Thesis: DigitalOcean is a genuine growth re-acceleration story with textbook leading indicator divergence. At 8.2x run-rate P/S with a clear path from 18% to 30% revenue growth, this is the cheapest credible acceleration story in cloud. The debt is the only thing keeping me from making it a top-5 position.
This is where the story lives. Quarters down, years across:
| CY22 | CY23 | CY24 | CY25 | Trend | |
|---|---|---|---|---|---|
| Q1 (Mar) | -- | +1.3% | +2.1% | +2.8% | Accelerating |
| Q2 (Jun) | +5.2% | +2.8% | +4.2% | +3.8% | Stable |
| Q3 (Sep) | +13.6% | +4.3% | +3.1% | +5.0% | Reaccelerating |
| Q4 (Dec) | +7.2% | +2.1% | +3.2% | +5.6% | Accelerating |
Three of four quarters show clear reacceleration in CY25 vs CY24 and CY23. Q4 went from 2.1% to 3.2% to 5.6% -- that 5.6% QoQ annualises to 24.4%. This is not subtle.
| CY23 | CY24 | CY25 | Direction | |
|---|---|---|---|---|
| Q1 (Mar) | 29.7% | 11.8% | 14.1% | Turned up |
| Q2 (Jun) | 26.8% | 13.3% | 13.6% | Stable |
| Q3 (Sep) | 16.4% | 12.1% | 15.7% | Turned up |
| Q4 (Dec) | 11.0% | 13.3% | 18.3% | Accelerating |
The last 6 quarters of YoY growth: 13.3% -> 14.1% -> 13.6% -> 15.7% -> 18.3% -> guided 18-19% for Q1 FY26. That is a 500bps YoY acceleration in Q4 vs prior year. This is a company that bottomed at 11% YoY growth in Q4 CY23 and is now at 18% with management targeting 25%+ exit rate for CY26 and 30% for full-year CY27.
| CY22 | CY23 | CY24 | CY25 | |
|---|---|---|---|---|
| Q1 (Mar) | 127.3 | 165.1 | 184.7 | 210.7 |
| Q2 (Jun) | 133.9 | 169.8 | 192.5 | 218.7 |
| Q3 (Sep) | 152.1 | 177.1 | 198.5 | 229.6 |
| Q4 (Dec) | 163.0 | 180.9 | 204.9 | 242.4 |
| Full Year | 576.3 | 692.9 | 780.6 | 901.4 |
| FY YoY | -- | 20.2% | 12.7% | 15.5% |
FY25 at 901M, reaching * *1B annualised monthly revenue** in December. The acceleration from 12.7% to 15.5% FY growth continues to 21% in FY26 per guidance, then 30% in FY27.
This is textbook. Revenue at 18% YoY. Now look at the leading indicators:
| Indicator | Value | Growth | vs Revenue (18%) |
|---|---|---|---|
| RPO | $134M | +500% YoY, +121% QoQ | +482pp gap |
| $1M+ Customer ARR | $133M | +123% YoY | +105pp gap |
| AI Customer ARR | $120M | +150% YoY | +132pp gap |
| $500K+ Revenue | 17% of rev | +97% YoY | +79pp gap |
| $100K+ Revenue | 28% of rev | +58% YoY | +40pp gap |
| $100K+ Customers | 635 | +26% YoY | +8pp gap |
| Incremental Organic ARR | $51M (record) | Record high | -- |
| NDR | 101% | Up from 97% (Q3 CY24) | Inflected >100% |
Every single leading indicator is growing faster than aggregate revenue. RPO at +500% YoY vs revenue at +18% is the widest divergence I have seen since AXON in mid-2023 (where RPO was growing 61% vs revenue at 34%, and revenue subsequently accelerated to 40%+).
The NDR inflection is particularly important. From 96% in Q3-Q4 CY23 to 97% through CY24 to 99% to 100% to 101%. This was the company's Achilles heel for years -- customers were shrinking. Now they are expanding again. And the top cohort ($1M+) has 115% NDR.
| Tier | % of Revenue | YoY Growth | NDR |
|---|---|---|---|
| $1M+ Customers | 14% | +123% | 115% |
| $500K+ Customers | 17% | +97% | 106% |
| $100K+ Customers | 28% | +58% | -- |
| DNE (>$500/mo) ARR | $640M (66%) | +30% | 102% |
| Sub-$500/mo (implied) | ~34% | ~flat | <100% |
The growth engine is the top-end. Large customers are growing 3-7x faster than the company average. This is the classic pattern of an upmarket motion working -- the company is graduating customers into higher tiers while attracting larger new logos. The bottom third (sub-$500/mo) is essentially a legacy drag being diluted out.
