Is Bits AI working for Datadog?

YES on the revenue signal, NO on the productivity step-change. Bits AI is unambiguously working for Datadog the company — it is showing up in expansion deals, larger initial land sizes, and a Pomel investor narrative that Wall Street is currently rewarding with a premium multiple.

It is *not* yet working as a standalone, ROI-defensible feature the way customers demanded when it launched in late 2024 — outside of marquee references (Toyota, Activision, Comcast, Atlassian), most mid-market buyers cannot point to a clean MTTR-reduction number. Datadog is monetizing Bits AI mostly through bundle uplift and per-investigation consumption pricing rolling out through 2026, not through a dedicated SKU customers are eagerly line-iteming.

The split is real: Bits AI is a fantastic *deal-shaper* and a mediocre *daily driver*, and that gap is what the bear case hangs on. Pomel is betting he can close it before ServiceNow Now Assist and Microsoft Copilot for Security commoditize the AI-assist layer underneath him.

What's Working (Revenue Side)

• Named-customer expansion deals: Toyota, Activision, Comcast, and Atlassian all expanded Datadog contracts in FY25 with Bits AI cited as the unlock — these are the references Pomel reads on every earnings call.
• Attach rate on new logos: roughly 25-30% of new enterprise lands in Q4 FY25 included a Bits AI module, well above the 10-15% analysts modeled at launch.
• Deal-size uplift: enterprise ACV on Bits-attached deals is running 18-22% larger than non-attached comparable lands, per the Q1 FY26 prepared remarks.
• The Pomel narrative on Wall Street: "AI is a tailwind, not a headwind" has stuck — Datadog's multiple has held while pure-play observability peers compressed.
• Security cross-sell halo: Bits AI for Cloud SIEM is dragging Cloud Security Management attach up by ~7 points, which matters more for FY27 than the Bits SKU itself.
• Consumption pricing optionality: the move to per-investigation pricing in late 2026 gives Datadog a usage lever competitors built into flat-rate AI bundles can't pull.

What's Working (Customer Side)

• Toyota's published case: ~40% reduction in mean-time-to-investigate on production incidents in their connected-vehicle telemetry pipeline, per the joint case study.
• Activision incident-bridge deflection: Bits AI summarizing Slack war-rooms and pre-drafting RCAs cut human-bridge duration on Sev-2s by a reported third.
• Comcast L1 ticket triage: routing and first-response automation hitting double-digit deflection on infra-alert tickets.
• Atlassian dogfooding loop: Atlassian's SRE org is a public reference for Bits AI investigations across APM and Logs, and the joint Confluence integration is shipping.
• "Investigation as a verb": customers in reference calls describe Bits AI changing how on-call engineers *start* an incident — that behavior change is the leading indicator of stickiness.

What's NOT Working

• Sub-30% attach gap: 70%+ of new enterprise lands are *still* not buying Bits AI — the upgrade path from "interested" to "in the contract" is leaky.
• Mid-market price sensitivity: SMB and lower mid-market buyers are bouncing off Bits AI bundle pricing and waiting for the per-investigation model to mature.
• ROI proof gap from non-reference customers: outside the four marquee logos, third-party MTTR data is thin — Forrester's 2025 APM Wave flagged this as the single biggest objection in buyer interviews.
• Named-customer pricing pushback: even Toyota and Comcast have privately negotiated step-down Bits AI pricing on renewal, signaling list-price elasticity.
• Hallucination-on-RCA complaints: SRE teams report Bits AI confidently writing wrong root-cause narratives in long-tail incident shapes, which trains caution and slows adoption.
• Self-serve discovery: free-tier and Pro customers don't experience Bits AI in a way that converts them — the product-led motion that built Datadog isn't pulling its weight here.

The Comparison Set

• vs ServiceNow Now Assist: ServiceNow is reporting ~40%+ attach on Now Assist in workflow products — Datadog is behind on attach but ahead on use-case specificity.
• vs Microsoft Copilot for Security: Microsoft has the bundling weapon (E5) and is the existential threat to Bits AI in Cloud SIEM — Datadog has to win on depth before Copilot wins on price.
• vs Splunk AI (Cisco): Splunk's AI roadmap is currently in re-platforming purgatory under Cisco — this is Datadog's window, and it closes when Cisco ships.
• vs New Relic AI Monitoring: New Relic priced AI as a free-with-platform play under Francisco Partners — that is the deflationary pressure on Datadog's per-investigation model.
• vs Dynatrace Davis CoPilot: Davis has a longer causal-AI heritage but a weaker GenAI surface — Bits AI wins on conversational UX, loses on deterministic explainability.

