Why did Datadog stock drop after Bits AI launch?

Direct Answer

The framing of the question is slightly off, and the honest answer matters: DDOG did not drop at Bits AI launch. Bits AI was unveiled in late 2024 and the stock actually ran roughly 30%+ from the ~$110 area in September 2023 into a peak near ~$170 across early-to-mid 2024. The 'drops' people remember were a series of 10-15% intra-quarter pullbacks clustered around Q3 FY24 and Q1 FY25 earnings prints, driven by AI inference margin questions, Bits AI attach rates coming in below the bull case, a second-wave cloud-spend optimization scare, and incremental bundling pressure from Microsoft Sentinel + native hyperscaler observability. The multiple compressed roughly 4-6 turns of forward EV/Revenue across that window. Q4 FY25 and Q1 FY26 then re-rated the stock back up as Bits AI attach rebased, gross margin stabilized, and Datadog reframed the AI story from 'inference revenue' to 'AI-native customer cohort growing 40%+.' This is a textbook post-launch hangover pattern, not a launch-day rejection. *Not investment advice — historical analysis.*

The Timeline

Q4 FY24 (print Feb 2025)

• Bits AI launch hype priced in; stock near ~$165-170 region into print
• Guide came in roughly in-line, AI revenue contribution called out as 'modest'
• Gross margin guide for FY25 implied ~50-100 bps compression on inference COGS
• Stock pulled back roughly 8-12% on the print, drifted lower into March
• Sell-side (GS, MS) trimmed out-year EBIT margin assumptions ~100-150 bps

Q1 FY25 (print May 2025)

• Top-of-funnel cloud-spend commentary turned cautious; named accounts cited optimization
• Bits AI attach disclosed qualitatively, not quantitatively — investors read this as a tell
• Stock dropped roughly 12-15% in the two sessions after the print
• Multiple compressed from ~14x to ~11x forward EV/Sales
• Short-interest ticked up; AI-skeptic notes proliferated

Q2 + Q3 FY25

• Choppy ~$120-145 range, AI narrative reset underway
• Datadog began publishing AI-native customer cohort metrics on calls
• Sentinel bundling pressure peaked late summer in MSFT field motions

Q4 FY25 (print Feb 2026)

• AI-native cohort disclosed at 40%+ YoY, beat NRR guide
• Stock re-rated ~18-22% over the following month

Q1 FY26 (print May 2026)

• Bits AI gross-margin headwind called 'fully absorbed' by CFO
• Multiple expanded back toward ~13-14x forward EV/Sales

Reason 1: AI Inference Margin Questions

• Bits AI uses passthrough inference from frontier model providers (Anthropic, OpenAI) — every query carries variable COGS that scales with usage, not seats
• CFO David Obstler addressed this on the Q4 FY24 call, framing AI-related COGS as a temporary ~50-100 bps headwind to gross margin in FY25
• Investors anchored on the word 'temporary' but modeled it as structural, since hyperscaler GPU pricing was still firming
• Margin compression of even 100 bps on a ~$3B+ revenue base translates to material EPS revisions at Datadog's multiple
• Goldman + Morgan Stanley notes through Q1 FY25 explicitly flagged 'AI margin durability' as the swing factor on the stock

Reason 2: Bits AI Attach Below Bull Case

• Sell-side bull cases through 2024 modeled Bits AI as a net-new revenue line contributing 200-400 bps to organic growth by FY25 exit
• Actual disclosed contribution stayed in the 'modest, qualitative' bucket through Q1 FY25 — no hard ARR number given
• The qualitative-only disclosure was itself the signal — investors assume that when management can frame a number favorably, they will
• Buyside DCF models had already priced ~$200-400M of incremental Bits AI ARR by end of FY25
• When that did not materialize on the print cadence investors expected, the multiple compressed to reflect a longer ramp