AI Customer ARR = $120M, growing 150% YoY. But the critical detail: 70% comes from inference services + core cloud products, not bare metal GPU rental. Only 30% is bare metal.
This matters enormously. Bare metal GPU rental is a commodity -- any Neo Cloud can do it. But inference + core cloud is an integrated platform play with higher margins and higher switching costs.
CEO Srinivasan: "We do not just rent GPUs. We run production AI. We are not a GPU landlord. We are an AI cloud platform."
The economics prove it: DOCN generates $22M ARR per megawatt vs neo-clouds at $9-12M per MW. That 2x+ efficiency gap IS the platform premium.
| Metric | Current | 2026 Addition | 2027 Target |
|---|---|---|---|
| Data center capacity | Existing | +31 MW across 3 facilities | Full utilization |
| Revenue growth | 18% YoY | 21% FY / 25%+ exit | 30% FY |
| ARR per MW | $22M | Expected ~$20M (AI mix) | Improving with utilization |
Capacity timeline:
CFO Steinfort: "If all we did was fill those up, we would hit 30% for the full year in 2027."
Demand > Supply: CEO confirmed demand "far exceeds supply." This is supply-constrained growth, not demand-constrained.
| CY22 | CY23 | CY24 | CY25 | |
|---|---|---|---|---|
| Gross Margin % | ||||
| Q1 | 62.9% | 56.5% | 59.1% | 61.4% |
| Q2 | 64.3% | 60.3% | 61.0% | 59.9% |
| Q3 | 62.7% | 60.3% | 60.2% | 59.6% |
| Q4 | 63.1% | 58.9% | 61.5% | 58.7% |
| Op Margin % | ||||
| Q1 | -10.7% | -19.7% | 6.2% | 17.9% |
| Q2 | -5.9% | -0.9% | 11.6% | 16.3% |
| Q3 | 4.7% | 20.0% | 12.4% | 19.6% |
| Q4 | -7.0% | 6.1% | 15.9% | 16.0% |
Gross margin: ~60% and compressing slightly. AI workloads carry lower gross margins than core cloud. The 31MW capacity ramp will add depreciation before revenue, further pressuring GM in H1 2026. Known, temporary headwind.
Operating margin: Extraordinary improvement. From consistently negative in CY22 to solidly positive 12-20% range in CY24-25. GAAP operating income went from -26MinCY22to+157M in CY25.
| Metric | FY23 | FY24 | FY25 | FY26 Guide |
|---|---|---|---|---|
| Adj EBITDA ($M) | $278M | $328M | $375M | ~$400M (37% midpoint) |
| Adj EBITDA Margin | 40% | 42% | 42% | 36-38% |
| Non-GAAP EPS | $1.43 | $1.92 | $2.12 | 0.75−1.00 |
| Adj FCF ($M) | $156M | $135M | $168M | ~$174M (16% midpoint) |
| SBC % of Revenue | ~17% | 12% | 9% | Declining |
EBITDA margins guided down to 36-38% for FY26 due to capacity ramp timing. Physics problem, not a business problem. By 2027, targeting Rule of 50 (30% growth + 20% unlevered FCF margin).
SBC: 17% to 12% to 9% of revenue. EBITDA-less-SBC is 33% -- 80th percentile in software.
| Item | Amount | Notes |
|---|---|---|
| Cash | $254.5M | |
| 2026 Convertible Notes | $325M | Due Dec 2026 -- must repay |
| 2030 Convertible Notes | $606M | Issued 2025 |
| Credit Facility | $376M drawn | $800M total capacity |
| Total Debt | $1,296M | |
| Net Debt | $1,041M | |
| Net Leverage | 3.2x EBITDA | Expected >4x near-term |
| Finance Leases | $130.5M | New data center capacity |
This is the concern. $1.3B in debt for a $970M ARR company is material.
