Elastic N.V. (NYSE: ESTC) — The Cheapest It Has Ever Been, on Profits It Did Not Earn

An independent fundamental research note. Report date: 2026-07-31 · Price: $65.79 (2026-07-31 close) · Market cap: ~$6.89B · Enterprise value: ~$6.11B Fiscal year end: April 30 · Latest reported period: FY2026 (ended 2026-04-30), reported 2026-05-28 Coverage status: Initiation · Sector: Information Technology — Infrastructure Software (Search, Observability & Security Analytics)

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information, not investment advice, and no reader should act on it without their own research. The analysis that follows takes no position and carries no price target — it discusses valuation only as embedded expectations and scenarios.

Verdict: HOLD / accumulate-on-weakness below the mid-$50s — a genuinely cheap asset attached to the weakest business in the cheap-software cohort. Explicitly NOT a short. Conviction: medium-low. Directional fair-value zone ~$62–78 (≈3.0–3.7× forward EV/sales on FY27E revenue of ~$1.99B, which is roughly where a mid-teens grower with 76% gross margins and real-but-stock-funded cash flow deserves to trade). Constructive-accumulation zone below ~$55 (~2.5× forward sales, where the ~$779M of net cash plus a 14× forward FCF multiple do most of the underwriting for you). Bear floor ~$40–48 — where the stock actually traded four months ago, and where it goes again if net expansion breaks below 105%. Bull case ~$95–110 if the +28% RPO and +43% non-current RPO convert into re-accelerating revenue and stock compensation falls below 13% of sales.

Two facts define this security, and they point in opposite directions. First: the cheapness is real. ESTC sits at the 6.8th percentile of its own eight-year valuation history — the 5.4th on price/sales, the 2.0th on price/book. It trades at 3.07× forward EV/sales against Dynatrace at ~5.8× and Datadog at ~17–18×. The stock closed at $70.00 on its IPO day in October 2018 and sits at $65.79 today, having grown revenue 10.9-fold in between. That is a genuine, quantifiable de-rating, not a narrative one. Second: the profits are not real. FY2026’s celebrated $367.8M of GAAP net income was manufactured by a $434.9M release of deferred-tax valuation allowances — the 10-K says outright that Elastic “would have incurred net losses in such years as well without the releases.” Strip it and you find a GAAP operating loss of $33.5M, the ninth consecutive one. And the $321.8M of free cash flow everyone cites is 92.7% funded by $298.4M of stock-based compensation; owner free cash flow — cash after paying the engineers in cash-equivalent — is roughly $23.4M on $1.74B of revenue, a 1.3% margin.

So what is it? My framing is early-stage bottoming in a structurally disadvantaged franchise — not a falling knife (the stock is +38% off its April all-time low with the 21-day EMA above the 50-day and the 200-day just reclaimed), and not yet momentum (Momentum factor loading −0.41, rs_12m −25%, Quality loading −0.26, and — most revealing — a Value loading of essentially zero, meaning the market has de-rated the stock without yet re-classifying it as a value asset). The thing that keeps me from being more constructive is not the valuation, it is the moat: Elastic is the only name in this cohort whose core engine has a free, Apache-2.0, Linux-Foundation-governed, hyperscaler-distributed drop-in substitute in OpenSearch. Dynatrace and Datadog do not have that problem. That is why Elastic earns 76% gross margins where GitLab earns 88%, why net expansion has been stuck at 110–112% for three years after peaking at 117%, and why it is the #3-or-worse player in each of the three markets it fights in simultaneously. The discount is earned. The question is whether it is over-earned, and at 3.07× forward sales with $779M of net cash and a backlog growing at twice the rate of revenue, I think it modestly is — which is a HOLD, not a table-pound.

Conviction: medium-low. Flips bullish if net expansion rate ticks above 115% while stock compensation falls below 14% of revenue — that combination would prove the data-gravity moat is pricing-powerful and that the FCF is becoming the shareholders’ rather than the employees’. Flips bearish if the FY27 guide of +14.6% is cut at the Q1 print on August 27, or if net expansion breaks below 105% — either would confirm that the RPO acceleration was multi-year-contract mix rather than demand, and turn “cheapest-ever” into the value trap it superficially resembles. Tag: the search company that gave its engine away — and is finally being paid for what it kept.

📈 Stock Price Action — Five-Year Event Map

Elastic’s five-year chart is a near-complete round trip through the software bubble and out the other side. From a November 2021 all-time high of $186.78, the shares fell 76.8% to an all-time low of $43.30 on 2026-04-10, and have since rebounded 51.9% to $65.79. The 52-week range is $43.30–$94.47; the stock is −64.8% from its high and −6% from its 2018 IPO-day close of $70.00, eight years and a 10.9× revenue increase later. Beta is 1.35, annualized volatility ~59%, and roughly 60% of the stock’s variance is idiosyncratic — this is a name that trades on its own guidance, not on the tape.

Cycle narrative. (1–2) The 2021 peak was a multiple, not a business: Elastic was growing ~40% with a −20% GAAP operating margin, and when rates rose that combination was repriced hardest. The 2022 collapse is almost entirely factor-driven — Elastic’s fundamentals did not deteriorate, its cost of capital did. (3) The 2023 recovery was the first “Search AI” trade: the Q2 FY24 print on 2023-11-30 delivered a beat-and-raise attributed to vector-search and RAG demand, and the stock gained 37% in a single session — the largest one-day move in its history. (4) August 2024 is the most important scar in the file: the Q1 FY25 miss was self-inflicted, caused by a botched sales-territory re-segmentation, and it cost 26.5% in a day. It is why the market discounts management’s forward commentary. (5–6) The FY26 prints in November and February were both received as deceleration and both cost ~15%. (7) The April 2026 low of $43.30 marked the maximum-pessimism point: ~2.3× EV/sales for a 17%-growing business with $779M of net cash. (8) The Q4 FY26 print on 2026-05-28 broke the sequence — total RPO accelerated to $1.98B (+28%), non-current RPO grew +43%, and the stock has rallied 51.9% off the low. Note the pattern the whole table describes: Elastic moves ±12–37% on essentially every earnings report. That volatility is a fact an owner must underwrite in its own right; the next test is 2026-08-27.

1. Executive Summary

Elastic N.V. sells the Elastic Search AI Platform — the commercial distribution of Elasticsearch, plus Kibana, Logstash and Elastic Agent — into three markets: search and AI context/retrieval, observability, and security analytics (SIEM and endpoint). FY2026 revenue was $1,739.3M, +17.3%, of which 94% is subscription and 48% is Elastic Cloud. It is a Dutch N.V. filing US domestic forms, with 46% of revenue outside the United States and 4,019 employees before a ~7% workforce reduction announced in June 2026.

The business is real and the cash flow is real: $326.9M of operating cash flow on $5.1M of capital expenditure, a growing customer-prepayment float (deferred revenue +21.4%), 1,720 customers above $100K of annual contract value, 240+ above $1M, and a contracted backlog (RPO $1.98B, +28%; non-current RPO +43%) growing roughly twice as fast as recognized revenue. Gross margin is 76.1% and improving.

Three things are equally real and cut the other way. First, Elastic has never earned a GAAP operating profit — FY2026’s was −$33.5M, and the reported net income of $367.8M is the arithmetic product of a $434.9M deferred-tax valuation-allowance release in the Netherlands, UK and California, which the 10-K concedes is the only reason the year was not a loss. The Dutch and Californian portions rest on a prospective “committed tax planning action in fiscal 2027,” not on demonstrated profitability. Second, the free cash flow belongs largely to employees: $298.4M of stock-based compensation (18% of revenue) against $321.8M of FCF leaves roughly $23.4M of owner free cash flow, and the diluted share count rose from 103.7M to 107.2M despite $340.1M of buybacks. Third, the competitive position is structurally capped: Elastic relicensed Elasticsearch away from Apache 2.0 in 2021, AWS forked it as OpenSearch, and that fork now sits under the Linux Foundation with 3,000+ contributors — a free, permissively-licensed, hyperscaler-distributed substitute for the core engine that no other company in this peer cohort faces.

The result is a business with a narrow, genuine, demand-side moat (data gravity and switching costs — visible in a 112% net expansion rate that would be sub-100% without them) that is nevertheless subscale in all three of its end markets and unable to convert its position into excess returns. Growth is decelerating — 17.3% delivered in FY26, 14.6% guided for FY27 — and management’s characterization of FY27 as “revenue acceleration” describes an intra-year shape, not the full-year step-down.

Valuation is where the argument gets interesting. At $65.79 the enterprise value is ~$6.11B (correcting for the $601.5M of marketable securities that standard screens omit), or 3.07× forward sales and ~14.3× forward company-defined free cash flow — the 6.8th percentile of Elastic’s own valuation history and the cheapest multiple in a peer group that includes Dynatrace (~5.8×), GitLab (~3.6× forward) and Datadog (~17–18×). On the reported metrics the stock is inexpensive. On owner economics — ~261× EV/owner-FCF — it is not cheap at all. The entire investment case therefore reduces to a single line item: whether stock-based compensation falls as a share of revenue while growth holds in the low-to-mid teens. Everything else is commentary.

No recommendation and no price target appears in this section or anywhere in the analysis that follows. The single exception is the clearly-labeled Claude’s Take block above.

2. Business Overview

2.1 What Elastic actually sells

Elastic’s product is a distributed document store and search engine — Elasticsearch — wrapped in a commercial platform. The company was founded in 2012 around the open-source Elasticsearch project (itself descended from Shay Banon’s Compass, and built on Apache Lucene), IPO’d on the NYSE in October 2018, and is incorporated in the Netherlands.

The technical stack is coherent and, in engineering terms, well-regarded:

• Elasticsearch — the engine. A document store and search/analytics engine handling textual, numerical, geospatial, structured and unstructured data, and — since the AI cycle — dense and sparse vectors.
• Kibana — the visualization, management and configuration interface.
• Logstash / Beats / Elastic Agent — the ingest layer: data pipelines and lightweight shippers that move data from edge machines into the store.
• Elasticsearch Relevance Engine (ESRE) — the retrieval layer combining classical text search with vector/AI retrieval and LLM integration.

On top of that engine, Elastic sells three solutions, which is how it goes to market and how the market thinks about it:

2.2 How it makes money

Revenue is overwhelmingly subscription: $1,634.5M of $1,739.3M (94%) in FY2026, with services at $104.9M (6%). Subscriptions are sold on resource-based pricing, and all tiers have access to varying levels of every solution — a deliberate choice that makes cross-solution expansion the primary growth engine rather than new-logo acquisition.

