Datadog pricing

About

Datadog is a publicly traded (NASDAQ: DDOG) observability and security platform that unifies infrastructure monitoring, application performance monitoring (APM), log management, real user monitoring, security, and — more recently — AI/LLM observability into a single product surface built around 40+ independently metered SKUs. Founded in 2010 by Olivier Pomel and Alexis Lê-Quôc and headquartered in New York, the company IPO’d in 2019 and has since grown into one of the largest independent observability vendors, reporting over $2.7 billion in annual revenue.

Datadog sells to teams of every size — from individual developers monitoring up to 5 free hosts to large enterprises running DevSecOps bundles across thousands of hosts and containers — competing most directly with New Relic, Dynatrace, Splunk (observability side), Grafana Labs, and the native cloud-provider monitoring tools (AWS CloudWatch, Azure Monitor, Google Cloud Operations). Its pricing page (datadoghq.com/pricing/list) is unusual among enterprise software vendors for publishing exact per-unit rates — annual, month-to-month, and on-demand — for essentially its entire product catalog rather than gating detailed pricing behind a sales conversation.

Pricing summary : How Datadog’s à la carte, per-host usage-metered pricing works

Datadog uses a hybrid, à la carte metering model — not one tiered plan, but 40+ independently priced products, each with its own billing unit:

1. Per-host/per-device products: Infrastructure, APM, Database Monitoring, Continuous Profiler, Workload Protection, and Edge Device Monitoring are billed per monitored host (or device) per month, each with its own Free/Pro/Enterprise-style tier ladder (e.g., Infrastructure: Free $0 → Pro $15 → Enterprise $23 per host/month, annual).
2. Per-volume products: Logs separate ingestion ($0.10/GB/month) from indexing ($1.06-$2.50 per million indexed events depending on 3- to 30-day retention); Cloud Security Management and Workload Protection add per-container pricing on top of per-host pricing.
3. Per-session/per-event products: RUM ($0.15/1,000 sessions), Synthetic Monitoring ($5/10K API test runs, $12/1K browser test runs), Custom Metrics ($5/100 metrics), and Custom Events ($2/100K events) meter product-analytics-style volume.
4. Newer AI-native meters: Agent Observability (LLM Observability) bills $160/month for the first 100,000 LLM spans plus $3.50 per additional 10,000 spans; AI Credits are sold in $500 bundles of 500 credits/month ($1.30/credit on-demand) to power Bits AI features.
5. Billing-term multiplier: almost every SKU publishes three side-by-side rates — billed annually (cheapest), billed month-to-month, and billed on-demand (most expensive, typically 15-25% above annual) — rather than a single list price.

What makes this different: instead of packaging value into a handful of named plans, Datadog publishes a flat public rate card for essentially its entire 40+ product catalog, letting a buyer self-serve-price any combination of infrastructure, logs, APM, security, and AI-observability products without contacting sales.

Pricing by product

Infrastructure Monitoring

Application Performance Monitoring (APM)

Log Management

Real User Monitoring, Synthetics & Continuous Testing

Security (Cloud Security Management, Workload Protection, Code Security)

AI / LLM Observability & AI Credits

Sales motions across products: self-serve / PLG for every metered SKU above — all list prices are publicly quoted and purchasable without a sales call; sales-led for Datadog Disaster Recovery (custom-quoted “Contact Sales”) and for negotiated multi-year/volume discounts on 500+ host deployments.

The billing-term multiplier: annual vs. month-to-month vs. on-demand

Nearly every SKU on Datadog’s price list publishes three rates for the identical entitlement rather than one list price. Using Infrastructure Pro as the example: $15/host/month billed annually, $18/host/month billed month-to-month, and $18/host/month billed on-demand (no contract) — the same $15-$18 spread repeats across almost all 100+ SKUs, meaning the effective discount for committing to annual billing is consistently in the 15-25% range regardless of which product a customer buys. This is a textbook case of the commitment-tier structure covered in our guide on usage invoicing and billing cycles.

Hidden costs : the classic Datadog “bill shock” mechanics

The per-host headline understates what teams running real infrastructure actually pay, because Infrastructure, APM, Logs, and Custom Metrics stack independently on the same hosts. Two documented patterns — high-cardinality custom metrics and the ingestion/indexing split on Logs — account for most of the community-reported “surprise bill” stories (see Pricing evolution and Strengths & weaknesses below). This is a live example of the AI cost unpredictability and bill-shock problem that FinOps teams increasingly track for usage-metered infrastructure, not just AI spend.