But manageable because:
Shares: ~105M diluted. Repurchased 34.9M shares ($1.6B) since IPO. Buybacks paused for growth investment.
Track record of underpromise/overdeliver:
Hit the 2027 target in Q4 2025 -- two full years early.
Introduced ARR-per-megawatt ($22M vs $9-12M neo-cloud) as competitive moat metric. Transparent about near-term margin pressure. "EBITDA less SBC at 80th percentile" framing shows sophistication.
Grade: A. CEO delivering transformational results. CFO providing clear visibility.
| Dimension | Neo Clouds (CoreWeave, Lambda) | DOCN |
|---|---|---|
| Primary service | Bare metal GPU rental | Full-stack cloud + inference |
| AI revenue mix | ~100% bare metal | 30% bare metal, 70% inference+cloud |
| ARR per megawatt | $9-12M | $22M |
| Profitability | Many pre-profit | 42% adj EBITDA margin |
| Target customer | Training-heavy, VC-funded | Post-PMF inference workloads |
OpenClaw validation: 30,000+ agents running with zero marketing spend. Organic pull = real product-market fit.
Open source tailwind: 30% of inference traffic on open source (OpenRouter data), at 90% lower cost/token. DOCN positioning as the platform for intelligent model routing.
| Metric | Value |
|---|---|
| Stock Price | $87.29 |
| Market Cap | ~$8.0B |
| Enterprise Value | ~$9.0B |
| Run-rate Revenue ($242.4M x 4) | $970M |
| Run-rate P/S | 8.2x |
| Run-rate EV/S | 9.3x |
| Run-rate Adj EBITDA ($99.3M x 4) | $397M |
| Run-rate EV/EBITDA | 22.7x |
| FY25 Adj FCF | $168M |
| P/FCF (trailing) | 47.6x |
| Non-GAAP EPS FY25 | ~$2.12 |
| P/E (Non-GAAP) | ~41x |
| FY26 Revenue Guide (midpoint) | $1,090M |
| Forward P/S (FY26) | 7.3x |
| Period | Growth | Margin | Rule of X |
|---|---|---|---|
| FY25 (actual) | 15% | 19% adj FCF | 34 |
| FY26 (guide) | 21% | 18% unlev FCF | 39 |
| FY27 (target) | 30% | 20% unlev FCF | 50 |
| Scenario | FY27 Rev | Multiple | Mkt Cap | Price | Return |
|---|---|---|---|---|---|
| Bull: 30% growth, 12x P/S | $1,417M | 12x | $17.0B | ~$153 | +75% |
| Base: 25% growth, 10x P/S | $1,363M | 10x | $13.6B | ~$122 | +40% |
| Bear: 20% growth, 7x P/S | $1,308M | 7x | $9.2B | ~$83 | -5% |
At 8.2x run-rate P/S with genuine acceleration from 18% to 30%, the risk/reward is asymmetric. Companies at 20-30% growth with 40%+ EBITDA margins typically trade at 8-12x NTM revenue. DOCN at 7.3x forward is below the peer median.
| Risk | Severity | Mitigant |
|---|---|---|
| Debt load ($1.3B, >4x near-term) | High | $310M OCF; no maturity post-2026 until 2030 |
| Capacity execution (31MW in 2026) | Medium | Committed contracts; realistic timeline |
| Gross margin dilution (AI mix) | Medium | EBITDA stable; lower COGS offset by lower opex |
| Competition (32 inference providers) | Medium | $22M ARR/MW vs $9-12M; 70% non-bare-metal |
| Customer concentration | Low | Top 25 = ~10% of revenue |
| Stock near 52-week high | Low | Fundamental support from acceleration |
DOCN is the cheapest credible growth re-acceleration story I can find in cloud infrastructure right now.
Position: Initiating at 3-5%. Add on pullback below $75 or continued execution in Q1 FY26. Deploy conference April 28 could be a near-term catalyst.
Triggers:
First analysis on DOCN. The leading indicator divergence reminds me of AXON in mid-2023 -- RPO and customer cohort growth massively outpacing revenue, with management systematically underpromising. In AXON's case, revenue accelerated from 34% to 40%+. The data says DOCN follows the same pattern.
-wsm
(No position in DOCN -- initiating)