The Demand Hinge — Is "Search AI" Bending the Curve, or Just Holding It?

The Demand Hinge — Is "Search AI" Bending the Curve, or Just Holding It?

The through-line set in /chapter-1 rests on one demand-side question, and /chapter-2 answered only half of it: the moat is real but narrow. The other half is whether that moat is converting into re-accelerating growth. This chapter takes the two gauges that decide it — and finds them pointing in opposite directions.

The trailing gauge has not moved. Elastic's Net Expansion Rate, the dollar-weighted measure of how much existing customers grow their spend, has been frozen at approximately 112% for eight consecutive quarters [1]. Two-plus years into the generative-AI wave that was supposed to lift it, the needle sits exactly where it did before.

The forward gauge just jumped. Remaining performance obligations — the contracted backlog that becomes future revenue — accelerated to 28% year-over-year growth in Q4 FY2026 (27% in constant currency), the fastest in the company's recent history, while reported revenue grew 16% [2]. Management staked an explicit revenue re-acceleration on that divergence. The investment case, in one line, is a bet that the second gauge leads the first.

The metric that hasn't moved

Net Expansion Rate is the cleanest read on whether Elastic's land-and-expand engine is gaining or losing torque. The multi-year arc tells a story of decline, then a floor — not a recovery.

Source: NER bottomed at 109% in Q3 FY2024 and has held at ~112% since Q1 FY2025; the FY2026 year-end reading of approximately 112% is per the FY2026 10-K [3]; the FY2024 year-end reading of approximately 110% is per the FY2024 10-K [4]. Quarterly cadence from Elastic quarterly results press releases, as reported.

The shape matters. A figure that fell from 117% (FY2023) to a 109% trough, then settled at 112% [5], is a business that stopped bleeding but has not re-ignited. At 112%, the average customer is growing roughly twelve cents on the dollar a year — respectable, but a long way from the 130% Elastic posted before the 2022 spending reset. A bull who claims AI is "transforming" the demand profile has to reconcile that claim with a number that has not flickered for two years. So far, AI shows up everywhere in the narrative and nowhere in the expansion rate.

The metric that just moved

The case for patience lives in the backlog. Because Elastic recognizes much of its revenue over the life of multi-year contracts, the order book turns before the income statement does. Through FY2026 the two diverged sharply: revenue growth drifted down toward the mid-teens while contracted commitments accelerated.

Source: total RPO (rpo_pct), current RPO (crpo_pct), and total revenue (rev_pct) constant-currency growth, Q1–Q4 FY2026 reconciliation tables — Q4 FY2026 results [6]; Q3 FY2026 results [7]; Q2 FY2026 results [8]; Q1 FY2026 results [9].

Total RPO constant-currency growth climbed quarter by quarter — 15%, 17%, 18%, then 27% — even as revenue slipped to 14%. Current RPO (the slice due within twelve months, the tightest leading indicator of next year's revenue) snapped from 15% to 20% in Q4 [10]. CFO Navam Welihinda laid out the causal chain without hedging: "The acceleration of CRPO is also what gives us confidence in our expected revenue acceleration over the next 12 months. As increasing commitment volumes accelerates constant currency CRPO, and then constant currency revenue, in that order" [11].

A skeptic should press two caveats before buying it. First, the Q4 backlog spike was flattered by mix: management said the quarter carried "a significantly larger mix of cloud commitments compared to historical patterns" [12], and cloud contracts book differently than self-managed ones. Second, RPO is a contracted commitment, not consumed revenue — for a usage-based model, a customer can commit and then under-consume. The bull and bear will argue this chart for the next year. But the direction is unambiguous, and it is the first hard evidence in two years that the demand engine may be re-engaging.

The engine underneath: AI is now a third of the franchise

What is driving the commitments is no longer a slideware claim. Elastic has begun disclosing the AI-attached cohort, and it is compounding visibly. The count of large customers — those spending over $100,000 a year — that use Elastic for AI rose through FY2026 from roughly 370, to over 470, to over 600.

Sources: 600+ $100k-ACV AI customers, now more than a third of the cohort — Q4 FY2026 transcript [13]; over 3,000 total AI customers and over 2,700 vector-database customers — Q3 FY2026 transcript [14]; 370+ $100k GenAI customers in Q2 — Q2 FY2026 transcript [15].

Source: 370+ ($100k GenAI cohort, Q2) per Q2 FY2026 transcript [16]; 470+ per Q3 FY2026 transcript [17]; 600+ per Q4 FY2026 transcript [18].

Three things make this cohort the load-bearing evidence for the bull case. First, penetration is climbing fast: AI use cases now reach more than a third of the $100,000-ACV base, up from 17% just a year earlier [19]. Second, the AI cohort grows faster than the rest — management quantifies it as "roughly 5%, a little over 5% faster than the rest of the cohorts," a gap they expect to persist and widen as adoption spreads [20]. Third, the use cases are broadening beyond the original vector-search niche: of the 470+ large AI customers in Q3, more than 410 used Elastic as a vector database, with the balance — and a fast-growing share of new demand — coming from agentic workflows, security, and observability [21].

The honest counterweight is magnitude. The CFO sizes the AI tailwind at "average to 5%" of growth today, with "opportunity to accelerate beyond that 5%" as penetration rises [22]. Five points of growth is a real tailwind, not a transformation — it is roughly the difference between a low-teens and a high-teens grower. For AI to bend the curve rather than merely defend it, that 5% has to become 8% or 10%. The disclosure shows the slope is right; it does not yet show the magnitude the bull needs.

Why the headline still decelerates

Set against accelerating backlog and a compounding AI cohort, Elastic's own FY2027 revenue guidance looks oddly cautious — and the gap is where the debate concentrates.

Source: reported revenue FY2023–FY2026 per company financial statements, as reported; FY2027 total-revenue guidance of $1.985B–$2.0B (14.6% growth at midpoint) per Q4 FY2026 transcript [23].

At the headline, FY2027 total revenue is guided to roughly 14.6% growth — a step down from FY2026's 17% [24]. Three things reconcile that with the accelerating backlog. First, the headline is dragged by monthly Elastic Cloud — the self-serve consumption line — which grew just 3% in Q4 and is a structural anchor on the blended number [25]. The cleaner signal, sales-led subscription revenue, is guided to grow 16.9% and to accelerate quarter by quarter through the year [26]. Second, the guide carries deliberate conservatism: the CFO cited "risk adjustment related to consumption, related to FX, related to timing of large deals and mix" [27]. Third, management raised the medium-term targets at the same time — reaffirming a 20%-plus sales-led growth goal for FY2029 and lifting the FY2029 operating-margin target from above 20% to approximately 25% [28].

So the same management team is guiding the next year down and the destination up. That is internally consistent only if you believe the backlog converts — which is precisely the bet.

What this means for the thesis, and what to watch

This chapter sharpens the through-line into a falsifiable proposition. Elastic's demand is not deteriorating — consumption is "stable," the AI cohort is compounding, and the order book is accelerating [29]. But it is not yet re-accelerating where it counts: the expansion rate is flat and reported growth is guided lower. The bull owns the leading indicators; the bear owns the lagging ones. The next twelve months adjudicate between them.

The two numbers to watch are therefore narrow and concrete: constant-currency current RPO growth (is the Q4 jump a trend or a mix-driven blip?) and Net Expansion Rate (does it finally break 112%?). Everything else in the AI story — the customer logos, the vector-database downloads, the agentic features — is a leading argument for those two figures. The reader now knows exactly where to look. The cash economics that turn this growth into owner value — and the stock-based compensation that taxes it — are the natural next subject.