Datadog
Usage-based + subscription SaaS; land-and-expand across 30+ products on one platform; gross-margin ~80%, scales with cloud + AI workloads
The thesis on this name
State of AI Compute
Datadog is the leading cloud-native observability and security SaaS platform — unified metrics, logs, traces, APM, and increasingly security + AI-workload monitoring on a single usage-based platform.
State of Enterprise AI SaaS
Pure consumption-priced observability that STRUCTURALLY BENEFITS from the same AI movement threatening seat-based peers: agents, AI apps and inference generate more logs/traces/metrics, not fewer, so Datadog's bill scales with machine workloads rather than headcount. Q1'26 rev +32% accelerating (first $1B quarter), 4,550 $100k+ customers up from 3,770, AI-native cohort a real secular driver (Datadog Q1'26 release, May 2026). The platform land-and-expand (now 25+ products) is the canonical durable-compounder shape — high NRR, no per-seat cannibalization vector.
State of Enterprise AI SaaS
The cleanest consumption-model winner — AI workloads are a direct usage tailwind (more agents, more telemetry), revenue reaccelerated to 32% and AI-native customers now >10% of revenue growing >100%.
State of Enterprise AI SaaS
Purest 'AI grows my bill, not my seat-risk' name; +32% accelerating. Started at 7%, staged +2% add on sub-$200 weakness.
Earnings, margins, COGS & capex
Asset-light, ~80%-gross-margin usage-based SaaS that re-accelerated to 32% growth in Q1 FY26 on AI-workload and enterprise demand. ARR crossed $4B; >27% FCF margin. The model converts cloud + AI-inference consumption into recurring observability spend with strong incremental margins.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~20¢ is cost of goods and ~58¢ operating expense, leaving ~22¢ of operating profit.
Revenue trend
Margins
Stable, high
Stable; FY26 guide 22-23%
Strong, durable
COGS structure
Primarily public-cloud hosting (AWS/GCP/Azure) compute + storage to ingest and process customer telemetry, plus support — the ~20% COGS; scales with ingested data volume, partly offset by efficiency engineering.
Capex
Minimal — asset-light, runs on rented public cloud rather than owned data centers; capex ~3-4% of revenue. The cloud-infra bill sits in COGS, not capex.
Latest earnings
Beat — rev $1.01B vs ~$960M est; non-GAAP EPS $0.60 vs ~$0.51 est (~18% beat); raised FY26 guide
Q2 FY26: $1.07-1.08B (+29-31%), non-GAAP op income $225-235M. FY26: $4.30-4.34B (+25-27%), non-GAAP op income $940-980M (22-23% margin), non-GAAP EPS $2.36-2.44.
- Revenue
- $1.006B (+32% YoY)
- Total customers
- ~33,200 (vs ~30,500 YoY)
- $100k+ ARR customers
- 4,550 (~90% of ARR)
- Net revenue retention
- low-120s% (up QoQ)
- Free cash flow
- $289M (~29% margin)
- Cash + securities
- $4.8B
Growth drivers
- AI-native + AI-integration adoption (20% of customers / 80% of ARR now use AI integrations; AI-native cohort still a meaningful but lumpier driver)
- Multi-product land-and-expand — ~83% of customers use 2+ products, security + data products as new vectors
- Enterprise + large-deal momentum (4,550 customers at $100k+ ARR, ~90% of ARR; record new-logo bookings, more than doubled YoY)
- Cloud migration tailwind + OpenTelemetry standardization (double-edged)
- Net revenue retention back to low-120s%, up sequentially
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-02-18. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
The cleanest public-market pure-play on AI + cloud observability: re-accelerating to 32% growth at ~$4B ARR with 80% gross / 27% FCF margins, a net-cash balance sheet, and a multi-product expansion engine that turns every new cloud and AI workload into recurring telemetry spend.
