Executive Summary

DigitalOcean is at $68.16, up 53% in a year and 29% in the past month. It's trading at 95% of its 52-week range, 14% above mean analyst targets, with 17.4% short interest — and reports Q4 earnings on February 24.

The thesis is simple: DOCN is a real cloud infrastructure company executing an AI pivot that is accelerating revenue growth from 8% to 16% YoY over four quarters. AI revenue is doubling. The highest-spending customers are growing 72% YoY. They pulled forward their 2027 growth target by a full year.

But at $68, the stock is above the capped call cap price on its $625M convertible notes. Every dollar higher means more dilution. Buybacks collapsed from $500M/year to $100M. The CPTO who architected the AI pivot just resigned. And NDR is still below 100%.

The asymmetry question is answerable: at this price, risk/reward is compressed. The stock is pricing in the bull case. A pullback to $55-58 would create the asymmetric entry the thesis deserves.

The Business: What's Actually Happening

DigitalOcean runs cloud infrastructure for developers and SMBs. Think AWS for people who don't want to hire a DevOps team. 640,000+ customers, average cohort retention of 4.5 years, $919M ARR as of Q3 2025.

The company is executing on three growth vectors simultaneously:

1. Revenue acceleration. Quarterly revenue went from $178M (Q1 2024, +8% YoY) to $230M (Q3 2025, +16% YoY). Q3 delivered $44M of incremental organic ARR — the highest in company history. FY2025 guided at $896-897M (+15%). Management expects 18-20% growth in 2026, pulled forward from the original 2027 target.

2. Upmarket expansion. Customers spending $100K+/year grew revenue 41% YoY and now represent 26% of total revenue. $1M+ ARR customers grew 72% to $110M ARR. The company signed its first-ever 8-figure committed contracts in Q3-Q4. These are not toy accounts — Fal (generative AI for Canva, Shopify, Perplexity), NewsBreak (40M MAU), Bright Data, VPN Super.

3. AI infrastructure. This is the swing factor. AI revenue "more than doubled YoY" for five consecutive quarters. CFO Matt Steinfort said on the Q3 call that AI "will get into the mid-teens, maybe even high teens as a percent of revenue" — a forward-looking projection, not a current disclosure. Based on the Q4 2024 anchor of "3 points of overall growth contribution," current AI revenue is likely $80-120M annualized. Over 19,000 agents built on the platform, 7,000+ in production. The company offers GPU droplets (bare metal and virtual), inference optimization (flash attention, FP8 quantization, speculative decoding), and a GenAI platform for deploying agents.

Importantly, AI revenue is not yet included in the NDR calculation. NDR at 99% reflects only core cloud. The AI layer is entirely additive to reported metrics — which means actual underlying expansion in the installed base is better than reported.

Source: DOCN Q3 2025 earnings call (Nov 5, 2025), 10-Q filed Nov 5, 2025.

The Numbers

Profitability is solid. Q3 2025: 60% gross margin (+100bps YoY), 43% adj. EBITDA margin, 37% unlevered adj. FCF margin. Operating income for 9M 2025 was $118M, tracking to beat the $91M full-year 2024 figure before Q4 even reports. Non-GAAP EPS of $2.05 guided for FY2025.

FCF has improved dramatically. Trailing 12-month adjusted FCF margin reached 21%, up from 17% in FY2024. Part of this is equipment financing timing ($28M moved off-balance-sheet via lease in Q3), but operating cash flow generation is genuinely strengthening — $252M for 9M 2025 vs. $211M prior year. One caveat: new operating right-of-use assets jumped to $161M for 9M 2025 vs. $28M prior year — a 5.8x increase in lease commitments for the datacenter expansion. These costs haven't fully flowed through yet. Expect 2026 COGS to step up materially as these leases commence ahead of revenue ramp.

Source: DOCN 10-Q filed Nov 5, 2025.

Capital Structure: Where the Complexity Lives

This is not a clean balance sheet company anymore.

Net debt/adj. EBITDA is ≈2.7x. CFO guides mid-3x by end of 2026 as new datacenter capacity comes online with upfront costs before revenue ramps.

The debt has real cost now. Interest expense jumped from $6.9M (9M 2024) to $9.5M (9M 2025) and will run higher in 2026 with the full-year impact of the $380M term loan at ≈6.1%. That's ≈$23M in annual interest expense — not catastrophic, but it's a new line item for a company that previously had zero-coupon converts and minimal interest cost.

The Convertible Cap Problem

The 2030 convertible notes convert at $39.17/share. The company bought capped calls to offset dilution up to $66.51/share. The stock is currently at $68.16.

The stock is above the cap. Here's what that means:

• Below $39.17: No dilution. Notes repaid in cash.
• $39.17 to $66.51: Capped calls fully offset conversion dilution. No net impact.
• Above $66.51: Dilution flows through. The capped calls cap out and the company is exposed to incremental share creation on every dollar above $66.51.