Deployment splits two ways, and the split matters enormously to the reported numbers:

• Elastic Cloud (Hosted + Serverless) — Elastic runs the deployment on AWS, Azure or GCP across 55+ public-cloud regions. 48% of FY2026 revenue (46% FY25, 43% FY24, 40% FY23, 35% FY22, 27% FY21). Grew +22% in FY26; the “Annual Elastic Cloud” component grew +28%.
• Elastic Self-Managed — the customer runs it, on-premises or in their own cloud account. The residual 52%, growing +14%.

The revenue-recognition asymmetry between the two is the single most under-appreciated mechanical feature of this P&L. Self-managed licence revenue is recognized partly upfront when the licence is delivered, with the remainder ratable. Cloud revenue ramps over the contract term, recognized on consumption or ratably. The consequence: a quarter in which the sales force closes an unusually cloud-heavy mix of commitments reports lower in-quarter revenue while building a larger forward book. Management stated exactly this about Q4 FY2026, attributing the cloud-heavy mix to US federal civilian agencies adopting Elastic-powered SIEM-as-a-Service through the Cybersecurity and Infrastructure Security Agency (CISA). This is a real effect, not an excuse — but it is also unfalsifiable from outside, and it is the sort of explanation that is easy to reach for. We treat it as Interpretation, corroborated by the RPO acceleration.

2.3 Customers and concentration

The customer base is broad at the bottom and concentrating at the top — the classic land-and-expand shape:

Two things stand out. The $100K cohort has decelerated from 30%-plus growth to a stable but pedestrian ~14%. And net expansion has been flat at 110–112% for three consecutive years after resetting down from 117%. Expansion is the engine of this model; a flat, low-teens expansion rate is the arithmetic ceiling on organic growth absent new-logo acceleration.

Concentration deserves a flag: one channel partner accounted for 11% of total revenue in FY2026 (12% in FY2025) and 11% of net accounts receivable at year-end. Elastic does not name it; the economics strongly suggest a cloud-marketplace or global-distribution relationship. A double-digit revenue concentration in a single intermediary is a real, if second-order, risk — it is a channel, not an end customer, so the underlying demand is diversified, but the commercial terms are not.

Geographically, 46% of revenue is outside the United States (44% FY25, 42% FY24) — a steadily internationalizing mix that adds currency translation noise to every reported growth rate (FY26 Q4: +16% reported, +14% constant currency).

2.4 Verdict

A coherent, genuinely recurring, capital-light business model. 94% subscription, 76% gross margin, $5.1M of annual capital expenditure against $1.74B of revenue, and a growing float of customer prepayments. The model is not the problem. The two structural features to carry forward are (a) the cloud/self-managed recognition asymmetry, which makes any single quarter’s revenue a poor signal, and (b) the flat 112% net expansion rate, which is where the competitive question shows up in the financials.

3. Industry Dynamics

3.1 Three markets, no home

The defining structural fact about Elastic is that it competes in three large, separately-contested markets simultaneously, and leads none of them. This is usually presented by the company as platform breadth. It is more accurately described as a three-front war fought by a $6.1B enterprise value against opponents who are, in each theatre, larger and more focused.

Search and AI context. This is Elastic’s ancestral home and the market where its position is strongest. The rise of retrieval-augmented generation has made “where does the LLM get its context” a first-order enterprise question, and Elastic’s answer — the data is already in Elasticsearch, so bring the model to the data — is genuinely compelling. Management reports over 600 customers with >$100K ACV now using Elastic’s AI capabilities, more than one-third of that cohort. Against that: the managed vector-database segment is dominated by Pinecone (roughly 70% share of a ~$2.55B market growing ~22% annually), and the deeper threat is that vector search is becoming a feature rather than a product — pgvector inside Postgres, Atlas Search inside MongoDB, native retrieval inside Databricks and inside every hyperscaler’s AI stack. Elastic’s advantage here is real (it retains a measurable edge on filtered vector queries, where metadata filtering and approximate-nearest-neighbour search must execute together — the workload that actually matters in production RAG) but it is an engineering advantage in a commoditizing layer.

Observability. Elastic is a credible #3 or #4. Our prior work on Datadog found that Datadog “wins consolidation against point tools (Dynatrace, New Relic, Splunk/Elastic) repeatedly” — Elastic appears in that sentence as one of the point tools being consolidated against. Datadog grows at 32% with 80% gross margins; Dynatrace runs a 29% non-GAAP operating margin; Grafana/Prometheus is free and ubiquitous; and CloudWatch/Azure Monitor are bundled. Elastic’s log-analytics cost position is a genuine selling point — it is materially cheaper per ingested terabyte than Datadog — but competing on price in a market with a free open-source alternative and two better-capitalized commercial leaders is a structurally poor place to stand.

Security / SIEM. This is where Elastic has had its most visible recent wins, and the CISA SIEM-as-a-Service relationship is a legitimate flagship — US federal civilian agencies migrating off incumbent SIEMs onto an Elastic-powered service is exactly the kind of reference that compounds. But the opponents are formidable: Cisco/Splunk owns the installed base, Microsoft Sentinel is bundled into E5, CrowdStrike and Palo Alto (XSIAM at >$600M ARR growing 100%) are both attacking with far larger security-native distribution.

3.2 The capital-cycle read

Applying the Marathon lens: infrastructure and observability software is in the late stage of a capital-attraction cycle. A decade of near-free capital produced extraordinary supply-side entry — dozens of observability vendors, dozens of vector databases — and high returns duly attracted the capital that competed them away. The sector’s de-rating is the market recognizing that: Dynatrace at the 5.9th percentile of its own valuation history, GitLab at the 7.8th, Elastic at the 6.8th. These are not idiosyncratic stories; they are one story told three times.

The Marathon signal to watch is supply withdrawal — consolidation, shutdowns, headcount reduction, and the return of pricing discipline. That is now visible. Elastic’s own ~7% workforce reduction (announced 2026-06-23, framed as “evolving its organization to better align its teams with working in an age of AI automation”) is one data point; the sector-wide pivot from growth-at-all-costs to Rule-of-40 discipline is the broader one. For surviving incumbents with intact pricing, the late capital cycle is when returns normalize upward.

But here is the wrinkle that specifically disadvantages Elastic. Supply withdrawal works because competitors exit when capital dries up. OpenSearch will not exit. A Linux Foundation project with 3,000+ contributors and 400+ contributing organizations does not shut down for lack of venture funding; AWS does not stop distributing it because returns compressed. The free substitute is structurally immune to the capital cycle. So the cyclical tailwind that should lift Dynatrace and Datadog lifts Elastic less, because Elastic’s most durable competitive constraint is not a funded competitor.

3.3 Regulation and structural factors

There is no meaningful sector-specific regulation constraining Elastic’s economics — no reimbursement regime, no rate regulation, no licensing bottleneck. Three secondary regulatory factors are worth noting: (a) public-sector procurement is becoming a genuine growth channel (CISA, FedRAMP-adjacent requirements), which favours vendors able to support air-gapped and self-managed deployment — a real Elastic advantage over pure-SaaS competitors; (b) data-residency and sovereignty rules in the EU similarly favour the hybrid/self-managed model; and © evolving international trade policy is explicitly cited in the 10-K as a factor that could shift mix from cloud back to self-managed in affected geographies.

3.4 Verdict

A structurally mediocre position in three structurally attractive markets. The profit pools are large and growing, and AI is a genuine demand tailwind rather than a narrative one. But Elastic occupies the contested middle of all three markets rather than owning one, the engine layer beneath all three is subject to free substitution, and the capital-cycle recovery that should benefit the sector benefits Elastic less than its proprietary peers. Structurally bad position; cyclically improving conditions.

4. Competitive Position

4.1 The licensing decision and its consequences

This section is the heart of the analysis, and it begins in January 2021.

Elasticsearch was released under Apache 2.0 — a permissive open-source licence. Amazon Web Services offered a managed Elasticsearch Service built on that code, competing directly with Elastic Cloud while, in Elastic’s view, contributing little upstream. In January 2021 Elastic relicensed Elasticsearch and Kibana to a dual SSPL / Elastic License v2 model, neither of which is OSI-approved open source, with the explicit purpose of preventing hyperscalers from offering the software as a service.

AWS responded within three months by forking Elasticsearch 7.10.2 into OpenSearch under Apache 2.0. In September 2024 AWS donated OpenSearch to the Linux Foundation, creating the OpenSearch Software Foundation with a multi-vendor technical steering committee — a governance change that materially reduced the “it’s just an AWS project” objection and broadened contribution. In August 2024 Elastic added AGPLv3 back as a licence option, an implicit acknowledgement that the 2021 decision had cost it open-source goodwill. As of 2026, OpenSearch is at version 3.6.0 (released April 2026) with 400+ contributing organizations and 3,000+ active contributors.

Why this matters more than any other single competitive fact: Elastic is the only company in its peer cohort whose core engine has a free, permissively-licensed, foundation-governed, hyperscaler-distributed, drop-in-compatible substitute. Datadog’s agent is proprietary. Dynatrace’s OneAgent is proprietary. MongoDB is SSPL-licensed but has no comparably-backed fork. A prospective Elastic customer evaluating log analytics can, at zero licence cost, deploy OpenSearch — or simply click “Amazon OpenSearch Service” in the AWS console.

4.2 The counterweight — Elastic did not lose the engineering race

It would be lazy to stop there, and the evidence does not support the maximal bear case. Since the April 2021 fork, commit activity has run roughly 7:1 in Elasticsearch’s favour — approximately 41,000 commits to elastic/elasticsearch versus ~5,800 to opensearch-project/OpenSearch as of May 2026. Elasticsearch retains a meaningful performance edge in filtered vector search, and Elastic has shipped the AI-era features first: ESRE, the Jina V5 omni family for multimodal search, Agent Builder (now generally available), Cross Project Search on serverless, and MCP applications for security and observability that embed Elastic-powered workflows directly into Claude, VS Code and similar tools.

The honest reading: the fork constrained Elastic’s pricing power; it did not take Elastic’s technical lead. Those are different injuries. The first is chronic and permanent; the second would have been fatal and has not happened.

4.3 Naming the moat under the Greenwald taxonomy

Applying Competition Demystified rigorously — barriers to entry are what matter, and there are only three genuine kinds:

Supply-side / cost advantage: ABSENT. Elastic’s gross margin is 76.1% — the lowest in its peer cohort (Datadog ~80%, GitLab 87–89%, Dynatrace ~82%). The reason is structural and disclosed: Elastic Cloud carries third-party hosting cost paid to the very hyperscalers it competes with, and the 10-K states plainly that continued cloud mix shift “will continue to have a modest unfavorable impact on our gross margin.” Elastic has a negative cost advantage relative to peers, and it rents its infrastructure from its competitors.