- Growth re-accelerated to 32% with raised guidance and record new-logo bookings — the demand pause of 2023 is behind it
- AI is a structural tailwind: more workloads, agents, and inference = more telemetry to monitor; AI-integration usage now spans 20% of customers / 80% of ARR
- Land-and-expand + new categories (security, data, AI monitoring) extend TAM well beyond core APM
- Rule-of-50+ profile (32% growth + 27% FCF margin) with $4.8B net cash funds buybacks/M&A and downside protection
- Embedded, sticky platform with low-120s% NRR — expansion compounds inside the base
A ~24x-trailing-sales premium that prices in years of 25%+ growth and margin expansion, against a usage-based model exposed to cloud-cost optimization, hyperscaler bundling, OpenTelemetry commoditization, and lumpy AI-native concentration — any growth wobble re-rates the stock hard.
- Valuation: ~24x trailing / ~19x forward sales (P/S ~21x per Simply Wall St) leaves no margin for error vs peers at single-digit P/S
- Usage-based revenue is two-sided — the same cloud spend that drives growth can be optimized away in a downturn (2023 precedent)
- Hyperscalers (AWS/Azure/GCP) and Cisco-Splunk can bundle 'good-enough' observability and undercut on price
- OpenTelemetry lowers switching costs over time — customers can swap the backend without re-instrumenting
- AI-native cohort concentration (incl. largest customer) is lumpy; high-single-digit YoY in Q1 FY26 shows the risk
- High SBC + dilution; GAAP profitability much thinner than the non-GAAP headline
What it is worth
Comps (EV/Sales, P/S vs observability peers) + reverse-DCF sanity check on the implied growth/margin.
~$140-175
growth decelerates to high-teens on cloud-cost optimization + hyperscaler/OTel pressure, AI-native cohort stalls; multiple compresses to ~10-12x sales. (~$50-62B)
~$230-260
25-27% growth per guide, NRR low-120s%, FCF ~28-30%; multiple gradually de-rates as growth normalizes. (~$80-92B)
~$300-340
sustained 30%+ growth on AI-workload + security attach, NRR holding 120s%, FCF margin to ~32%; multiple holds at premium. (~$110-120B mkt cap)
At ~$82.6B market cap on ~$3.43B FY25 revenue, DDOG trades ~24x trailing / ~19x forward (FY26 guide ~$4.32B) sales — a steep premium to Dynatrace (~mid-single-to-high-single-digit P/S) and the software median (~3x). The price embeds ~mid-20s% revenue CAGR for ~5 years AND continued FCF-margin expansion toward ~30%+. That is achievable given the AI/cloud tailwind and re-acceleration, but it is fully priced — the reverse-DCF leaves no cushion for a growth stumble or a cloud-optimization cycle.
SWOT
Strengths
- Category-leading unified platform — single agent/pane across metrics, logs, traces, APM, security; high switching cost once embedded
- Best-in-class SaaS financials: ~80% gross margin, 27% FCF margin, net-cash balance sheet, GAAP-positive
- Multi-product land-and-expand engine; >83% of customers on 2+ products; new-product attach drives expansion
- Direct beneficiary of cloud + AI-workload growth — observability spend scales with compute consumption
Weaknesses
- Usage-based model exposes revenue to customer cloud-cost optimization (the 2023 'optimization' demand shock)
- Concentration in AI-native cohort — its largest customer sits here; this cohort's growth is lumpy (only high-single-digit YoY in Q1 FY26)
- Heavy stock-based compensation inflates non-GAAP vs GAAP; dilution overhang
- Premium valuation leaves no margin for a growth stumble
Opportunities
- AI observability / LLM-monitoring as a new product category (monitor inference, agents, model performance, cost)
- Security (Cloud SIEM, CSPM, app security) cross-sell into the installed base — a second large TAM
- Move up-market / international + multi-year enterprise commitments
- Consolidation of point tools (logs, SIEM, RUM) onto one bill — vendor-rationalization tailwind
Threats
- Hyperscalers bundling native observability (CloudWatch, Azure Monitor, Google Cloud Ops) at lower/zero marginal cost
- OpenTelemetry commoditizing instrumentation, lowering switching costs and enabling cheaper/open-source backends
- Cisco (Splunk+AppDynamics), Dynatrace, Grafana, New Relic, Elastic competing hard on price/openness
- Macro-driven IT-budget tightening + renewed cloud-cost optimization compressing usage growth
Moats, dependencies & bottlenecks
Moats
One agent, one pane across metrics/logs/traces/APM/security; ripping it out is costly and risky for ops teams.