Archetype 1 — mid-size team hit by tag-driven custom-metric cardinality

A 200-host fleet on Infrastructure Pro that tags one API-latency metric with a customer_id carrying 1,000 unique values can turn a routine metrics bill into a five-figure line item, per multiple independent cost-optimization write-ups (Sysdig, Sawmills.ai) citing real customer cardinality blowouts:

A single poorly scoped tag (customer_id, user_id, request_id) can add more to the monthly bill than the entire host fleet’s base Infrastructure charge — which is why Datadog ships “Metrics without Limits” as a dedicated cardinality-control feature rather than leaving it to chance.

Archetype 2 — Logs ingested broadly, indexed selectively

A team piping 5 TB/month of application logs into Datadog but only indexing (making searchable/alertable) a 20% slice for 15-day retention:

The lesson this table teaches: ingestion and indexing are separate meters on purpose, so a team that forgets to filter what it indexes pays twice — once to get logs in, again to make them searchable — and a well-tuned indexing filter can cut the second charge by 80%+ without losing raw log retention.

Want to estimate your own Datadog bill? Use the Datadog pricing calculator to model your monthly cost based on host counts, log volume, and the specific products your team turns on.

Pricing evolution : stable core meter, expanding SKU catalog

Cadence

Tracked range: 2022-01–2026-09. Quarters shown above reflect dated, source-confirmed events. Other quarters in this range are not independently verified and are not claimed stable; the one data point confirmed unchanged across the full window is the core Infrastructure per-host rate ($0/$15/$23).

Notable changes

• 2022-01-21 — Earliest confirmed reference point: Infrastructure Free/Pro/Enterprise pricing identical in structure and rate to the 2026 rate card; Pro’s included-integrations count has since grown from “500+” to “1,000+” and Enterprise’s “Premium support” line was replaced by a “Governance Console” feature at some unverified point between 2022 and 2026.
• 2023-05-04 — Datadog CFO David Obstler discloses a “large upfront bill that did not recur,” analyst-estimated at ~$65M, from a Q1 2022 crypto-industry customer on the Q1 2023 earnings call.
• 2024-06-26 — LLM Observability reaches general availability (Datadog press release, syndicated via PR Newswire).
• 2025-06-10 — Bits AI SRE, Dev, and Security Analyst agents unveiled at DASH 2025.
• 2025-12-02 — Bits AI SRE reaches general availability to all customers; AI Credits introduced as the shared billing currency for the Bits AI product family (Datadog press release, syndicated via Nasdaq).

The $65M bill disclosure in detail

Datadog’s usage-based, per-host/per-GB model means a customer’s bill scales directly with their own infrastructure footprint — which is exactly what made this event notable. On the Q1 2023 earnings call (2023-05-04), CFO David Obstler referenced a “large upfront bill for a client in Q1 2022 that did not recur at the same level or timing in Q1 2023.” A JPMorgan analyst calculated the bill at roughly $65 million; the customer was widely reported — via journalist Gergely Orosz’s sourcing of multiple Coinbase engineers, and picked up by The Pragmatic Engineer and The New Stack — to be Coinbase. Datadog’s CEO explained on the same call that the customer’s business “was cut in three or four” amid the 2022 crypto downturn, and that Datadog “restructured their contract” to keep them as a customer at a lower spend level going forward. The story generated two separate Hacker News front-page threads (179 points and 97+ points with 200+ comments), and remains one of the most-cited data points in “usage-based pricing means unpredictable bills at scale” discussions across the observability community — a useful real-world companion to the entitlement-to-credits shift many usage-metered vendors are now navigating.

What’s unique : à la carte metering at 40+-product scale

1. The rate card is the sales page. Where most enterprise observability and security vendors gate detailed pricing behind a sales call, Datadog publishes exact per-unit rates — annual, month-to-month, and on-demand — for essentially its entire catalog at datadoghq.com/pricing/list. A buyer can self-serve-price any combination of Infrastructure, Logs, APM, Security, and AI products without ever talking to sales, which is unusual at Datadog’s scale (NASDAQ: DDOG, $2.7B+ revenue) and stands in contrast to competitors like Dynatrace and Splunk, whose enterprise tiers are typically quoted.