Economies of scale plus customer captivity: ABSENT. This is the advantage that produces durable excess returns, and it requires being large relative to the served market. Elastic is subscale in all three of its markets. It cannot amortize R&D over a larger base than Datadog in observability, than Splunk or Microsoft in SIEM, or than the combined vector-database field in AI retrieval. Spending $451.9M on R&D (26% of revenue) is a large absolute number and a small relative one against those opponents.

Demand-side / customer captivity: PRESENT, MODERATE. This one is real. Petabyte-scale search indices sitting in production ingest paths are genuinely expensive and risky to migrate: the data has gravity, the queries are embedded in applications, the dashboards are embedded in operational runbooks, and the on-call engineers know Kibana. The financial fingerprints are visible — 112% net expansion, 21.4% deferred-revenue growth, $807.9M of FY26 revenue recognized out of the opening deferred balance, and a >$5M-ACV customer cohort growing 30%. Customers who commit deepen; they do not churn casually.

4.4 The moat test

The test we apply: if a moat claim cannot be tied to a financial outcome that would deteriorate without it, it is not a moat.

Elastic passes — narrowly. Given that a free, functionally-comparable, foundation-governed substitute exists and is one click away inside the largest cloud, a net expansion rate of 112% is only explicable by genuine switching costs. Absent them, a rational cost-minimizing CIO facing a renewal would migrate, and expansion would be below 100%. The moat is doing measurable work.

But observe what it is not doing. Elastic has never produced a GAAP operating profit in eight years as a public company. FY2026: −$33.5M. FY2025: −$54.7M. FY2024: −$125.0M. FY2023: −$187.9M. Return on invested capital is not meaningfully positive on any undistorted basis. A moat that cannot generate an accounting return over an entire business cycle is, by our standards, a moat that is just wide enough to prevent value destruction and not wide enough to create value.

Compare the market-share-stability test: stable shares indicate a moat, churning shares indicate none. Elastic’s share is not collapsing — it is grinding forward at 14% customer growth and 112% expansion. But it is also not consolidating the way a genuine leader’s does (Datadog at +32%, Palo Alto’s XSIAM at +100%).

4.5 Head-to-head against the cohort

(Peer figures are drawn from each company’s own public filings and earnings releases as of June–July 2026 and are as-of those dates; they are directional cross-checks, not synchronized comparables.)

The table tells a consistent story: Elastic is the cheapest name in the cohort and also the weakest on almost every quality metric that matters — lowest gross margin, only one still posting GAAP operating losses, thinnest net-cash cushion relative to market capitalization, and a net expansion rate that is merely in line with the most-de-rated peer rather than ahead of it. The discount is directionally earned. The investment question is one of degree.

4.6 Verdict

A narrow, genuine demand-side-captivity moat built on data gravity and switching costs — sufficient to hold the installed base, insufficient to earn an operating return. No cost advantage (the worst gross margin in the peer set, with infrastructure rented from competitors). No scale advantage (subscale in all three markets). And a free, Linux-Foundation-governed substitute for the core engine that structurally caps pricing at the layer where Elastic’s technology is most differentiated. This is a contested market with real but modest differentiation, not a durable franchise.

5. Growth History and Forward Opportunities

5.1 The historical record

Two trends run in opposite directions and define the company. Growth has decayed monotonically from 70% to 17%, and is guided to 14.6%. Losses have narrowed monotonically from −40% to −1.9% of revenue. Elastic has spent eight years trading growth for margin, and has not yet arrived at profitability — it is projected to cross into GAAP operating profitability in FY2027, nine years after listing.

Growth has been overwhelmingly organic. Cumulative cash spent on acquisitions across the period is modest — $24.4M (FY20), $119.9M (FY22), $19.1M (FY24), $36.8M (FY26) — against $356.4M of goodwill. There is no roll-up here and no acquired-revenue illusion. That is a genuine mark of quality: the 10.9× revenue increase since FY2018 was built, not bought.

5.2 Where the growth comes from now

The composition has shifted decisively toward expansion within existing accounts and toward cloud:

• Elastic Cloud +22%, self-managed +14%; cloud is now 48% of revenue and rising ~2 points a year.
• “Annual Elastic Cloud” +28% — the committed (rather than on-demand) portion, which is the higher-quality half.
• Customers >$5M ACV +30%; >$1M ACV +14% (240+, with 30+ net additions); >$100K ACV +13.9% (1,720).
• Net expansion 112%, flat for three years.

The clear pattern: growth is increasingly concentrated in the largest accounts. The >$5M cohort grows at 30%, the >$1M at 14%, and the >$100K at 14%. Elastic is going up-market successfully — that is a legitimate strategic achievement — but it also means growth is becoming more dependent on fewer, larger relationships, and it means the mid-market land motion is not contributing much.

5.3 The backlog signal

The most genuinely encouraging datum in the FY2026 file is the divergence between backlog growth and revenue growth:

Contracted future revenue is growing at roughly 1.6× the rate of recognized revenue, and the long-dated portion at 2.5×. Management attributes this to a step-change in multi-year commitments, and stated that these were secured “without any material change in our discount practices.” If that is true, it is a strong forward signal: customers signing longer contracts at unchanged prices is precisely what deepening switching costs look like.

The skeptical reading, which must be stated: RPO growth of 28% against revenue growth of 17% can also be produced by a shift in contract duration — persuading customers to sign three-year rather than one-year deals inflates total RPO immediately without adding a dollar of annual spend. Non-current RPO growing 43% is consistent with exactly that. The discriminating evidence would be an acceleration in CRPO (the 12-month portion), which did occur — from 15% constant-currency in Q3 to 20% in Q4 — and that is genuinely supportive. But one quarter of CRPO acceleration is not yet a trend, and management’s own FY27 revenue guide of +14.6% implies they do not expect the backlog to convert into accelerating full-year revenue.

5.4 Forward opportunities

AI context / RAG (the real one). Over 600 customers with >$100K ACV now use Elastic’s AI capabilities — more than one-third of that cohort, up meaningfully year on year, and including 40+ serverless customers newly captured. The strategic logic is sound: as enterprises deploy agentic systems, the retrieval layer becomes load-bearing, and moving petabytes to a purpose-built vector store is unattractive versus adding vector search to the store the data is already in. Elastic’s published claim of a 70% reduction in tokens used versus naive RAG, if it generalizes, is a concrete economic argument. This is the highest-value optionality in the story.

Public sector. The CISA SIEM-as-a-Service relationship is expanding across civilian agencies displacing incumbent security vendors. Federal SIEM is a large, sticky, reference-rich pool, and Elastic’s ability to serve air-gapped and self-managed deployments is a genuine structural advantage against pure-SaaS competitors.

Serverless. Elastic Cloud Serverless removes cluster-sizing friction and should widen the funnel below the enterprise tier. Too early to size.

Security consolidation. Displacing legacy SIEM on cost-per-ingested-terabyte is Elastic’s most repeatable enterprise sales motion, and management called Q4 “a record Q4 for $1 million deals.”

5.5 Verdict

Medium-quality growth, honestly earned but decelerating. The positives are substantive: entirely organic, increasingly enterprise, with a backlog growing at 1.6× the rate of revenue and a genuine AI demand tailwind that is showing up in customer counts rather than only in slideware. The negatives are equally substantive: growth has fallen from 70% to a guided 14.6% with no inflection yet visible in reported revenue, net expansion has been flat at 110–112% for three years, and the $100K customer cohort has grown at almost exactly 14% for three consecutive years. This is a company growing in line with its market, not taking share from it.

6. Financial Quality

6.1 The headline that isn’t

Elastic reported net income of $367.8M for FY2026 — its largest ever, on a 21.1% net margin. The number is arithmetically correct and economically meaningless. Here is the bridge:

The pre-tax result was a $2.3M loss. The entire reported profit is a tax benefit, and the tax benefit is the release of deferred-tax valuation allowances: $390.5M in the Netherlands, $23.7M in the United Kingdom, $20.7M in California — $434.9M gross.

The 10-K does not hide this. Its Risk Factors state, in terms: “Although we had net income of $367.8 million and $61.7 million for the years ended April 30, 2026 and 2024, respectively, we would have incurred net losses in such years as well without the releases of valuation allowances against deferred tax assets.” The CFO quantified it on the Q4 call as “a onetime benefit of $435 million to our GAAP net income” that “did not impact any of our operating results, non-GAAP diluted earnings per share, adjusted free cash flow or cash and cash equivalents.” Management’s disclosure here is exemplary; the point is that the screens do not read the disclosure.

Two further observations sharpen this. First, this is the second time: FY2024’s reported net income of $61.7M sat on a −$184.5M tax benefit against a −$125.0M GAAP operating loss. Elastic has now produced GAAP “profitability” twice, both times out of the tax line. Second, and more substantively, the basis for the largest component is unusual. The UK release rested on the conventional test — three years of cumulative income. But the 10-K states that “the release of the valuation allowance in the Netherlands and California was supported by the implementation of a committed tax planning action in fiscal 2027 that is expected to generate future taxable income in each jurisdiction.” In plain English: the $411M Dutch-plus-Californian release is predicated on a prospective internal restructuring that has not yet happened, forecast to generate taxable income that does not yet exist. That is a defensible position under ASC 740 if the planning action is genuinely committed and prudent-and-feasible — but it is a materially softer basis than demonstrated profitability, and it creates a live risk of partial re-establishment if the action does not execute as planned.

Practical consequence for valuation screens: trailing GAAP EPS of ~$3.49 and the resulting 18.4× P/E are artifacts. Any percentile, multiple or comparison built on Elastic’s GAAP earnings is invalid. The usable valuation reads for this name are price/sales and price/book.

6.2 The second headline that isn’t

Elastic generated $326.9M of operating cash flow and $321.8M of free cash flow in FY2026 — an 18.5% FCF margin, or the company-defined “~20% adjusted free cash flow margin.” Capital expenditure was $5.1M, which is genuinely remarkable capital intensity: 0.3% of revenue. The cash is real, it is in the bank, and the balance sheet shows it.

But stock-based compensation was $298.4M — $308.2M including related employer taxes, or 17.7% of revenue. The 10-K’s own percentage table shows SBC at 18% of revenue in FY2026 and FY2025, and 20% in FY2024.

SBC equals 92.7% of free cash flow. This is the single most important number in the file. It means the reported cash generation is very nearly a transfer from shareholders to employees routed through the share count, and that the “20% FCF margin” and the “1.3% owner margin” describe the same year.

This is not unique to Elastic — our prior work found GitLab’s SBC at approximately 100% of its FCF and Datadog’s at ~22% of revenue, and flagged the same issue in both. It is endemic to the category. But two things make it more consequential here than at Datadog: Elastic’s reported FCF margin is lower to start with (18.5% vs ~27%), and Elastic’s growth rate is lower (17% vs 32%), so there is less prospect of out-growing the charge.

