Built by Claude Fable

DigitalOcean Holdings: 4.0/10 | $98.72

Deep dive pages: Financial Trends | Thematic Exposure | Management Quality | Investor Sentiment | Concerns, Catalysts & Risks

Prepared 2026-10-16. Price and market cap ($11.61B) are the FMP /stable/quote print for 2026-10-16 (fmp_quote.json). The Bloomberg MCP was not connected, so FMP is the fallback, and no figure was assumed. EV is $12.34B (FMP key-metrics-ttm). Fiscal year = calendar year. Last print was FY2026Q2 (2026-08-04); the next is FY2026Q3 on 2026-11-04.

Data source flag (read first): Daloopa was not connected in this session (claude.ai connector OAuth not authorized), so this report has no daloopa.com/src citations. Phase 3 could not fetch Daloopa to fill gaps either. Fundamentals are SEC-derived FMP /stable statements, cited [FMP-IS], [FMP-CF], [FMP-BS] (raw files in dim1_data/). Company KPIs come from the 8 FMP earnings call transcripts in transcripts/ ([T-FYyyyyQn]). Consensus is FMP analyst-estimates, because Visible Alpha and Bloomberg were also unavailable. Before this score is published, re-tie every figure to Daloopa once Brooker re-authorizes the connector in claude.ai connector settings.

Bottom line: DOCN fails all three quality gates (no oligopoly position, FCF that is not positive and growing, and a management track record gate scored NO), so the composite is capped at 4.0/10: Speculative / Does Not Meet Quality Bar (AVOID). The raw weighted composite is 4.30. Revenue growth is accelerating sharply (12% in Q1'24, 29% in Q2'26, guided 32% to 34% for Q3'26 and "50% plus" for FY27) on AI inference demand, and management has beaten its revenue guide 8 quarters in a row. But DOCN has under 1% share in every market it competes in, and is a price taker against the AWS/Azure/GCP oligopoly and a CoreWeave 10x its size. The growth is also being bought: GAAP gross margin is down about 490 bps YoY, GAAP operating income has declined YoY for 2 quarters, TTM FCF is about $15M (1.5% margin), diluted shares are up 25.8% YoY, and equipment finance leases total $578M. This is a capacity-led growth story, not a quality compounder.

Quality gate

Gate result: all three NO. Maximum composite 4.0/10. Flag: "Speculative / Does Not Meet Quality Bar."

Gate tension, surfaced rather than overridden: the management NO sits next to an 8/10 management score. If that gate were flipped to YES, DOCN would have two NOs, a 5.5 cap, and an uncapped 4.30 composite. That changes the score by only 0.3, and the conclusion (below the quality bar) does not change.

Gate flag: Speculative / Does Not Meet Quality Bar. Recommendation: AVOID.

What would change the view:

• FCF gate: the 2027 capacity (155 MW committed) delivers the "50% plus" growth at about 40% adj. EBITDA with "high-teens" adj. FCF margins, and GAAP FCF after lease principal inflects positive and grows for 2 or more quarters [T-FY2026Q1], [T-FY2026Q2].
• Per-share discipline: diluted share growth falls back below 10% YoY and sequential debt growth (equipment financing) stops outpacing revenue.
• Thematic gate: this one is structural and unlikely to flip. DOCN would need a defensible leading share in a definable sub-market (for example, managed inference for AI-native software companies), backed by sourced share data.
• Nearest test: the FY2026Q3 print on 2026-11-04. Guide is $304M to $307M (+32% to 34%) and adj. EBITDA of 38% to 39%; watch sequential lease and debt growth and FCF after principal.

Open gaps for follow-up: Daloopa re-tie of all fundamentals; Bloomberg / Visible Alpha consensus and revision trends; FY2025 10-K geographic split; reconciling FMP FY27 EBITDA ($437.7M, ~24% margin) with management's ~40% adj. EBITDA framework; Priority 1 internal sources (SharePoint, Excel models, OneNote, Outlook) not available this run.

Reviewed 2026-10-16 against pm-review and investing-principles.

Verdict: the AVOID call stands, but the 4.0 is built on a shaky gate call and several unreconciled numbers. A cleaner score is about 4.3 to 4.6 (management gate flipped to YES, dilution penalty re-tiered). That is still below the quality bar, so the recommendation does not change. What matters more is that the report hides its most important finding: core (non-AI) cloud is growing about 11%, not "mid-teens." All of the acceleration is AI. That should be the headline.

Derivation: AI customer ARR was $234M at Q2'26, +212% YoY, so about $75M a year ago. Total ARR was $875M at Q2'25, so core was about $800M. At Q2'26, core is $1,125M minus $234M, or about $891M. That is +11% YoY core growth. AI added about $64M of the $93M record incremental ARR (69%). The "accelerating" story is really an AI capacity build sitting on a decelerating, commodity core.