At today's price of $68.16, the net dilution is modest — roughly 400K incremental shares above what the capped calls cover. But the math gets worse as the stock rises:

This creates a perverse dynamic: the higher the stock goes, the more dilutive the convert becomes. It's a drag on EPS expansion that gets worse as the thesis plays out. For a stock that's already pricing in acceleration, this is a headwind consensus may underappreciate.

Buyback Collapse

The capital allocation story has shifted dramatically:

At $68/share, the $100M authorization buys back ≈1.5M shares — barely offsetting annual SBC dilution of ≈3M shares. The aggressive share count reduction that took DOCN from ≈130M to ≈91.5M diluted shares is over. Capital is now directed at datacenter expansion and debt management.

Source: DOCN 10-Q filed Nov 5, 2025, Note 6 (Debt) and Note 8 (Stockholders' Equity).

AI Revenue: The Critical Unknown

Management refuses to disclose exact AI revenue. This is either prudent (AI is commingled with cloud services and hard to separate) or strategic (the number is smaller than the narrative implies).

Here's what we can infer:

CFO on Q3 call: AI "will get into the mid-teens, maybe even high teens as a percent of revenue." Note the tense — this is forward-looking, describing trajectory, not a disclosure of current AI revenue. The anchoring data point is from Q4 2024: management said AI/ML contributed "3 points of overall growth" in FY2024, implying roughly $24M of incremental AI revenue on ≈$800M in total revenue. AI ARR as of Q4 2024 was growing 160%+ YoY. Working from that base, current AI revenue is likely $80-120M annualized — meaningful but lower than the $130-160M figure you'd get from reading the CFO's quote as a current disclosure.

If AI doubles again YoY in 2026 (as it has each quarter since platform launch), that's $160-240M — roughly 15-22% of projected ≈$1.06B FY2026 revenue. This would be the growth engine that justifies acceleration from 16% to 20%+. But the range is wide because management won't quantify the base.

But there's a margin problem. AI infrastructure is GPU-at-cost-plus. CFO flagged "modest headwind to gross margins" from AI mix in 2026. If AI goes from ≈15% to ≈25% of revenue with lower margins, gross margin could drop from 60% to 57-58%. That's not fatal, but it compresses the operating leverage story that justified the multiple expansion from $25 to $68.

The other question: is AI revenue durable? Management says workloads are "predominantly inferencing, not training" — which is good (production workloads are stickier than experimentation). But inference is also more commoditized. GPU providers are abundant. DOCN's differentiation has to come from the software stack (Gradient AI platform, agent lifecycle management), not the raw compute.

Source: DOCN Q3 2025 earnings call, Q4 2024 earnings call.

Management Risk: The CPTO Departure

Bratin Saha, Chief Product and Technology Officer, resigned November 20, 2025. The 8-K says it was "not the result of any disagreement." The company reaffirmed guidance the same day.

This matters. Saha was referenced extensively in earnings calls as the architect of the AI strategy and product roadmap. The CFO repeatedly credited "Bratin and his team" for product execution. Losing the CPTO during the most critical phase of the AI infrastructure buildout is a real risk.

Worth noting: Saha sold 13,010 shares ($415K) on September 3, 2025 — just 78 days before his resignation. This doesn't prove anything on its own (executives sell for many reasons), but the timing is notable context alongside the boilerplate 8-K language. No other insider has made discretionary purchases — all recent insider transactions are automatic director equity awards.

No replacement has been announced as of the filing date. The Feb 24 earnings call should address this. If the replacement is a strong hire from a major cloud/AI company, the market will move on. If there's an interim arrangement or the role stays vacant, it signals deeper challenges.

Source: DOCN 8-K filed Nov 26, 2025; Form 4 filings.

NDR: The Lagging Indicator

Net dollar retention has improved from 96% to 99% over four quarters but remains below 100%. This means the existing customer base — measured by the metric — is still contracting on a dollar-weighted basis.

Management's explanation is reasonable: a long tail of small customers ($10-15/month) with naturally declining spend weighs down the blended NDR, while larger customers (Scalers+) are expanding meaningfully. AI customers aren't included yet.

But reported NDR below 100% matters for sentiment and peer comparisons. Cloud investors screen for NDR > 100% as a quality signal. Until DOCN crosses that threshold — likely when AI customers are incorporated into the calculation, possibly in 2026 — the stock will carry this discount.

Competitive Position: The Sandwich

DOCN occupies an awkward middle. From above, AWS/Azure/GCP simplified offerings (Lightsail, App Service) increasingly target the same developer SMB audience. From below, Hetzner and Vultr offer significantly cheaper raw compute. From the side, Linode (Akamai) has developer-friendly UX backed by infrastructure scale.

DOCN's differentiation is real but narrow:

• Simplicity of developer experience — genuinely easier than AWS
• Community and educational content — 640K+ customers, organic acquisition engine
• Twin-stack positioning — cloud + AI on one platform without hyperscaler complexity
• Average cohort retention of 4.5 years — sticky despite low switching costs

No competitor mentions DigitalOcean in their earnings calls. This is instructive: DOCN is below the competitive radar of hyperscalers. The threat isn't competitive attack — it's competitive irrelevance if hyperscalers make their own products simple enough.