Is SBC a real cost? Yes, unambiguously, and the share count proves it. Weighted diluted shares went 103.7M (FY25) → 107.2M (FY26), and FY27 guidance is 107.5–108.5M — despite $340.1M of buybacks in FY2026. Elastic spent more on repurchases than the entire SBC charge and the share count still rose. If SBC were not a real economic cost, that would not happen.

6.3 Margin structure and operating leverage

Operating leverage is real but slow. Sales and marketing plus G&A has fallen 9.5 points over four years, and gross margin has risen 3 points despite the cloud mix shift — a genuinely good outcome, since the 10-K warns cloud mix should pressure it. R&D ticked up in FY2026 (24.7% → 26.0%), which management frames as investing into the AI opportunity. The uncomfortable arithmetic: after four years of disciplined expense reduction, the GAAP operating margin is still negative.

Note the quarterly detail, which is less flattering than the annual: GAAP operating income by quarter in FY2026 was −$9.4M, −$8.2M, +$0.6M, −$16.4M. The company had exactly one marginally GAAP-profitable quarter and then went backwards in Q4. Sequential revenue in the back half was $449.9M → $450.7M — flat.

6.4 Balance sheet

The balance sheet is sound but not fortress-like. Net cash of ~$779M is ~11% of the market capitalization — meaningful, but well short of GitLab’s ~25% or Dynatrace’s no-funded-debt position. The single debt instrument, $575M of 4.125% senior notes, does not mature until July 2029 and costs ~$25M a year against $1.37B of liquid assets; there is no refinancing risk on any relevant horizon. The notes trade below par ($545.9M fair value versus $575M principal), which is a mild credit-market signal worth noting but not alarming for a company with net cash.

A methodological caution that matters here: standard screens — including ROIC.ai’s published enterprise value — net only cash and equivalents ($768.7M) against debt, ignoring the $601.5M of short-term marketable securities. That understates net cash by $601.5M and overstates enterprise value by the same amount. Every valuation figure in the valuation section uses the corrected $6.11B.

6.5 Working capital and revenue quality

Receivables grew to $464.4M from $375.6M (+23.6%) against revenue growth of 17.3% — running ahead, which bears watching, though the +11% concentration in a single channel partner’s receivable balance partly explains lumpiness. Days sales outstanding of roughly 97 days is high but typical for enterprise software with annual invoicing.

Two positives on revenue quality. Deferred revenue grew 21.4%, ahead of revenue — the customer float is expanding, which is the opposite of a pull-forward. And deferred contract acquisition costs were a net drag: $163.7M capitalized against $111.1M amortized, a −$52.6M headwind to operating cash flow. Elastic is not using commission capitalization to flatter its cash flow this year; if anything the sales-capacity build is suppressing reported FCF relative to the underlying run rate.

6.6 Returns on capital

Return on invested capital is not meaningfully computable on a basis that means anything, and we decline to publish a flattering one. The FY2026 ROE of ~29% that screens will report is the tax benefit divided by equity. On an operating basis, NOPAT is negative — a −$33.5M operating loss against ~$1.87B of total capital. Elastic has not earned its cost of capital in any year of its public life. Stated plainly: this is a business that has consumed capital for eight years and is only now approaching the break-even line.

6.7 Verdict

Economics improve with scale, but far too slowly, and the reported profitability is not real. The genuine positives: 76.1% gross margin and rising, 0.3%-of-revenue capital intensity, a growing deferred-revenue float, no commission-capitalization games, and $779M of net cash. The disqualifying negatives for anyone underwriting reported figures: no GAAP operating profit in eight years, a headline net income that is 100% a valuation-allowance release predicated partly on a future tax restructuring, and free cash flow that is 92.7% funded by stock compensation, leaving ~1.3% of revenue as owner cash. The correct summary is that Elastic is a business at the threshold of genuine profitability that has not yet crossed it, being described as though it already has.

7. Capital Allocation

7.1 The record

M&A: restrained, and to management’s credit. Cumulative cash spent on acquisitions since FY2020 is roughly $200M — $24.4M (FY20), $119.9M (FY22), $19.1M (FY24), $36.8M (FY26) — against $356.4M of goodwill and $1.74B of revenue. There is no serial-acquirer roll-up, no large platform deal at a bubble multiple, and no acquired-revenue illusion inflating the growth rate. In a sector littered with value-destructive 2021-vintage acquisitions, not doing anything stupid is a genuine achievement and should be scored as such.

Debt: conservative and well-timed. One issuance — $575M of 4.125% senior notes in July 2021, at the bottom of the rate cycle, due 2029. That is close to optimal timing. Interest cost is ~$25M a year against $1.37B of liquid assets.

Buybacks: new, well-timed, and structurally neutralized. In October 2025 the board authorized Elastic’s first-ever repurchase program, $500M with no expiration. In FY2026 the company repurchased $340.1M — approximately 4.4 million shares, using ~68% of the authorization, ahead of its stated goal of half in FY26. Q4 alone was ~$40M for ~650,000 shares, an average of roughly $61.5. Management’s stated policy is to return 50% of free cash flow via repurchases unless acquisitions require the cash.

The timing deserves credit. Elastic is buying back stock at the 6.8th percentile of its own valuation history, at an average price near multi-year lows. This is the opposite of the 2021-vintage software buybacks executed at 15–20× sales. If you are going to repurchase shares, this is when.

7.2 The treadmill

And yet the arithmetic is sobering:

Elastic spent more on buying back stock than its entire stock-compensation charge, and the diluted share count still rose 3.4%. The repurchase program is not a return of capital; it is a dilution-offset. Without it, dilution would run roughly 4% annually; with it, roughly 0.7–1%. That is a worthwhile improvement — compounding 1% dilution instead of 4% is worth a great deal over a decade — but it must be described accurately. No capital is being returned to shareholders. Cash is being converted into the maintenance of a roughly-constant share count.

7.3 Incentives — the governance heart of the matter

This is where the analysis becomes uncomfortable, and it explains a great deal about the preceding sections.

Per the FY2025 proxy statement, the executive annual cash-incentive plan was weighted 35% Total Revenue and 30% Total Cloud Revenue. The performance share unit plan’s performance metric was total revenue, with attainment scaled 80%/100%/120% of plan producing 50%/100%/200% payout. FY2025 outcomes: Cloud revenue $688M against a $669M target (102.8% attainment, 107.1% payout); Total revenue $1,483M against a $1,485M target (99.9%). One hundred percent of target PSUs were earned for FY2025; 97% for FY2024.

There is no return-on-capital metric. No free-cash-flow-per-share metric. No margin metric in the PSU plan. And — critically — no share-count or dilution metric anywhere.

This is the incentive structure that produces exactly what Elastic has produced: revenue compounding at 35% annually since FY2018, stock compensation at 18–20% of revenue, a diluted share count up from 61.9M (FY18) to 107.2M (FY26) — a 73% increase — and a share price 6% below its 2018 IPO-day close. Management was paid, correctly and as designed, for growing revenue. Shareholders were not paid at all. The gap between those two outcomes is the definition of a capital-allocation problem, and it is written into the compensation plan.

The October 2025 special CEO award is a material and genuine correction. The board granted Ashutosh Kulkarni a one-time PSU award with a grant-date fair value of $29.3M, up to 456,491 shares, over a five-year term, structured as follows:

Hurdles are measured on a 30-consecutive-trading-day average, modified by relative total shareholder return against the Russell 3000 median (a 20% reduction and forfeiture if below median), and gated on continuous service as CEO for four to five years. The award is genuinely 100% at risk.

This is well-designed pay — share-price-gated, rTSR-modified, long-dated, heavily back-weighted to the hardest hurdle. It is precisely the metric that was missing. But note where it stands. The $86.61 baseline was the 30-day average from 2025-08-29 to 2025-10-10. The stock is $65.79 — 24% below the baseline — and would need to appreciate +85% from here merely to reach Tranche 1, and +201% to reach the 40% tranche. The award is deeply out of the money, which cuts two ways: alignment is maximal (Kulkarni earns nothing without a near-doubling), and retention risk is elevated (an award this far underwater has limited holding power).

7.4 Verdict

Mixed, and improving from a poor base. The affirmative case: no value-destructive M&A across eight years, conservatively timed debt, and a first-ever buyback initiated at genuinely depressed prices with a sensible 50%-of-FCF policy. The case against: a decade of paying executives on revenue and cloud revenue while issuing 18–20% of revenue in stock, a diluted share count up 73% since the IPO, buybacks that offset dilution rather than returning capital, and a share price below its first-day close. Management has demonstrated that it allocates capital to growth. It has not yet demonstrated that it allocates capital to the share. The October 2025 CEO award is the first structural evidence that the board understands the distinction.

8. Changes and Headwinds — Last Two Years

FY2025 (May 2024 – April 2025) — the self-inflicted wound. The defining event was the Q1 FY2025 print on 2024-08-29, when Elastic cut guidance after a botched sales-territory re-segmentation disrupted the field organization. The stock fell 26.5% in a single session ($103.64 → $76.19). This was not a demand event; it was an execution event, and it is the principal reason the market applies a discount to management’s forward commentary. The recovery through FY2025 (+14.8% on the Q2 print, +14.9% on Q3) was the market re-underwriting execution, before the Q4 FY25 print cost another 12.1%.

October 2025 — the capital-allocation pivot. Within four days the board authorized the first buyback in company history ($500M, 2025-10-09) and granted the CEO a $29.3M share-price-gated PSU award (2025-10-13). Taken together with the Financial Analyst Day held the same month — at which the 50%-of-FCF return policy and the Rule-of-40-by-FY29 target were laid out — this was a deliberate repositioning of Elastic from a growth story to a capital-returns story. It is the most important governance development in the file.

November 2025 and February 2026 — two deceleration scares. The Q2 FY26 print (2025-11-20) cost 14.7%; the Q3 FY26 print (2026-02-26) cost 15.4%, with constant-currency CRPO growth of 15% read as a slowdown. The stock fell from $94.47 (November) to an all-time low of $43.30 on 2026-04-10.

May 2026 — the FY2026 close and the tax release. The Q4 print (2026-05-28) beat across the board: FY26 revenue +17.3%, non-GAAP operating margin 16.4% (up 120bp), adjusted FCF margin ~20%, RPO +28% and non-current RPO +43%. The stock rose 12.3%. The same print contained the $435M valuation-allowance release that produced the illusory GAAP profit, and FY2027 guidance of +14.6% — a deceleration that management framed as intra-year “acceleration.”