2. Every product is its own hybrid pricing model, stacked. Rather than one company-wide pricing model, Datadog runs 40+ independent per-product meters side by side — per-host (Infrastructure, APM), per-GB-plus-per-event (Logs), per-session (RUM), per-committer (Static Code Analysis), per-container (Workload Protection) — and a customer’s total bill is the sum of whichever subset they turn on. This lets a five-person startup and a Fortune 500 buy from the identical rate card, scaling spend with adoption breadth rather than forcing an all-or-nothing plan choice.

3. AI features get their own meter, layered on top of the base catalog. Rather than folding Bits AI and LLM Observability into existing per-host pricing, Datadog introduced AI Credits as a second, parallel consumption axis — sold in 500-credit bundles, consumed by Bits Chat/Investigation/Code/Agent Builder, separate from the per-span Agent (LLM) Observability meter. A customer’s AI spend is fully decoupled from their host count, so an org can have a small infrastructure footprint but heavy Bits AI usage (or vice versa) without either meter distorting the other.

4. The annual/monthly/on-demand spread is a structural constant, not a promotion. The ~15-25% surcharge for committing month-to-month or on-demand instead of annual repeats identically across nearly all 100+ SKUs — it functions as a standing default discount for commitment rather than a limited-time offer, making Datadog’s effective “list price” always somewhat higher than the headline number shown first on the page.

Strengths & weaknesses

Billing UX : Named controls on Datadog’s pricing surfaces

• Billed Annually / Billed Month-to-Month / Billed On-Demand columns — the price list (datadoghq.com/pricing/list) shows all three rates side-by-side for nearly every SKU rather than a single toggle, so a buyer can compare the commitment-vs-price tradeoff for one product at a glance.
• Datadog Site selector (US1/US3/US5, EU, etc.) — a dropdown at the top of both the main pricing page and the price list that switches the region/data-residency site a customer’s account runs on; pricing itself doesn’t change by site, but the control is prominent on every pricing surface.
• Allotments Calculator (datadoghq.com/pricing/allotments/) — a “Search all products” tool that, given a parent product and a desired host/container count, computes how many bundled units (custom metrics, ingested spans, containers, workflow executions, etc.) are included per host/hour, with a Monthly/Hourly toggle on the “Total Estimate” panel.
• Left-rail product category navigator on the main pricing page — the default view shows only the Infrastructure category’s tiers; switching products (APM, Logs, Security, AI, etc.) re-renders the tier grid rather than scrolling to a new section, so the visible price grid always reflects exactly one selected product family.
• “Multi-Year/Volume discounts available” banner — flagged prominently above the fold on the main pricing page as a “Contact Us” path for 500+ host deployments, distinct from the self-serve annual/monthly/on-demand rates below it.
• DevSecOps SKU exclusivity rule — per Datadog’s own pricing FAQ, DevSecOps Pro/Enterprise cannot be purchased for a subset of hosts (it applies to all existing hosts and containers) and cannot be combined with standalone Infrastructure or Cloud Security SKUs — an entitlement constraint enforced at checkout, not just documented in fine print.
• Asterisked per-container allotment note — rows like Container Monitoring, Custom Metrics, and the Cloud Security/Workload Protection container add-ons carry a footnoted asterisk: “For existing customers, please refer to your latest billing agreement for included allotments of this product,” meaning the published on-demand per-unit rate may not apply to grandfathered contracts.

Strategic wins : why the à la carte rate card keeps working at scale

1. Publishing the full rate card removes sales friction at every deal size

Datadog is one of the few public infrastructure-software companies at its scale to publish exact, unit-level pricing for essentially its entire catalog rather than gating it behind “Contact Sales.” That transparency lets engineering teams self-serve a cost estimate before ever talking to a salesperson, which is a meaningful usage-based pricing advantage in a developer-tools market where buyers increasingly research and provision without a rep in the loop.

2. Per-product metering let the catalog grow to 40+ SKUs without a repricing crisis

Because each product (Infrastructure, Logs, APM, RUM, Security, AI) is its own independently metered unit, Datadog has been able to add entirely new product lines — most recently LLM Observability (2024) and Bits AI (2025) — without renegotiating the pricing of existing products. This is the practical benefit of choosing hybrid, per-product pricing over a single company-wide plan: expansion revenue comes from adoption breadth, not price increases on the base meter.