The "agentic cloud" positioning is the strategic bet to escape the sandwich. If DOCN becomes the default platform for SMBs deploying AI agents — the way Heroku was for web apps — it creates a genuine moat through ecosystem lock-in. But this is early. 19,000 agents built, 7,000 in production. Interesting traction, not yet proven durability.

Factor Decomposition: What's Driving Returns

Before valuation, a reality check on what's actually moving this stock. DOCN's factor profile:

Idiosyncratic variance is 65% — below the 75% threshold where we'd say returns are driven primarily by stock-specific factors. Nearly a quarter of DOCN's price movement is riding the tech sector. The 53% 1Y return and 29% 1M surge include meaningful factor contributions. Beta to SPX is 1.77 (high sensitivity).

This doesn't invalidate the thesis, but it means any entry should account for the reality that DOCN is partly a levered tech sector bet. If XLK corrects, DOCN likely falls harder than idiosyncratic fundamentals warrant — which, incidentally, is what creates the $55-58 entry opportunity.

Valuation: Cheap Relative, Fair Absolute

AKAM (Akamai, which owns Linode — a direct DOCN competitor) is the most structurally comparable peer: similar gross margins, IaaS business model, compute platform with cloud ambitions. AKAM trades at ≈4x revenue with 5% growth — showing what the market pays for a mature IaaS business without the growth acceleration narrative. DOCN's premium to AKAM is earned by the 16%→20% growth trajectory, but the gap to NET/DDOG reflects structurally lower margins and narrower moat.

DOCN trades at a steep discount to SaaS cloud peers on EV/Revenue. That discount is earned: lower gross margins, slower growth, narrower moat, higher debt. But the discount has been compressing rapidly as the growth acceleration story gains traction — 8x revenue for a company accelerating from 13% to 18-20% growth is not expensive.

The 17.4% short interest deserves attention. It's 5-6x the short interest of cloud peers like NET (3.3%) and DDOG (2.9%). AKAM, also an IaaS company under competitive pressure, carries 11.1%. At 3.6 days to cover, the short position is not trivially sized. The stock is trading 14% above the mean analyst price target of $59.58 and above 10 of 14 analyst targets. Someone is paying to be short here — likely the combination of elevated valuation vs. fundamentals, convert dilution above the cap, and the growth deceleration risk if AI revenue doesn't sustain doubling.

The problem is timing. At $68, the stock is pricing in the 18-20% acceleration. If the Feb 24 call delivers exactly what was previewed, the reaction may be "sell the news." To push higher, DOCN needs to exceed the 20% threshold or deliver an AI revenue disclosure that reframes the narrative.

Asymmetry Assessment

Scenario Analysis

Expected value of ≈$67 against a $68 stock price. Asymmetry is essentially zero at current levels. The bull case requires multiple things to go right (20%+ guide, AI disclosure, CPTO replacement, short squeeze). The bear case only needs one thing to go wrong (guidance miss, margin compression, or macro deterioration).

Where Asymmetry Improves

At $55-58, the math shifts:

A pullback to $55-58 — possible on an in-line quarter that disappoints elevated expectations — creates the entry point where asymmetry favors longs.

What to Watch on February 24

Seven items will determine the next leg:

1. 2026 revenue guidance range. 18-20% confirmed, or tighter? The street needs specifics.
2. AI revenue disclosure. Will management finally quantify AI? If AI is $40M+/quarter, the stock reprices.
3. CPTO successor. Named replacement or interim? Quality matters more than speed.
4. RPO inflection. Did the 8-figure committed contracts materially increase RPO from $47M?
5. Gross margin guidance. AI mix pressure — how much? 200bps compression is priced; 400bps is not.
6. NDR trajectory. Crossing 100% would remove a sentiment overhang.
7. 2026 converts. How will they handle the $312M remaining balance maturing Dec 2026?

Verdict

DigitalOcean is a legitimate business executing an AI pivot with real infrastructure, real customers, and accelerating growth. The agentic cloud positioning is directionally correct — inference workloads for AI-native companies is a genuine demand vector that hyperscalers serve poorly at the SMB/mid-market level.

But at $68, the stock is pricing in the bull case. The convertible dilution above $66.51 creates an escalating headwind. The CPTO departure introduces execution risk at the worst possible moment. Buybacks can no longer support the share count. And the Feb 24 catalyst is 10 days away with expectations elevated.

This is not a name to chase. The underlying business is strong enough that it will create entry points. Post-earnings weakness on in-line results, a macro-driven selloff, or a quarterly miss in 2026 as DC expansion costs ramp ahead of revenue — any of these would create the asymmetric setup at $55-58 where the risk/reward genuinely favors new longs.

For existing holders: hold through earnings. The 18-20% growth guide is real, the AI revenue stream is durable, and the upmarket motion is working. There's no thesis-invalidating risk here — just unfavorable entry-point math.

Epistemic state: Noisy signal. The Feb 24 call resolves multiple open gaps simultaneously (AI quantification, CPTO replacement, 2026 guidance specifics). Post-call, this should collapse to either a clear bull or a clear pass. Size for uncertainty until then.