June 2026 — the restructuring. On 2026-06-23 Elastic committed to a plan to reduce its workforce by approximately 7% (8-K Item 2.05, filed 2026-06-24), described as “evolving its organization to better align its teams with working in an age of AI automation,” with continued hiring in customer-facing go-to-market roles. Against a base of 4,019 employees this implies roughly 280 positions. Headcount had grown 13.6% in FY2026 (3,537 → 4,019), so this partially reverses a single year’s hiring. The charge has not yet been quantified and will appear in Q1 FY2027 results on 2026-08-27.

Persistent headwinds through the period. (1) OpenSearch’s institutionalization — the September 2024 Linux Foundation donation converted a vendor fork into a neutral industry project, materially strengthening it as a substitute. (2) AI-disruption sentiment — the market’s 2026 anxiety that general-purpose LLMs and agentic systems erode the value of dedicated data-layer software, which drove the sector-wide de-rating that took DT, GTLB and ESTC all to their cheapest-ever multiples. (3) Currency — with 46% of revenue offshore, reported growth has diverged from constant-currency growth by up to 2 points.

8.1 Verdict

On balance these developments strengthen the shareholder position while weakening the growth thesis. The capital-allocation pivot (buyback, CEO award, 50%-of-FCF policy, Rule-of-40 target) is a real, durable improvement in how the business is run for owners. The restructuring is evidence of cost discipline arriving. But the two deceleration scares, the flat sequential revenue in the back half of FY2026, and a FY2027 guide of +14.6% are evidence that the growth engine is downshifting faster than the margin engine is upshifting. The 2024 execution failure remains the reason to discount forward guidance.

9. Risk Analysis

The two risks that actually matter are #1 and #2, and they are related. The free substitute caps what Elastic can charge; the stock-compensation burden determines whether what it does charge reaches shareholders. A reader who wants to hold one thought about the downside should hold this one: Elastic’s competitive position prevents it from raising price, and its cost structure prevents its cash flow from becoming the owners’. Risk #3 is the one the market trades on; risks #1 and #2 are the ones that determine the terminal value.

10. Valuation Discussion — Embedded Expectations

No price target and no recommendation appear in this section. The subjective view is confined to the labeled Claude’s Take block.

10.1 The enterprise-value bridge

A note on method. Standard data providers — including ROIC.ai — compute Elastic’s enterprise value netting only cash and cash equivalents, ignoring the $601.5M of short-term marketable securities. That produces an EV of ~$6.71B and overstates every multiple by roughly 10%. The corrected figure is used throughout.

10.2 The multiple set

The GAAP P/E and its 13.1st percentile must be discarded for the reasons set out in the Financial Quality section — the earnings are a tax artifact. When GAAP EPS is distorted, the P/B and P/S percentiles are the valid reads, and they are 2.0th and 5.4th respectively. On its own eight-year history, Elastic has essentially never been cheaper.

10.3 What the price requires to be right

This is the analytically important question, and the answer is narrower than it first appears.

At 3.07× forward EV/sales and ~14.3× forward company-defined free cash flow, the market is not pricing a growth story. It is pricing something close to a mid-teens grower with real cash flow — a fair, unexcited multiple. If Elastic simply delivers its own FY2027 guide (+14.6% revenue, 19% non-GAAP operating margin, 21.5% adjusted FCF margin) and holds that shape, today’s price is not demanding.

But the entire case runs through one line item. At 18% of revenue, stock-based compensation is the whole distance between a 20%-FCF-margin business and a 1.3%-owner-margin business. The valuation therefore embeds a specific, checkable assumption: that SBC falls materially as a share of revenue while growth holds in the low-to-mid teens. Consider the sensitivity on FY2027E revenue of $1,992.5M:

(Adjusted FCF held at the guided 21.5% margin in all rows; SBC varied. Illustrative sensitivity, not a forecast.)

Note that even holding SBC flat at FY26’s ratio, owner FCF improves markedly in FY2027 — from ~$23M to ~$86M — because revenue grows faster than the fixed portion of the charge and because FY2026’s owner FCF was suppressed by the $52.6M net commission-capitalization drag. The direction of travel is favourable. But at 71× owner FCF on an unchanged SBC ratio, the market is paying for the improvement to continue, not merely to begin.

10.4 What the market is pricing correctly

• The deceleration. 17.3% delivered, 14.6% guided, net expansion flat at 112% for three years, $100K-customer growth stuck at ~14%. A low-teens grower does not deserve a growth multiple.
• The absence of GAAP profitability. Eight consecutive years of operating losses is a fact about the business model, not an accounting artifact.
• The free substitute. A permissively-licensed, foundation-governed fork of the core engine is a permanent cap on pricing power and shows up directly in the peer-worst gross margin.
• The three-front subscale position. Being third in three markets is worse than being first in one.

At 3.07× forward sales, none of this looks mispriced. This is a case where the low multiple is substantially explained by the fundamentals.

10.5 What the market may be pricing incorrectly

• The backlog. Total RPO +28% (the fastest in four years) and non-current RPO +43%, against reported revenue growth of 17.3% and a 14.6% guide. Contracted future revenue is compounding at 1.6× the rate of recognized revenue. Constant-currency CRPO accelerated from 15% (Q3) to 20% (Q4). If a meaningful share of that converts, the FY2027 guide is conservative. Counter-argument, which is serious: multi-year contract-duration mix alone can produce this pattern, and non-current RPO growing at 43% is exactly the fingerprint of duration extension.
• The cloud-mix recognition drag. The mechanical asymmetry described in the Business Overview means an unusually cloud-weighted commitment quarter depresses reported revenue while improving the forward book. Q4 FY2026 was such a quarter (the CISA public-sector wins). Reported growth understates commercial momentum to the extent this is genuine.
• The tax asset itself. Set aside the accounting theatre: the released valuation allowances correspond to real net operating losses that will shelter real future cash taxes. Elastic’s cash tax rate should stay low for years. That is worth something and is invisible in a P/S multiple.
• The buyback’s accretion. Repurchasing at an average ~$61.5 — at the 6.8th own-history percentile — is meaningfully accretive per share in a way the same dollars spent in 2021 were not.

10.6 Scenario framework

Illustrative five-year framings, not targets or recommendations.

Bear — the free substitute wins slowly. OpenSearch and hyperscaler-native alternatives continue to cap pricing; net expansion breaks below 105%; growth fades to 8–10%; SBC stays at 18%; the Rule-of-40 target slips past FY2029. On 2.0–2.5× forward sales this implies an enterprise value materially below today’s. Note the stock traded at ~2.3× EV/sales as recently as 2026-04-10.

Base — the guide is roughly right. Growth 13–15%; non-GAAP operating margin marches from 16.4% toward the guided 19% and on to 20–22%; SBC drifts to 14–15% of revenue; GAAP operating profitability arrives in FY2027 and holds; Rule of 40 achieved around FY2029 as management targets. On 3.0–3.5× forward sales, the multiple is roughly unchanged and earnings growth does the work — a mid-teens compounding of enterprise value with the share count roughly flat.

Bull — the backlog converts and the charge normalizes. The +28% RPO and +43% non-current RPO translate into re-accelerating recognized revenue; AI context demand broadens well beyond the current 600 customers; growth returns to 18–20%; SBC falls below 13% as the workforce reduction and AI-driven internal productivity flow through; owner FCF converges toward reported FCF. On 4.5–5.0× forward sales — still a discount to Dynatrace today — this is a substantially higher enterprise value.

10.7 Peer cross-check

Elastic is the cheapest name in the cohort and, on gross margin, growth, GAAP profitability and balance-sheet cushion, the weakest. The relative discount to GitLab (3.07× vs 3.6× for a business growing 23% at 88% gross margins) and to Dynatrace (3.07× vs 5.8× for a business at a 29% operating margin) is directionally justified. The valuation question is not whether Elastic should trade at a discount — it should — but whether a discount of this size to a de-rated peer group is proportionate to the quality gap.

10.8 Verdict

The market is pricing Elastic as a low-teens grower with capped pricing power and unresolved cost structure, and that description is accurate. The cheapness is real and measurable — 6.8th own-history percentile, 3.07× forward sales, ~14.3× forward reported FCF, with $779M of net cash — but it is cheapness that has been earned by genuine structural deficiencies, not conferred by a market error. The single embedded expectation that determines whether today’s price is correct is whether stock-based compensation declines as a share of revenue. Every other variable is second-order.

11. Variant Perception

11.1 The consensus belief

Consensus holds that Elastic is a decent, unspectacular infrastructure-software asset that has permanently lost the growth premium it enjoyed in 2021 and again in the 2023 “Search AI” trade; that it is structurally disadvantaged by OpenSearch and by hyperscaler bundling; that it will grow in the mid-teens and eventually earn a mid-teens operating margin; and that at 3× forward sales this is roughly fairly priced. The factor evidence corroborates that this is the market’s actual positioning rather than merely its stated view: Momentum loading −0.41, Quality loading −0.26, rs_12m −25.2%, one-year Sharpe −0.56, and a five-year annualized return of −15.5%. No one is enthusiastic about this stock.

11.2 The most revealing single datum

Elastic’s Value factor loading is approximately zero (−0.02).

This is the most interesting statistical fact in the file. A stock sitting at the 6.8th percentile of its own eight-year valuation history with $779M of net cash and 14× forward free cash flow ought to load positively on Value. It does not. It loads on Market (+1.40), Cloud Computing (+1.04), and negative LowVolatility (−0.78) — it trades as high-beta cloud-software exposure, full stop.

The interpretation: the market has de-rated Elastic without re-classifying it. It is no longer priced as a growth stock, but it is not yet held by value investors either. It sits in the ownership gap — abandoned by the growth cohort that owned it at $186, not yet acquired by the value cohort that would own it at 14× cash flow. Factor-similar peers confirm the placement: the model groups ESTC with Samsara, Braze, DocuSign, Nutanix, Snowflake, Amplitude, Asana, MongoDB, GitLab and Box — de-rated mid-cap SaaS — and not with Datadog, CrowdStrike or Palo Alto. Both durable value recoveries and durable value traps are drawn from precisely this population. The factor tape tells you where the stock sits; it does not tell you which kind it is.

11.3 The strongest bull case

Elastic is the cheapest way to own the AI data layer, and the backlog is already telling you the demand is there.