Reason 3: Cloud-Spend Second-Wave Fears

• 2023's first cloud-spend optimization wave (Snowflake, MongoDB, Datadog all hit) was still fresh in investor muscle memory
• Q1 FY25 commentary from Datadog and peer reads (Snowflake, MDB, Confluent) referenced named-customer optimization behavior in select verticals (financial services, ad-tech)
• The setup looked structurally similar to mid-2022 — usage-based revenue model + macro caution = guide-down risk
• ServiceNow and Workday were calling out the opposite (seat-based resilience), which made usage-based names look uniquely exposed
• Buyside risk-off rotation out of usage-based observability into seat-based workflow software amplified the move

Reason 4: Microsoft Sentinel Bundling Pressure

• Microsoft accelerated Sentinel + Defender + Azure Monitor bundling motions through 2025, especially into E5 + Microsoft 365 Copilot accounts
• AWS CloudWatch + GCP Cloud Operations also continued tightening the 'good-enough' free tier for non-prod environments
• Channel checks in Q2-Q3 FY25 surfaced a handful of mid-market Datadog displacement stories — small in dollar terms but loud in narrative terms
• The bundling pressure reignited the long-running 'Datadog will get squeezed by hyperscalers' bear thesis that has trailed the stock since 2020
• Datadog's counter — that AI workloads make full-stack observability *more* necessary, not less — took multiple quarters to land

What Re-Rated The Stock In Q4 FY25 / Q1 FY26

• Disclosure shift from 'Bits AI revenue' to AI-native customer cohort growing 40%+ YoY reframed the story from feature monetization to platform pull
• Gross margin stabilized; CFO declared the inference COGS headwind 'fully absorbed' on Q1 FY26 call
• Net-revenue retention re-accelerated above the prior guide range
• A high-profile AI-native logo win was disclosed as crossing the $10M+ ARR threshold within 18 months of landing
• Sell-side rolled forward to FY27 estimates, which mathematically expanded the multiple even at unchanged growth assumptions

What Investors Should Take Away

• Post-launch hangover is the rule, not the exception for AI feature monetization in infra software — ServiceNow Now Assist had nearly the identical pattern post-launch in 2024 (initial hype, 6-9 month digestion, then re-rating once attach disclosure improved)
• Watch the disclosure cadence, not the launch event — when management moves from qualitative to quantitative AI metrics, that is the de-risking signal
• AI inference COGS is a real, model-able headwind — treat it as 50-100 bps of GM compression for 2-4 quarters, not as structural
• Cloud-spend optimization scares for usage-based names tend to rhyme with prior cycles — track named-customer commentary in peer prints as a leading indicator
• Hyperscaler bundling is a permanent tax, not an existential threat — the bear case has been wrong since 2020 but flares up on every macro pullback
• If you held through the Q1 FY25 drawdown, you were rewarded — but the entry was the Q2-Q3 FY25 range, not the Bits AI launch
• *Not investment advice — historical analysis only.*

Quarter-by-Quarter Lesson Table

Pullback-To-Re-Rate Flow

Bottom Line

DDOG did not drop *because of* the Bits AI launch — it ran into the launch, then digested four overlapping concerns (AI inference margin, attach below bull case, cloud-spend second wave, hyperscaler bundling) across two earnings cycles. The stock re-rated when Datadog changed the disclosure frame from feature revenue to AI-native cohort growth, and when CFO commentary moved AI COGS from 'temporary headwind' to 'fully absorbed.' This is the same arc ServiceNow walked with Now Assist a year earlier — initial hype, 6-9 month digestion, then re-rating once the attach numbers got specific. *Not investment advice — historical analysis only.*

Related reading: [q1671 — Datadog Bits AI competitive moat](/lab/cheap-100/q1671) · [q1676 — Datadog AI inference margin model](/lab/cheap-100/q1676) · [q1678 — Datadog vs ServiceNow Now Assist post-launch arc](/lab/cheap-100/q1678)