Dashboards, alerts, runbooks, and historical telemetry are embedded in customer workflows; OTel erodes this slowly.
>83% of customers on 2+ products; each new product (security, AI monitoring) raises attach and stickiness.
Default choice for cloud-native teams; reinforced by fast product cadence and ecosystem integrations.
Ingest scale lets Datadog optimize cloud COGS and fund R&D faster than subscale rivals.
Dependencies
Supplier (compute/storage) + channel + competitor Datadog runs ON the hyperscalers (COGS) and sells THROUGH their marketplaces, yet they ship native rivals (CloudWatch/Azure Monitor/Cloud Ops) — supplier, partner, and competitor at once.
Usage-based revenue tracks customers' cloud/AI consumption; a cost-optimization cycle directly slows growth.
Revenue concentration Largest customer + several big AI-native names; cohort growth is lumpy quarter to quarter.
Ecosystem standard Reduces instrumentation lock-in; Datadog embraces it but it lowers backend switching costs over time.
Advantages
- Single unified platform / single agent
- Best-in-class 80% gross / 27% FCF margins with net cash
- Re-accelerating 30%+ growth at $4B ARR scale
- Multi-product attach + low-120s% NRR expansion engine
- Direct, structural AI + cloud workload tailwind
Weaknesses
- Premium valuation with no error margin
- Usage-based revenue is two-sided (optimization risk)
- AI-native concentration is lumpy
- High SBC / dilution; thin GAAP profit
Bottlenecks
- Cloud-cost optimization cycles capping usage growth
- OpenTelemetry lowering switching costs / commoditizing instrumentation
- Enterprise sales cycle length for large multi-product deals
- Cloud-infra COGS efficiency at higher ingest volumes
Top signals & trends
Top signals
Re-accelerating expansion inside the base is the cleanest read on demand recovery.
Lumpy; watch for renegotiation or insourcing by the largest AI customers.
Top-of-funnel strength supports durable multi-year growth.
Aggressive bundling/price cuts would pressure net-new and renewals.
Second-act categories extending TAM beyond core observability.
Trends
New telemetry surface (inference, agents, model cost/quality) — a structural growth vector and new product category.
More distributed systems = more to monitor; core secular tailwind.
Expands the addressable instrumentation base but lowers backend lock-in — double-edged.
Enterprises collapsing point tools onto one platform favors the broad-suite leader.
Pressure to trim observability spend caps usage-based upside in downturns.
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.
Primary cloud-infra supplier (compute/storage in COGS) + marketplace channel; also a competitor via CloudWatch.
Cloud-infra supplier + co-sell channel; competitor via Azure Monitor.
Cloud-infra supplier + marketplace; competitor via Cloud Operations suite.
Indirect — GPU/AI-infra that powers the AI workloads Datadog increasingly monitors (ecosystem driver, not a direct vendor).
Representative of the AI-native cohort driving inference-monitoring demand (illustrative of cohort, not disclosed by name).
4,550 customers at $100k+ ARR (~90% of ARR); ~33,200 total — broad, cloud-native installed base.
Representative cloud-native digital-commerce platform — the type of high-telemetry customer Datadog serves.
Closest pure-play observability rival; AI (Davis) + unified platform, trades at a much lower multiple.
Splunk (SIEM/observability) + AppDynamics (APM) under Cisco — enterprise-incumbent bundle and security adjacency.
Native cloud observability bundled into Azure; supplier + channel + competitor.
Native AWS monitoring; 'good-enough' for AWS-only shops and price-bundled.
Open-source-rooted observability + search/SIEM; screens cheap on valuation.
Consumption-priced APM/observability; no longer public.
Open-source LGTM stack (Loki/Tempo/Mimir) — the open/cheap alternative pressuring per-host economics.