The bull argues that reported revenue is a lagging, mix-distorted indicator and the leading indicators have already inflected: total RPO +28% (best in four years), non-current RPO +43%, constant-currency CRPO accelerating from 15% to 20%, multi-year commitments secured “without any material change in our discount practices,” a record Q4 for $1M deals, the >$5M ACV cohort +30%, and 600+ customers now using AI capabilities — a third of the $100K cohort, from a standing start. The strategic logic is sound and getting stronger: as enterprises deploy agentic AI, the retrieval layer becomes load-bearing, and moving petabytes to a purpose-built vector store is economically irrational versus adding vector search where the data already lives. Elastic retains the engineering lead over OpenSearch (7:1 on commits) and a measurable edge on filtered vector search — the workload that matters. Meanwhile the capital-allocation posture has genuinely changed: a first-ever buyback executed near the lows, 50% of FCF committed to repurchase, a ~7% cost reduction, GAAP operating profitability arriving in FY2027, and a CEO whose entire incentive now requires the stock to double. At 3.07× forward sales with $779M of net cash, you are paying nothing for the AI leg and rather little for the security franchise displacing legacy SIEM across the US federal government.

11.4 The strongest bear case

You are buying a company with no operating profit, no owner cash flow, and a free competitor — and calling it cheap because a tax accountant had a good quarter.

The bear argues that the entire “profitable, cash-generative Elastic” narrative is constructed from accounting: GAAP net income is 100% a valuation-allowance release, partly predicated on a future tax restructuring; the operating line is still negative after eight years; and the celebrated 20% FCF margin is 92.7% funded by stock compensation, leaving 1.3% for owners. The share count rose 3.4% in a year in which the company spent $340M on buybacks — shareholders are running to stand still. Underneath the accounting the business is decelerating: net expansion flat at 112% for three years, $100K-customer growth stuck at 14% for three years, back-half FY2026 revenue flat sequentially, and a FY2027 guide of +14.6% that management describes as “acceleration.” Worst of all, the competitive geometry is unfixable: Elastic gave away its engine in 2021, AWS forked it, the Linux Foundation now governs it, and the fork is one click away inside the largest cloud on earth — which is precisely why Elastic earns the lowest gross margin in its peer group while renting infrastructure from its own competitors. It is subscale against Datadog, subscale against Microsoft and Splunk, and subscale against Pinecone and pgvector. A cheap multiple on a business that has never earned its cost of capital is not value; it is a value trap with a good story about AI.

11.5 The five assumptions that actually matter

11.6 Where we think consensus may be offsides

Consensus is correct on the structural critique and may be over-extrapolating the growth deceleration. Three specific points:

First, the factor evidence suggests positioning is washed out rather than crowded. A −0.41 Momentum loading, a −0.56 one-year Sharpe, and a −64.8% drawdown from peak describe a stock that has already been sold by everyone inclined to sell it. That is a materially different setup from the crowded-long condition we flagged at Datadog and JFrog. Washed-out positioning does not make a business good, but it changes the asymmetry of surprises.

Second, the backlog/revenue divergence is unusually wide and consensus is treating it as noise. RPO growing at 1.6× revenue and non-current RPO at 2.5× is a large signal to dismiss entirely, particularly when accompanied by CRPO acceleration and an explicit management claim of unchanged discounting.

Third, and most importantly, the market appears to be extrapolating SBC forever. At a 3.07× sales multiple, no credit is being given for the possibility that a company that just cut 7% of its workforce, put a share-price-gated award on its CEO, and committed half its free cash flow to buybacks might also normalize its stock-compensation ratio. That is the cheapest option embedded in the security.

Where consensus is probably right: the moat critique. We find no basis to argue that OpenSearch is not a permanent constraint on Elastic’s pricing power, and the 76.1% gross margin is the arithmetic proof.

12. Fact vs. Interpretation

13. Open Questions

1. What is the “committed tax planning action in fiscal 2027”? The 10-K discloses that the $411M Netherlands-plus-California valuation-allowance release depends on it, but not what it is. Its execution, timing and durability are unknown to outside investors, and a failure would force partial re-establishment.
2. How much of the +28% RPO growth is new demand and how much is contract-duration extension? Non-current RPO growing at 43% — faster than total RPO — is consistent with duration mix. Elastic does not disclose weighted-average contract duration. This is the single most valuable disclosure the company could add.
3. What is the actual paid-deployment split between Elasticsearch and OpenSearch? No reliable public data exists. Elastic’s gross margin and flat NRR are indirect evidence of substitution pressure; direct measurement is unavailable.
4. Who is the 11%-of-revenue channel partner, and on what terms? The 10-K discloses the concentration but not the counterparty. If it is a hyperscaler marketplace, the strategic implications (dependence on a competitor for distribution) differ materially from a conventional distributor.
5. What is the restructuring charge for the ~7% workforce reduction, and what run-rate saving does it produce? Not quantified in the 8-K. First disclosure will be Q1 FY2027 on 2026-08-27.
6. Does the FY2027 non-GAAP operating margin guide of ~19% embed the restructuring benefit? The guide was issued 2026-05-28; the restructuring was committed 2026-06-23. If not embedded, the guide may prove conservative; if embedded, it was pre-announced margin.
7. What is the trajectory of SBC in absolute dollars post-restructuring? A 7% headcount cut concentrated outside go-to-market should mechanically reduce SBC, but Elastic has not guided to it. This is the variable the valuation most depends on.
8. How much revenue do the 600+ AI-using customers represent? Elastic discloses the customer count using AI capabilities but no associated revenue, ACV uplift, or attach-rate economics. Without that, the AI thesis is unfalsifiable.
9. Why has no insider bought a share on the open market? With the stock 65% off its high, a $500M buyback running, and a CEO whose PSUs need a double, the complete absence of code-P purchases is conspicuous even allowing for trading-window constraints.
10. Is the Q4 FY2026 cloud-mix explanation for flat sequential revenue genuine or convenient? It is mechanically plausible and consistent with the RPO data, but it is not verifiable from outside. Q1 FY2027 is the test.

14. What Must Be True

14.1 For the bull case

14.2 For the bear case

The single most important test on either side is Bull #1 / Bear #3 — the stock-compensation ratio. It is the same test read from two directions, it is disclosed annually and inferable quarterly from the share count, and it determines whether Elastic’s cash flow ever becomes the shareholders’. An investor who tracks one number on this name should track that one.

APPENDIX A — Standard Diligence Questionnaire

Report date: 2026-07-31 · Price: $65.79 · Latest reported period: FY2026 (ended 2026-04-30) A supplemental due-diligence questionnaire. Answers are labeled Fact / Interpretation / Assumption where the distinction matters.

General

What thoughtful questions have other investors asked about this company?

The recurring institutional questions on Elastic, evidenced by the topics management chooses to pre-empt on earnings calls and by the price reaction to each print, cluster into five:

1. “Is the open-source fork killing you?” — the single most persistent question since AWS forked Elasticsearch in April 2021. Management rarely addresses OpenSearch by name. (Interpretation: the silence is itself informative; the gross-margin evidence is in the Competitive Position section.)
2. “Why is net expansion stuck?” — 117% → 110% → 112% → 112%. Investors want to know whether the plateau is macro optimization (recoverable) or competitive (structural).
3. “When does GAAP profitability arrive, and does it stay?” — asked every year since FY2022. Q4 FY2026’s relapse from +$0.6M to −$16.4M gave the question fresh life.
4. “What is the real free cash flow?” — the SBC-versus-FCF question. This is the question we regard as decisive, and it is asked less often than it should be.
5. “Is the RPO acceleration real demand or longer contracts?” — the newest question, arising from the Q4 FY2026 print. Management has been asked and has answered only that discounting did not materially change.

A sixth question deserves to be asked and largely is not: why has no insider bought a share on the open market while the stock fell 65% from its high and the company itself repurchased $340M?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Neither, because there are no operating earnings. (Fact) Elastic has posted a GAAP operating loss in every fiscal year of its public life, FY2026 included (−$33.5M). If the question is asked of margins, they are at a cyclical and secular high — the −1.9% GAAP operating margin is the best in company history, up from −40.0% in FY2020 — and are guided higher (~19% non-GAAP for FY2027 versus 16.4% delivered). If asked of reported net income, FY2026’s $367.8M is an artificial all-time high produced by a $434.9M tax-allowance release and will not repeat. (Fact)

Driven by the external environment or internal actions? Predominantly internal. The margin improvement is a deliberate multi-year expense-discipline program: sales-and-marketing plus G&A fell from 61.5% of revenue (FY22) to 52.0% (FY26). The external environment has been broadly neutral-to-mildly-negative — a maturing observability market, aggressive hyperscaler bundling, and a free forked substitute. The one meaningful external tailwind is the AI/RAG demand cycle. (Interpretation)

How stable are revenues? Structurally very stable, tactically volatile. 94% of revenue is subscription; $807.9M of FY2026 revenue was recognized out of the opening deferred-revenue balance; deferred revenue grew 21.4%. (Fact) But reported quarterly revenue is unstable because of the cloud/self-managed recognition asymmetry — self-managed licences recognize partly upfront, cloud ramps — so commitment mix alone can swing an in-quarter number, as management stated occurred in Q4 FY2026. (Fact, with management attribution as Interpretation) The market’s reaction is correspondingly violent: every major print in five years moved the stock 12–37% in a session.

Outlook for products/services? Solid but not accelerating. FY2027 guidance is $1.985–2.000B, +14.6%, with sales-led subscription revenue +16.9%. (Fact) Elastic Cloud (+22% in FY26) continues to outgrow self-managed (+14%) and is now 48% of revenue. Management guides to intra-year acceleration (Q1 +13.1% building through the year) on the strength of a backlog growing at 28%.

How big will this market be — growing, shrinking, domestic or international? Growing, and international. Elastic addresses three markets: search/AI-retrieval (the vector-database segment alone was ~$2.55B in 2025 growing ~22% annually, and the broader enterprise search/context market is far larger), observability (a mature multi-tens-of-billions market growing high single digits to low teens), and security analytics/SIEM (growing low-to-mid teens with an active displacement cycle). (Fact/Interpretation) 46% of Elastic’s revenue is already outside the United States and rising (44% FY25, 42% FY24). (Fact) The relevant caution is that a growing market is not the constraint here — Elastic’s growth is decelerating within growing markets, which is a share statement, not a market statement.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More. Elastic faces three separate competitive sets simultaneously and is the #3-or-worse player in each. In observability: Datadog (+32% growth), Dynatrace, Grafana/Prometheus (free), Cisco/Splunk, plus bundled CloudWatch and Azure Monitor. In SIEM: Cisco/Splunk’s installed base, Microsoft Sentinel bundled into E5, CrowdStrike, and Palo Alto’s XSIAM (>$600M ARR growing 100%). In AI retrieval: Pinecone (~70% of the managed vector segment), Weaviate, Qdrant, and — the more serious threat — vector search becoming a feature inside Postgres (pgvector), MongoDB (Atlas Search), Databricks and every hyperscaler stack. (Fact/Interpretation) The countervailing force is the capital cycle: venture funding has withdrawn and the sector is consolidating, which reduces the number of funded competitors. It does not reduce OpenSearch.