1. Gaps to the pm-review

• Price talk throughout, which pm-review explicitly bans. Dim 4 leans on the stock price: a 14.8% drop on Oct 14, "up ~336%," "127% YTD," the 52-week range, "~78% implied upside to avg PT," insider sale prices, and "shares fell after the Q2 beat." Part of the 4d sub-score (4/10) comes from price momentum. Remove it, and rebuild 4b/4d on estimate revisions and ratings changes instead.
• YoY is missing from several tables.
  • The KPI table shows ARR levels, incremental ARR and adj. FCF with no YoY growth column. ARR YoY (+29%) and incremental ARR YoY ($93M vs $32M, +191%) are the most important growth rates in the report, and neither is shown.
  • The annual table gives YoY only for revenue. It needs YoY for FCF, CFO, capex and op income.
  • The segment table has no YoY for core cloud. "Mid-teens" is unsourced and, per the math above, likely wrong.
• Margin trend is not shown as bps change YoY by quarter for adj. EBITDA and FCF margin. The table shows levels and labels the trend "flat to down." Show the bps: Q2'26 adj. EBITDA 40% vs 41% is -100 bps, and Q1'26 is flat.
• The Adj. Op Margin proxy column misleads. GAAP OI plus SBC gives 30.1% for Q2'26 vs 24% company-reported. The analyst admits it is wrong and leaves it in anyway. Drop the column, or use company figures only and mark the earlier quarters n/a.
• Sentiment inversion is never addressed. Is this worse to better? Growth is worse to better (12% to 29%). Margins, FCF and per-share are better to worse. Consensus is already enthusiastic: 68% Buy, FY27 street revenue +54% above management's 50%+ floor. Under our framework, consensus at or above the guide on the headline driver means the easy money on the growth inflection is gone. Say that explicitly.
• Management contrarianism is identified (ARR per MW / software attach) but not quantified. What ARR/MW does consensus imply vs management? $1,820M FY27 revenue over 155 MW gives a testable implied number. Without it, "real, specific gap" is an assertion.
• Street consensus is missing on key metrics beyond revenue and EPS. There is no consensus for gross margin, adj. EBITDA (the FMP 24% vs mgmt 40% conflict is flagged but not resolved), FCF, or capex. There are no revision trends at all.
• Explanation of why margins compress is qualitative only. It needs a bridge: how much of the 490 bps GM decline is AI mix vs startup cost vs depreciation on new MW? Management gave about $100M of startup costs, so allocate it by quarter.

