DigitalOcean is a cloud infrastructure provider historically focused on SMBs, developers, and startups. It's now pivoting to position itself as an "agentic inference cloud" — targeting AI-native companies that need GPU inference capacity alongside traditional cloud primitives (compute, storage, networking). Revenue is usage-based with recurring characteristics. ~640,000 customers, top 25 = ~10% of revenue.
| | Q124 | Q224 | Q324 | Q424 | Q125 | Q225 | Q325 | Q425 | | | Mar-24 | Jun-24 | Sep-24 | Dec-24 | Mar-25 | Jun-25 | Sep-25 | Dec-25 | |---|---|---|---|---|---|---|---|---| | Revenue ($m) | 184.7 | 192.5 | 198.5 | 204.9 | 210.7 | 218.7 | 229.6 | 242.4 | | YoY % | 11.8% | 13.3% | 12.1% | 13.3% | 14.1% | 13.6% | 15.7% | 18.3% | | QoQ Rev Add ($m) | 3.8 | 7.8 | 6.0 | 6.4 | 5.8 | 8.0 | 10.9 | 12.8 | | ARR ($m) | 739 | 781 | 798 | 820 | 843 | 875 | 919 | 970 | | NDR % | 97% | 97% | 97% | 99% | 100% | 99% | 99% | 101% | | GM % [GAAP] | 59.1% | 61.0% | 60.2% | 61.5% | 61.4% | 59.9% | 59.6% | 58.7% | | EBITDA Margin % | 40.2% | 42.4% | 43.7% | 41.9% | 41.0% | 40.9% | 43.5% | 41.0% | | FCF Margin % | 18.6% | 19.4% | 13.2% | 17.9% | 16.3% | 26.1% | 37.0% | 11.1% | | SBC ($m) | 22.9 | 21.8 | 22.9 | 22.9 | 19.4 | 21.1 | 19.8 | 20.0 |
| Metric | Value |
|---|---|
| Stock Price | ~$87 |
| Market Cap | $9.0B |
| Enterprise Value | $10.4B |
| Trailing P/S (TTM $901M) | 10.0x |
| Forward P/S (FY26E $1.09B) | 8.3x |
| EV/Gross Profit (TTM $540M) | 19.3x |
| EV/EBITDA (TTM $375M) | 27.7x |
| Net Debt | $1.04B |
| Shares Diluted | ~111M |
| Tier | ARR | YoY Growth | NDR |
|---|---|---|---|
| Total | $970M | +18% | 101% |
| DNE (>$500/mo) | $640M | +30% | 102% |
| $100K+ (635 customers) | 28% of rev | +58% | -- |
| $1M+ | $133M | +123% | 115% |
| AI Customers | $120M | +150% | -- |
Let me be clear: DOCN fails two of my six first-filter criteria, and that matters.
| Criteria | Status | Notes |
|---|---|---|
| Revenue growth >30% | FAIL | 18% YoY. Guided 21% for FY26. |
| Recurring revenue model | PASS | Usage-based cloud, high recurring nature |
| Gross margins >70% | FAIL | 59% GAAP. Declining. AI mix headwind. |
| Not capital-intensive | BORDERLINE | PP&E up 36% YoY to $589M. Finance leases ramping. |
| Not too complicated | PASS | Understandable business |
| No single-customer dependency | PASS | Top 25 = ~10% of revenue |
The numbers have to match the theory. An 18%-growth cloud company with 59% gross margins and $1.3B in debt is a fundamentally different animal than the SaaS companies I typically invest in.
Revenue acceleration is genuine. Four consecutive quarters of YoY acceleration: 12.1% -> 13.3% -> 14.1% -> 13.6% -> 15.7% -> 18.3%. The incremental ARR record of $51M in Q4 is meaningful — up from $24M a year ago.
Large customer traction is impressive. $1M+ customer ARR at $133M growing 123% YoY. $100K+ customers at 635, up 26%. The company is successfully moving upmarket. This is the most compelling part of the story.
AI customer revenue is real, not vaporware. $120M AI ARR growing 150%, and 70% of that is inference services + core cloud, not bare metal GPU rental. The inference differentiation narrative has some substance behind it.
NDR recovery. Bottomed at 96% in Q3 FY23. Now 101%. Still below the 120%+ I want to see from best-in-class, but the trend is right.
Management executing on the pivot. Paddy Srinivasan has repositioned this company from "simple cloud for developers" to "agentic inference cloud" with credible early traction. The OpenClaw example — 30,000 agents deployed with zero marketing — is the kind of organic pull you want to see.
Profitability is real. $375M EBITDA on $901M revenue (42% margin). Non-GAAP net income $218M for FY25. This isn't a money-burning hope story.