How profitable is the business (ROIC, ROE)? Not profitable on any undisturbed measure. (Fact) GAAP operating income FY2026: −$33.5M. NOPAT is negative against ~$1.87B of total capital, so ROIC is negative and we decline to publish a flattering figure. The ~29% ROE that screens will report for FY2026 is the tax benefit divided by equity and is meaningless. Elastic has not earned its cost of capital in any year of its public existence.

The right profitability question for a software business of this shape is cash-based, and there the answer is: 18.5% reported free-cash-flow margin, ~1.3% owner free-cash-flow margin after stock compensation. (Fact) That gap is the whole analysis.

How profitable is the industry — how many competitors, what barriers to entry? The industry is profitable at the top and unprofitable in the middle. Datadog and Dynatrace earn genuine operating margins (Dynatrace ~29% non-GAAP); the second tier — Elastic, GitLab, Sumo Logic, Amplitude — largely does not. Barriers to entry at the application layer are moderate (brand, integrations, enterprise sales motion, data gravity in the installed base). Barriers at the engine layer are close to zero for Elastic specifically, because a free, Apache-2.0, Linux-Foundation-governed, drop-in-compatible fork exists and is distributed natively by the largest cloud provider. (Fact) This is the defining structural feature of the business.

Can the business be easily understood? Yes, unusually so. Elastic sells one engine, three solutions, two deployment models, on subscription. The complexity is not in the business model; it is in (a) the revenue-recognition asymmetry between cloud and self-managed, and (b) the accounting distance between reported net income, reported free cash flow, and owner economics. An investor who understands those two things understands Elastic.

Can it be undermined by foreign low-cost labor? Not in the conventional sense — this is intellectual property, not labor arbitrage, and Elastic already operates a globally distributed workforce (46% of revenue and a large share of engineering outside the US). The analogous threat is AI-driven labor substitution in software development itself, which cuts both ways: it lowers the cost for competitors and open-source contributors to replicate features (negative), and it is the explicit rationale for Elastic’s own 7% workforce reduction, framed as “working in an age of AI automation” (positive for margins). (Fact for the framing; Interpretation for the net effect)

Do brands matter? Moderately, and more than one might expect. “Elasticsearch” is close to a generic term for the category among developers, and the ELK stack (Elasticsearch, Logstash, Kibana) is a widely-known reference architecture. That mindshare is a genuine asset — it drives the free-tier funnel that converts into paid deployments. But brand is not pricing power here: the same developers who know the brand also know that OpenSearch runs the same queries for free. (Interpretation)

What is the nature of competition? Three distinct kinds, which is unusual and unhelpful. Against Datadog and Dynatrace: feature-and-workflow competition where Elastic competes primarily on cost per ingested terabyte — a price-led motion. Against Microsoft and AWS: bundling, where Elastic competes against something the customer has already paid for. Against OpenSearch: competition against free, where Elastic must justify the delta in managed operations, paid-tier features, and support. The third is the hardest, because the reservation price is zero.

Customers’ switching costs? Real and material, and the primary reason to own the equity. Petabyte-scale indices in production ingest paths are expensive and risky to migrate; queries are embedded in applications, dashboards in operational runbooks, and detection rules in security workflows. The financial fingerprints: 112% net expansion, 21.4% deferred-revenue growth, >$5M ACV customers +30%, and multi-year commitments secured (per management) without material discount concessions. (Fact) Our test — that a moat must tie to a financial outcome that would deteriorate without it — is passed: given a free substitute one click away in the AWS console, a 112% expansion rate is only explicable by switching costs. (Interpretation) But observe the ceiling: switching costs sufficient to retain are not sufficient to price, which is why eight years have produced no operating profit.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Three, and they are material.

1. The Elasticsearch codebase and developer mindshare. Internally developed software is expensed; ~$451.9M of FY2026 R&D (26% of revenue) and roughly $2.3B cumulatively since the IPO sits nowhere on the balance sheet. (Fact)
2. Net operating loss carryforwards. The accumulated deficit of $732.0M corresponds to real NOLs that will shelter real future cash taxes. The FY2026 valuation-allowance release recognized a portion of this as a deferred tax asset; the economic value is genuine even though the accounting entry was theatrical. (Fact/Interpretation)
3. The installed base and customer relationships. Because growth has been almost entirely organic — roughly $200M of cumulative acquisition spend against $1.74B of revenue — essentially no customer-relationship intangible has ever been capitalized. (Fact) A comparable business assembled by acquisition would carry billions of intangibles for the same economics.

Off-balance-sheet liabilities? Nothing alarming. Operating and finance leases are on-balance-sheet under ASC 842 ($20.7M of finance-lease obligations). The material unrecognized commitment is third-party cloud-hosting spend — Elastic Cloud runs on AWS, Azure and GCP, and multi-year committed-spend agreements with hyperscalers are standard in this industry. These flow through cost of revenue and are the mechanical reason gross margin is capped at 76%. (Fact for the margin effect; Assumption that committed-spend agreements exist at scale — the 10-K purchase-obligation table is the place to verify.) The other item to watch is the ordinary software-industry exposure to indemnification and IP claims, which is disclosed but not quantified.

How conservative is the accounting? Mostly conservative, with one aggressive election.

Conservative: revenue recognition is straightforward and heavily ratable; there is no percentage-of-completion judgment of consequence; deferred contract acquisition costs were a net drag in FY2026 ($163.7M capitalized against $111.1M amortized, −$52.6M), meaning commissions are suppressing rather than flattering reported cash flow; capital expenditure is trivial and nothing is being capitalized to manufacture margin; disclosure of the tax benefit’s non-recurring nature was explicit and repeated in three places.

Aggressive: the basis for the valuation-allowance release. The UK portion ($23.7M) rested on the conventional three-years-cumulative-income test. But the Netherlands ($390.5M) and California ($20.7M) portions — 94.5% of the total — were “supported by the implementation of a committed tax planning action in fiscal 2027 that is expected to generate future taxable income in each jurisdiction.” (Fact, direct quote) Releasing a $411M allowance on the strength of a restructuring that has not yet happened, forecast to produce income that does not yet exist, is defensible under ASC 740 only if the planning action is genuinely committed, prudent and feasible. It is a materially softer basis than demonstrated profitability, it is re-assessed quarterly, and it creates a live risk of partial re-establishment. (Interpretation)

How CapEx-hungry is the business? Barely at all — one of the genuinely attractive features of the model. FY2026 capital expenditure was $5.1M on $1,739.3M of revenue: 0.3%. (Fact) Free cash flow is 98.4% of operating cash flow. Infrastructure cost is rented (it sits in cost of revenue as hyperscaler hosting fees), which converts capital intensity into gross-margin pressure — a trade that favours cash conversion and disfavours margin, and explains the 76% gross margin versus GitLab’s 88%.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? FY2026: $326.9M operating cash flow, $321.8M free cash flow (18.5% margin); the company’s own “adjusted free cash flow margin” was ~20%, guided to 21.5% for FY2027. (Fact) Owner free cash flow after $298.4M of stock compensation was ~$23.4M. (Fact, arithmetic)

The stated philosophy, articulated at the October 2025 Financial Analyst Day and reaffirmed on the Q4 FY2026 call, is to return 50% of free cash flow through share repurchases unless attractive acquisitions require the cash. (Fact) In practice FY2026 saw $340.1M of repurchases — more than 100% of reported FCF, front-loading the new authorization.

Significant acquisitions recently? No, and this is to management’s credit. Cumulative acquisition spend since FY2020 is roughly $200M — $24.4M (FY20), $119.9M (FY22), $19.1M (FY24), $36.8M (FY26) — against $356.4M of goodwill. (Fact) There is no roll-up, no bubble-multiple platform deal, and no acquired-revenue illusion inflating the 10.9× revenue increase since FY2018. In a sector that destroyed enormous capital on 2021-vintage acquisitions, restraint is a real achievement.

Buying back shares? Yes — for the first time in company history. The board authorized $500M on 2025-10-09 with no expiration; $340.1M was repurchased in FY2026 (~4.4M shares, ~68% of the authorization), at an average of roughly $61.5 in Q4. (Fact) The timing is genuinely good: Elastic is buying at the 6.8th percentile of its own valuation history.

The critical qualification: weighted diluted shares still rose from 103.7M to 107.2M, and FY2027 guidance is 107.5–108.5M. (Fact) The program is a dilution-offset, not a return of capital — it converts cash into the maintenance of a roughly constant share count. Absent it, dilution would run ~4% a year; with it, ~0.7–1%. That is worth having, and it is not the same thing as returning capital. (Interpretation)

Issuing large amounts of new shares to insiders? Yes. $298.4M of stock-based compensation in FY2026 ($308.2M with employer taxes) — 18% of revenue, unchanged from FY2025 and down only from 20% in FY2024. (Fact) Diluted shares outstanding have grown from 61.9M (FY2018) to 107.2M (FY2026) — a 73% increase — while the share price fell from a $70.00 IPO-day close to $65.79. (Fact) This is the central capital-allocation failure of the business: the company has been financing a material share of its operating cost with equity issued at declining prices.

Compensation policy of directors/management? Per the DEF 14A filed 2025-08-28, the FY2025 executive annual cash-incentive plan was weighted 35% Total Revenue and 30% Total Cloud Revenue; the PSU plan’s performance metric was total revenue (80%/100%/120% attainment → 50%/100%/200% payout). FY2025 attainment: Cloud revenue $688M vs $669M target; Total revenue $1,483M vs $1,485M target. 100% of target PSUs earned for FY2025, 97% for FY2024. (Fact)

There is no return-on-capital metric, no free-cash-flow metric, no margin metric in the PSU plan, and no share-count or dilution metric anywhere. (Fact) This is the incentive structure that produced 35% annual revenue compounding, 18–20%-of-revenue stock compensation, a 73% increase in share count, and a share price below its IPO-day close. Management was paid exactly as designed. Shareholders were not.