3. Layering AI Credits on top of existing meters, not folding AI into them

Rather than raising Infrastructure or APM prices to subsidize Bits AI, Datadog metered it separately via AI Credits — a pattern also seen across the broader corpus as AI usage graduates onto its own dedicated meter. This keeps existing telemetry customers’ bills unaffected by AI adoption and lets Datadog price the AI product family on its own unit economics.

4. Renegotiating rather than losing the $65M customer

When a major customer’s spend became structurally unsustainable amid a 2022 industry downturn, Datadog’s public response (per its CFO and CEO on the Q1 2023 earnings call) was to restructure the contract rather than enforce the existing terms and risk churn. Retaining a marquee logo at a lower, sustainable price point is a stronger long-term signal to the market than the original $65M figure — and it is a rare instance of a usage-based vendor publicly narrating how it handles a bill that outgrew the customer’s ability to pay.

Areas to improve : where the à la carte model creates real friction

1. No native cost forecasting tool on the pricing surface itself

The Allotments Calculator (datadoghq.com/pricing/allotments/) estimates included allotments per host but does not model a full multi-product bill (Infrastructure + Logs + APM + Custom Metrics together) the way a real customer’s invoice combines them. Proposed fix: extend the calculator to accept a bundle of products at once and output a single blended monthly estimate, closer to what our own pricing calculator approach demonstrates is possible.

2. Cardinality-driven custom-metrics overages remain a self-inflicted trap

Datadog ships “Metrics without Limits” as a mitigation, but it is opt-in and requires the customer to already understand cardinality risk — the same knowledge gap that produces the tag-explosion bill-shock stories documented in Hidden costs above. Proposed fix: default new accounts into cardinality-limiting guardrails (with an explicit opt-out) rather than requiring proactive configuration, similar to how AI FinOps tooling increasingly defaults to cost governance and chargeback guardrails rather than leaving allocation to chance.

3. Logs ingestion-vs-indexing is not explained where a customer first encounters it

The ingestion/indexing split is one of the most common sources of confusion in third-party teardowns of Datadog’s pricing, yet the distinction isn’t surfaced prominently on the main pricing page — it only becomes clear once a customer reads the detailed price list. Proposed fix: add an inline example (like the one in this page’s Hidden costs table) directly on the Log Management pricing card, showing ingestion and indexing as two separate line items on a sample invoice.

4. AI Credits add a second forecasting axis without a combined estimator

Since AI Credits are billed independently of the host/GB meters that already dominate a Datadog invoice, a customer now has two largely independent consumption dimensions to project. Proposed fix: publish a combined “telemetry + AI” cost calculator so FinOps teams can model both axes together rather than reconciling two separate estimates by hand.

Key takeaways

1. A fully public, unit-level rate card scales past enterprise size. Datadog proves that a company generating $2.7B+ in revenue can still publish exact self-serve pricing for 100+ SKUs rather than gating everything behind sales — a model other infrastructure vendors evaluating usage-based pricing adoption should study directly.
2. Independent per-product metering supports large catalogs without repricing risk. Because each of Datadog’s 40+ products is billed on its own unit, new product launches (LLM Observability, Bits AI) don’t require renegotiating existing customers’ Infrastructure or APM rates.
3. New consumption axes should be layered, not blended, into existing meters. Datadog’s AI Credits sit alongside — not inside — its per-host and per-GB pricing, preserving legibility for customers who don’t yet use AI features while still monetizing those who do.
4. Cardinality and stacking meters are the real “bill shock” mechanism, not headline prices. Datadog’s list prices haven’t moved much in years; the community complaints are almost entirely about how independently metered products compound (custom-metric tags, ingestion-vs-indexing) rather than about any single sticker price changing.
5. Public willingness to renegotiate outsized bills builds durable trust. The ~$65M customer story could have been a reputational liability; instead, Datadog’s public earnings-call narration of restructuring the contract turned it into a case study in handling usage-based pricing’s biggest structural risk — a customer outgrowing what they can pay.