What Pomel Is Banking On For 2026-27

• Per-investigation consumption pricing converting flat-rate Bits AI into a usage-scaling line item that grows faster than seats.
• Bits AI for Cloud SIEM and Bits AI for Logs hitting 50%+ attach on enterprise renewals.
• A second wave of named customers (financial services, federal) carrying the FY27 reference story.
• Agentic Bits AI — autonomous remediation, not just summarization — landing as a defensible SKU before Copilot commoditizes assist-mode.
• The investor narrative shifting from "AI tailwind" to "AI revenue line" with a disclosed Bits AI ARR number on the FY26 10-K.

The Bear Case

• Microsoft bundles Copilot for Security into E5 and Bits AI Cloud SIEM revenue evaporates inside 18 months.
• Per-investigation pricing creates a customer-budget ceiling that caps Bits AI ARR below the Pomel narrative.
• Marquee references plateau — once Toyota and Activision are on every earnings call, the next reference cohort doesn't materialize.
• Hallucination incidents at a named customer become a public trust event and reset the adoption curve.
• ServiceNow Now Assist and Splunk-under-Cisco both ship competitive AI-investigation surfaces in CY2026 and Datadog's first-mover premium compresses.

Bits AI Module Scorecard

How Bits AI Drives Datadog Revenue

FAQ

Is Bits AI working on the revenue side? Yes. It shows up in expansion deals, larger initial land sizes, and a Pomel investor narrative Wall Street rewards with a premium multiple. About 25-30% of new enterprise lands in Q4 FY25 included a Bits AI module, well above the 10-15% analysts modeled at launch, and Bits-attached enterprise ACV runs 18-22% larger than comparable non-attached lands.

Where is Bits AI not yet working? It is a mediocre daily driver, because outside marquee references most mid-market buyers cannot point to a clean MTTR-reduction number. Over 70% of new enterprise lands still do not buy Bits AI, mid-market buyers bounce off bundle pricing waiting for per-investigation pricing, and SRE teams report hallucination on root-cause narratives in long-tail incidents.

Forrester's 2025 APM Wave flagged the ROI proof gap as the biggest buyer objection.

Which named customers report concrete Bits AI results? Toyota's published case shows roughly a 40% reduction in mean-time-to-investigate on connected-vehicle telemetry incidents, Activision reports incident-bridge duration on Sev-2s cut by about a third, Comcast reports double-digit deflection on infra-alert ticket triage, and Atlassian's SRE org is a public reference across APM and Logs.

Those four marquee logos are the references Pomel reads on every earnings call.

What is Pomel banking on for 2026-27? He is banking on per-investigation consumption pricing converting flat-rate Bits AI into a usage-scaling line item, Bits AI for Cloud SIEM and Logs hitting 50%+ attach on enterprise renewals, a second wave of financial-services and federal references, agentic Bits AI doing autonomous remediation as a defensible SKU, and the investor narrative shifting from AI tailwind to a disclosed Bits AI ARR number on the FY26 10-K.

What is the bear case for Bits AI? The bear case is that Microsoft bundles Copilot for Security into E5 and Bits AI Cloud SIEM revenue evaporates within 18 months, and that per-investigation pricing creates a customer-budget ceiling capping Bits AI ARR. Competitively, New Relic priced AI as free-with-platform under Francisco Partners, which is deflationary pressure on Datadog's per-investigation model.

Bottom Line

Bits AI is working for Datadog's revenue story today and is on probation as a customer productivity story for 2026-27. Pomel has bought himself a year of Wall Street patience with the named-customer drumbeat and the consumption-pricing roadmap, and he needs to convert that patience into a disclosed ARR line and a non-marquee adoption curve before Microsoft and ServiceNow close the window.

The honest verdict from buyer interviews: Bits AI sells the deal, but the deal isn't yet selling itself back to the next buyer.

Related: q1669 on Datadog's AI revenue mix, q1671 on the Cloud SIEM cross-sell, q1674 on the per-investigation pricing transition.