scale but complex and costly for SMBs - the ceiling DO's customers 'graduate' to.
Hyperscaler with deep enterprise + AI (OpenAI) integration; competes at the upper end and via free credits.
Hyperscaler with strong AI/data stack; targets developers and AI workloads.
Specialist GPU/AI cloud, far larger AI capacity and Nvidia access - direct competitor for AI/GPU workloads.
Fast-scaling AI/GPU cloud (ex-Yandex) targeting AI builders - overlaps DO's AI ambitions.
Edge compute (Workers), R2 storage, developer platform - competes for developer mindshare and serverless/AI-inference at the edge.
Linode gives Akamai a direct DO-style developer IaaS competitor plus edge/CDN.
Aggressive on AI/GPU capacity and pricing; competes for AI workloads upmarket.
Low-cost cloud storage/compute for SMBs/developers - niche overlap on storage.
European developer IaaS (Euronext Paris); competes on price/simplicity. Peers Vultr and Hetzner are private.
Developer PaaS/frontend platforms winning modern app deployment and AI-app hosting mindshare.