2. Gaps vs Known Skills (investing-principles)

• The management quality gate NO is not supported by our own principles. The principles define great teams by guidance accuracy, promise delivery and C-suite stability. Dim 3 shows 8/8 revenue beats, a 90% hit rate, a stable CFO since Jan 2023 and a CEO since Feb 2024. The "3+ year pairing" bar is not in investing-principles. The red-flag table itself marks "CEO/CFO change in last 2 years: No." Gate it YES and record the FCF reset and insider selling as watch items. Failing the gate here is a silent override of the analyst's own evidence.
• The promise-tracking rubric was violated. The rule is promises made 4+ quarters ago, scored against actuals. Three rows are Q3'25 promises (3 quarters old) that are scored on "tracking": the 2027 18% to 20% target scored against FY26, the FY26 EBITDA guide, and the FY26 FCF guide. "Tracking" is not a result. Strip them out and the clean rate is 16 of 17 (94%), which strengthens the gate YES case.
• NDR is contradictory across dimensions.
  • Dim 3 scores "NDR above 100% in 2025" as a Miss (held at 99%).
  • Dim 2 cites ~102% D&E NDR from the FY2025Q4 call.
  • Dim 5 says NDR was at a 3-year high of 102% when it was retired.
  • The KPI table shows Q4'25 NDR as n/a. If Q4'25 was 102%, the promise was a Hit. Resolve this with the transcript.
• The red flag "increasing revenue with declining FCF" is underweighted. Annual FCF fell from $110M (FY23) to $96M (FY24) to $41M (FY25) while revenue grew every year. That is a textbook Historical Pattern red flag. Dim 3 marks it "No (watch)" by testing quarterly noise instead of the annual trend. Flag it as Yes on the annual series, and make Dim 1 and Dim 3 agree.
• The "rising debt without rationale" test uses untrustworthy data.
  • The FMP total-debt series is inconsistent with the capital structure the report describes. Q4'25 total debt of $731M cannot hold $625M of 2030 converts plus a $380M Term Loan A plus finance leases.
  • Q2'26 shows only $609M of LTD even though the $472M equitization happened in July, after quarter end.
  • "Most 2026 notes repurchased" sits next to $312M of 2026 converts still outstanding. The debt penalty decision and the leverage narrative rest on these numbers. Tie them to the 10-Q.
• The dilution penalty may be an if-converted artifact. GAAP diluted shares jump around: 102.6M in Q3'24, then 94.4M in Q4'24, then 102.3M in Q1'25. That pattern is consistent with convert shares going in and out as anti-dilutive. Basic shares are +14.8%, which is the over-10% tier (-1), not -2. Use basic shares plus the economic dilution from the $888M raise, and show the math. That moves Dim 1 from 2 to 3.
• The valuation framework is incomplete. P/FCF (next year) is a required forward multiple and is missing. Given that FCF is the thesis breaker, it is the single most important multiple here. Show it on GAAP FCF, on company adj. FCF, and after lease principal. EV/EBITDA is also mixed-basis across peers (FMP consensus vs adjusted), so the "in line at 17x" conclusion is not like-for-like.
• "Leaders remain leaders" is applied to the wrong market. Failing DOCN against global hyperscaler share is right but lazy. The real question is whether DOCN leads the developer/SMB alt-cloud niche (vs Akamai/Linode, Vultr, Hetzner, OVH). If it holds 30%+ there, the thematic gate case changes. If it does not, the AVOID is airtight. Either way the report needs sourced niche share. The current share sources (tech-insider.org, shattered.io, Mordor) are weak secondary web sources.
• Data hierarchy. No Daloopa, Bloomberg, Visible Alpha or Priority 1 internal sources were used. That is disclosed, which is good. But under our "never fabricate market data" rule, price, market cap and EV must come from Bloomberg before publishing. FMP is a fallback, not a substitute. Also verify the implied share count: $11.61B / $98.72 is 117.6M shares, which does not match the 104.6M basic or 126.5M diluted. Price and EV cascade into every multiple.
• Internal inconsistencies left in the report.
  • FY26 adj. FCF guide is "9% to 12%" in Dim 1 but "11% to 13%" (latest) in Dims 3 and 5.
  • Committed capacity is "135 MW" in Dim 1 but "155 MW" in Dims 5 and composite.
  • Insider selling is reconciled to Form 4s in Dim 3 (~$17M officers plus $557M Access) but restated as unverified "GuruFocus ~$574M, worth checking" in Dim 5.
  • The revenue beat streak is "8 quarters" in places and "last 4" in others, with different bases (guide vs consensus) that are not labeled.

3. Action Items

1. Rewrite the bottom line around core vs AI. Add a table with 8 quarters of core cloud ARR vs AI ARR, each with YoY, plus each one's share of incremental ARR. The investable question is whether an 11%-growth commodity core plus a capital-hungry AI layer earns its cost of capital.
2. Flip the management gate to YES and re-run the cap (5.5). Fix promise tracking to 4+ quarter-old promises only. Resolve the NDR contradiction from the FY2025Q4 transcript.
3. Re-tier the dilution penalty using basic shares and the raise math. Re-score Dim 1 and show the arithmetic from initial score to final. The "10-grade plus three 1-grades equals 5" step is opaque.
4. Compute FCF after finance-lease principal from the CF statement for all 10 quarters and 5 years, with YoY. Add next-year P/FCF on all three FCF bases. This is the number that decides the FCF gate.
5. Build unit economics: capex per MW, ARR per MW (core vs AI), payback, and incremental ROIC on the 155 MW. Back out consensus-implied FY27 ARR/MW and compare it to management's "increasing" claim. That quantifies the one contrarian thesis.
6. Split AI growth into price and volume. About 30% GPU list-price increases are in the FY26 guide, so how much of the acceleration is pricing that reverses when supply loosens?
7. Tie the debt stack to the Q2'26 10-Q: converts by series, TLA, finance leases, and post-July pro forma. Re-test the debt penalty and the leverage claims.
8. Strip all price and stock-move content from Dim 4. Replace it with consensus revision history (Bloomberg/VA: FY26/FY27 revenue, EBITDA and EPS revisions over 90 and 180 days) and re-score.
9. Source niche market share for developer/SMB alt-cloud and managed inference from broker research or AlphaSense, not blog-tier web pages.
10. Re-tie before publishing: Daloopa for fundamentals (with src links), Bloomberg for price, market cap, EV and consensus, and the FY2025 10-K geographic split. Fix the 9%-12% vs 11%-13% FCF guide, the 135 vs 155 MW, and the insider-selling inconsistencies.
11. Add a sentiment-inversion paragraph: growth is worse to better, while cash and per-share are better to worse, and consensus already sits above guide. State plainly why that is not an inversion setup.

Score sensitivity after fixes: Dim 1 goes from 2 to 3 (+0.25) and the management gate goes to YES (cap 5.5). Composite is about 4.55. Still AVOID. The thesis does not hinge on these fixes, but the report should not reach the right answer with the wrong gate.