Gross margins are structurally limited and declining. 63% in FY22, 59% now, and the CFO acknowledged AI margins are lower than core cloud margins. As AI becomes a larger mix, gross margins face a structural headwind. This is the opposite of what I want — I want improving unit economics, not deteriorating ones. Every cloud company I've invested in successfully has had 70%+ gross margins.
18% growth doesn't clear my bar. Yes, it's accelerating. Yes, management guides 25%+ exit rate for Q4 2026 and 30% for FY27. But I don't invest based on hope. Right now the numbers say 18%, and the pathway to 30% depends on capacity ramp timing that has execution risk. Management itself flagged "supply chain and implementation timing risks."
The debt load is substantial. $1.3B in total debt, $1.04B net debt. $325M in convertible notes due December 2026 — this year — that needs to be refinanced or repaid. Management expects net leverage to temporarily exceed 4x. For a company with $901M revenue, carrying >1x revenue in net debt is a real constraint on capital allocation flexibility.
Diluted share count tells a story. Diluted shares went from 94.4M in Q4 FY24 to 111.5M in Q4 FY25 — an 18% increase, almost entirely from convertible note dilution. The company has spent $1.6B on buybacks since IPO, and the share count is barely down from when they started. That's not accretive capital allocation — it's fighting the tide.
Capital intensity is increasing. PP&E up 36% to $589M. Finance lease liabilities jumped from $4.6M to $130.5M in one year. Operating lease obligations up 44%. This is becoming a capital-intensive infrastructure business, which is exactly the profile I try to avoid.
Competitive intensity is extreme. 32 companies in SemiAnalysis benchmarking. DigitalOcean is competing against hyperscalers (AWS, Azure, GCP) on one end and Neo Clouds (CoreWeave, Lambda, Crusoe) on the other. The CEO's framing — "we put the cloud in Neo Cloud" — is clever positioning, but I need to see more evidence that this moat is durable.
FY26 margin compression. Adj EBITDA margin guided to 36-38% (from 42% in FY25). FCF margin guided 15-17% (from 19%). This is the cost of the capacity ramp. The Rule of 50 target for FY27 is aspirational — at 30% growth + 20% FCF margin = 50. Both numbers are projections, not results.
Non-GAAP adjustments are aggressive. "Adjusted FCF" adds back acquisition-related compensation (~$33M/yr). Unlevered FCF strips out interest expense. These aren't unusual for the industry, but they make the profitability picture look better than GAAP reality. Interest expense quadrupled from $9M to $18M in FY25.
At $87 per share, the market cap is $9.0B and EV is $10.4B.
P/S heuristic: Growth rate (18%) vs. trailing P/S (10x). Growth exceeds P/S, which by my heuristic suggests reasonable-to-fair valuation. But this heuristic was designed for high-growth SaaS companies with 75%+ gross margins. Applying it to a 59%-gross-margin cloud infra company with $1B net debt requires a discount.
Forward P/S: ~8.3x on FY26E revenue of 1.09B. 6.3xonFY27Erevenueof 1.4B (if they hit 30%). If the acceleration plays out, there's room for the stock to work. If it doesn't, 10x trailing P/S on 18% growth with 59% gross margins is expensive.
EV/GP: $10.4B / $540M = 19.3x. This is the more honest number for a lower-margin business. Compare to a SaaS company at 80% GM growing 25% that trades at 15x EV/GP — DOCN is not cheap on this basis.
Market cap sanity check. "Is this really a $9B company?" Possibly, if the AI inference pivot delivers sustained 25-30% growth. At $1.4B FY27 revenue with 40% EBITDA margins, you'd have $560M EBITDA, and 18x EBITDA = $10B EV. That math works, barely, if everything goes right. But there's no margin of safety in this valuation.
Analyst consensus. Average target $68, stock at $87. Trading 28% above consensus. Either analysts haven't caught up, or the market is pricing in upside that may not materialize.
Prior Beliefs (Going In):
Updated Beliefs (After Analysis):
Verdict: No position.
DOCN fails my first filter on two critical criteria: sub-30% growth and sub-70% gross margins. While the AI pivot story is interesting and the execution has been credible, the numbers don't yet match the theory.
For this to become investable by my framework, I would need to see:
If the company delivers on its Q4 2026 exit rate of 25%+ and demonstrates stable-to-improving margins, I'd revisit. But I don't chase narratives ahead of results.
I could be wrong. If AI inference demand is as robust as management describes, and DigitalOcean's integrated platform genuinely differentiates vs. bare metal competitors, this could be a 30%+ grower at 40%+ EBITDA margins, and the stock would be cheap here. But my experience says these transformation stories have a higher failure rate than the market prices in.
The numbers have to match the theory. Right now, they're heading in the right direction but haven't arrived.
Bear