UBP implications

1. À la carte metering is a viable alternative to tiered plans at massive product-line scale. Datadog’s 40+ independently priced SKUs show that usage-based pricing doesn’t require collapsing everything into three or four named tiers — a mature usage-pricing company can instead expose dozens of granular meters and let the buyer compose their own bill.
2. Stacking usage meters is the primary source of “bill shock,” not price increases. The Datadog pattern — stable core prices, expanding product surface — suggests that usage-based pricing’s reputational risk comes less from raising rates and more from customers under-anticipating how many independent meters apply to their footprint simultaneously.
3. AI monetization is emerging as its own metering layer on top of legacy usage products. Datadog’s AI Credits, introduced alongside its established per-host/per-GB meters, is an instance of a broader pattern: incumbent usage-based vendors adding AI as a parallel consumption axis rather than repricing their core product to absorb AI costs.

Sources

• Datadog pricing (accessed 2026-09-01)
• Datadog pricing list (full SKU rate card) (accessed 2026-09-01)
• Datadog Allotments Calculator (accessed 2026-09-01)
• Datadog AI Credits billing documentation (accessed 2026-09-01)
• Datadog LLM Observability GA press release (accessed 2026-09-01)
• Datadog Bits AI SRE GA press release (accessed 2026-09-01)
• Datadog blog (accessed 2026-09-01)
• Datadog changelog (accessed 2026-09-01)

Bottom line

Datadog’s headline per-host prices have barely moved in four-plus years — the story is the surrounding catalog, which grew from about 14 named products in 2022 to 100+ independently metered SKUs across 40+ products by 2026, adding AI Credits as a new consumption axis for Bits AI along the way. The real cost risk for buyers isn’t a price hike; it’s underestimating how many of those 40+ meters — custom-metric cardinality and log indexing chief among them — end up stacking on the same infrastructure.

Want to compare Datadog against other observability pricing? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Baseline capture: à la carte per-host + usage metering across 40+ products

Sep 2026

Initial wiki-capture pass. Datadog's public price list shows 100+ SKUs metered independently — per-host tiers for Infrastructure/APM/Database Monitoring, per-GB/per-event metering for Logs, per-session metering for RUM, and a newer per-span Agent (LLM) Observability product with a $160/mo entry tier.

Bits AI SRE reaches general availability; AI Credits introduced as the metering unit

Dec 2025

Datadog's press release (2025-12-02, syndicated to Nasdaq) announced Bits AI SRE generally available to all customers. AI Credits became the shared billing currency across Bits Chat, Investigation, Code, and Agent Builder — a second AI-specific meter stacked on top of the existing per-host/per-GB catalog, sold in 500-credit bundles on annual, monthly, or on-demand commitment.

Bits AI SRE, Dev, and Security Analyst agents unveiled at DASH

Jun 2025

At Datadog's DASH 2025 keynote (2025-06-10), Datadog introduced the Bits AI agent family — SRE, Dev, and Security Analyst — for a roughly six-month production preview with about 2,000 customers ahead of general availability.

LLM Observability reaches general availability

Jun 2024

Datadog's own press release (datadoghq.com/about/latest-news, 2024-06-26; syndicated via PR Newswire) announced general availability of LLM Observability, adding prompt/response tracing, evaluation, and sensitive-data scanning for generative-AI applications — the first product line metered on an AI-native unit (spans/requests) rather than hosts, GB, or events.

CFO discloses ~$65M anomalous customer bill on Q1 2023 earnings call

May 2023

On Datadog's Q1 2023 earnings call, CFO David Obstler described a 'large upfront bill that did not recur' from a Q1 2022 crypto-industry customer, which a JPMorgan analyst estimated at roughly $65 million — widely reported (Gergely Orosz's sourcing, The Pragmatic Engineer, The New Stack) to be Coinbase. Datadog's CEO said the company restructured the contract as the customer's business was 'cut in three or four.' Two Hacker News threads on the story reached 179 and 97+ points.

Earliest verified snapshot: Infrastructure already $0 / $15 / $23 per host

Jan 2022

Wayback Machine capture of datadoghq.com/pricing (2022-01-21) shows Infrastructure Free/Pro/Enterprise at $0/$15/$23 per host/month, annual-vs-on-demand ($18/$27) already the standard structure, and a left-rail catalog of about 14 named products (Infrastructure, Log Management, APM & Continuous Profiler, Database Monitoring, Synthetic Monitoring, Incident Management, RUM, CI Visibility, Serverless, NPM, Cloud SIEM, CSPM, Sensitive Data Scanner, Workload Security). Enterprise carried a footnoted 100-host minimum not present on the 2026 rate card.