The October 2025 special CEO award is a genuine structural correction. Ashutosh Kulkarni received PSUs with a $29.3M grant-date fair value, up to 456,491 shares, across four tranches over five years, gated on 30-consecutive-trading-day average share prices of $121.69 (+40%), $152.64 (+76%), $166.77 (+93%) and $198.15 (+129%) from an $86.61 baseline, modified by relative TSR against the Russell 3000 median and requiring continuous service as CEO for four to five years. (Fact) It is 100% at risk and well-constructed. At $65.79 it is 24% below its own baseline and requires +85% merely to reach Tranche 1 — maximal alignment, minimal retention power. (Fact/Interpretation)

Motivations of management? Founder Shay Banon remains on the board and files Form 4s; CEO Ashutosh Kulkarni has led since 2022 and now carries a $29.3M share-price-gated award. The observable behavior is consistent with a team that has pivoted, deliberately and recently, from growth-maximization to a balanced growth-and-returns posture — the buyback, the 50%-of-FCF policy, the Rule-of-40-by-FY2029 target, the 7% workforce reduction, and the CEO award are all October-2025-onward and all point the same direction. (Interpretation)

Against that, one uncomfortable observation: across the recent Form 4 corpus (filings clustered 2025-12-10, 2026-03-10 and 2026-06-10, covering Kulkarni, Welihinda, Banon, Exner, Herzog, Dodds and Bone), transaction codes are exclusively A (grant) and S (sale). There is not a single code-P open-market purchase. (Fact) No insider bought a share while the stock fell from ~$94 to $43.30 and the company itself spent $340M buying it. Trading windows and Dutch-issuer constraints mitigate the signal, but they do not erase it. (Interpretation)

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? None of these. Elastic N.V. is a Dutch naamloze vennootschap whose ordinary shares trade directly on the NYSE under ESTC — not an ADR. It files US domestic forms (10-K, 10-Q, 8-K, DEF 14A) with the SEC rather than 20-F/6-K. No K-1; holders receive standard Form 1099 treatment. (Fact)

Two Dutch-domicile considerations do apply and are disclosed in the 10-K risk factors: (a) Dutch dividend withholding tax would apply to any dividend or, absent an applicable exemption, to share repurchases — a real constraint on how capital can be returned; and (b) Dutch corporate law and the Dutch Corporate Governance Code afford shareholders different (generally weaker) rights than a Delaware corporation, including anti-takeover provisions, and US civil-liability judgments may not be enforceable in the Netherlands. (Fact)

Dividend policy? None, and none intended. The 10-K states Elastic “has not paid a cash dividend on our ordinary shares in the past and we do not intend to pay any cash dividends… in the foreseeable future.” (Fact) Capital return is entirely via the $500M repurchase authorization. Given the Dutch withholding-tax treatment of distributions, buybacks are the structurally preferred mechanism regardless.

How profitable is the business? Answered above and in the Financial Quality section. Summary: gross margin 76.1%; GAAP operating margin −1.9%; non-GAAP operating margin 16.4% (guided to ~19% in FY2027); reported FCF margin 18.5%; owner FCF margin ~1.3%. GAAP net margin of 21.1% is a tax artifact and should be disregarded. (Fact)

Is net income diverging from cash from operations? Yes, violently, and in both directions across the cycle — which is the tell.

FY2026 is the first year in which net income exceeded operating cash flow, and the reason is precisely diagnostic: the $370.1M non-cash tax benefit flowed into net income and was then subtracted back out of the cash-flow statement as a −$398.6M deferred-income-tax adjustment. (Fact) In every other year the divergence ran the other way, driven by the SBC add-back. The correct conclusion is that neither net income nor operating cash flow, taken alone, describes this business: net income is distorted by taxes, and operating cash flow is inflated by a non-cash charge that is nonetheless a real economic cost. Owner free cash flow — OCF, less capex, less SBC — is the only measure that reconciles to shareholder value, and it is ~$23.4M.

Risks & Downside

What factors would cause the stock to decline? In descending order of probability-weighted impact:

1. A cut to the FY2027 guide at the Q1 print (2026-08-27). Precedent: the Q1 FY2025 guidance cut cost 26.5% in a single session. Elastic moves double digits on virtually every print.
2. Net expansion breaking below 110% in the FY2027 10-K — a fourth consecutive flat-to-down year would convert the “plateau” narrative into a “decline” narrative.
3. Stock compensation failing to fall as a share of revenue, confirming that owner FCF stays near zero.
4. Evidence of competitive displacement — deceleration in the >$1M ACV cohort, or a public loss of a marquee observability or SIEM account.
5. A broader software/AI-disruption de-rating, to which ESTC is highly levered (beta 1.35, Cloud Computing factor loading +1.04, LowVolatility −0.78).
6. Partial re-establishment of the valuation allowance if the FY2027 tax planning action does not execute — non-cash, but a visible negative headline.

Risk of a catastrophic loss? Low over any relevant horizon, on balance-sheet grounds. Elastic holds $1,370.3M of cash and marketable securities against $591.6M of total debt — net cash of ~$778.7M, roughly 11% of the market capitalization. (Fact) The single debt instrument, $575M of 4.125% senior notes, does not mature until July 2029, costs ~$25M a year, and is comfortably covered by liquid assets. There is no covenant pressure, no refinancing wall, and no financing risk. The business generates positive operating cash flow before stock compensation and would generate positive cash flow even in a sharp revenue decline, because 94% of revenue is contracted subscription and 0.3% of revenue is capital expenditure.

The realistic severe-downside scenario is not insolvency but multiple compression on a decelerating business: ESTC traded at ~2.3× EV/sales and $43.30 as recently as 2026-04-10, a level it could revisit on a guidance cut. That is roughly a 34% decline from $65.79 — painful, not catastrophic. (Interpretation)

Chance of a total loss? Negligible. Net cash of ~$779M, $1.74B of contracted-subscription revenue, 1,720 customers above $100K of ACV, $1.98B of remaining performance obligations, and near-zero capital intensity. For a total loss you would need a simultaneous collapse of all three end markets and a wholesale migration of the installed base to OpenSearch — and even then the residual value of the codebase, the customer relationships and the NOLs would be substantial. (Interpretation) Note also that a business with these characteristics at 3.07× forward sales is a plausible strategic or private-equity target, which puts a further floor under the equity. (Assumption — no transaction interest has been reported; this is a structural observation, not a rumor.)

Recent News & Events

Has the business environment changed recently? Yes, in three ways over the past twelve months.

1. AI shifted from narrative to measurable demand. Elastic now reports 600+ customers with >$100K ACV using its AI capabilities — more than one-third of that cohort, including 40+ serverless customers newly captured. (Fact)
2. The sector de-rated violently and then partially recovered. ESTC fell to an all-time low of $43.30 on 2026-04-10 amid a broad “AI disrupts software” drawdown that also took Dynatrace and GitLab to their cheapest-ever multiples, then rallied 51.9% to $65.79. (Fact)
3. US public-sector demand became a genuine channel. The CISA SIEM-as-a-Service relationship is expanding across federal civilian agencies displacing incumbent security vendors, and management expects that momentum to continue into FY2027. (Fact for the relationship; Interpretation for its durability)

Significant acquisitions? No. FY2026 acquisition spend was $36.8M — immaterial against $1.74B of revenue. (Fact)

Change in accounting policies? No policy change. But the most significant accounting event in the company’s history occurred in Q4 FY2026: the release of deferred-tax valuation allowances in the Netherlands ($390.5M), the United Kingdom ($23.7M) and California ($20.7M) — $434.9M gross, a $370.1M net income-tax benefit — which converted a $2.3M pre-tax loss into $367.8M of reported net income. (Fact) This is an estimate change under ASC 740, not a policy change, and it is non-recurring. The 10-K states explicitly that Elastic “would have incurred net losses in such years as well without the releases.”

Recent changes — new markets, facilities, management?

• 2025-10-09: First-ever share repurchase program authorized, $500M, no expiration. (Fact)
• 2025-10-13: Special one-time performance-based equity award to CEO Ashutosh Kulkarni, $29.3M grant-date fair value, share-price and rTSR gated. (Fact)
• October 2025: Financial Analyst Day — capital-allocation policy (50% of FCF to buybacks) and the Rule-of-40-by-FY2029 target articulated. (Fact)
• 2026-06-23: Committed plan to reduce the workforce by approximately 7% (~280 of 4,019 employees), described as aligning teams with “working in an age of AI automation,” with continued hiring in go-to-market. Charge not yet quantified. (Fact)
• Product: Agent Builder reached general availability; Cross Project Search launched on serverless; the Jina V5 omni family shipped for multimodal search; MCP applications launched for security and observability, embedding Elastic-powered workflows into Claude, VS Code and similar tools. (Fact)
• Management: No CEO or CFO change. Ashutosh Kulkarni (CEO) and Navam Welihinda (CFO) remain in place; founder Shay Banon remains a Section 16 filer. (Fact)

Next catalyst: Q1 FY2027 results on 2026-08-27 — the first test of the FY2027 guide, the first disclosure of the restructuring charge, and the first read on whether Q4’s cloud-mix explanation for flat sequential revenue was genuine.

APPENDIX B — Source Appendix

Report date: 2026-07-31 · All sources accessed 2026-07-31 unless otherwise stated. Sources are ordered by evidentiary weight: primary regulatory filings first, then company communications, then quantitative data services, then third-party and industry sources. All sources listed are public and independently verifiable.

A. Primary regulatory filings (SEC EDGAR — CIK 0001707753)

The full trailing-60-month EDGAR corpus (452 filings from 2021-07-31 onward) was enumerated and reviewed. Form breakdown: 188 Form 4, 128 Form 144, 35 8-K, 15 10-Q, 5 10-K, 6 DEF 14A, 9 Form 3, 7 S-8, 5 Form 4/A, 1 8-K/A.

A.1 Annual reports (Form 10-K)

FY2026 10-K on EDGAR: https://www.sec.gov/Archives/edgar/data/1707753/000170775326000021/estc-20260430.htm

A.2 Current reports (Form 8-K) — material events

All 8-K filings are available on SEC EDGAR under CIK 0001707753.

A.3 Proxy statements (DEF 14A)

EDGAR: https://www.sec.gov/Archives/edgar/data/1707753/000170775325000080/estc-20250828.htm

A.4 Insider transactions (Forms 3, 4, 4/A)

Reviewed directly from EDGAR primary documents rather than any aggregator. The recent Form 4 clusters — 2025-12-10 (7 filings), 2026-03-10 (7 filings), 2026-06-10 (7 filings), covering Ashutosh Kulkarni (CEO), Navam Welihinda (CFO), Shay Banon (founder/director), Ken Exner, Carolyn Herzog, Mark Eugene Dodds and Jane E. Bone — were parsed for transaction codes.

Finding: transaction codes are exclusively A (award/grant) and S (sale). Zero code-P open-market purchases across the sampled corpus. The pattern is quarterly RSU vesting with associated sales on the 10th of March, June, September and December.

Index: https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001707753&type=4&dateb=&owner=include&count=100

A.5 Quarterly reports (Form 10-Q)

Fifteen filings covering FY2022 Q1 through FY2026 Q3 were reviewed and used to corroborate the quarterly revenue and GAAP operating-income series in Section 6.3. Quarterly figures in this report were sourced from ROIC.ai and reconciled against these filings and the corresponding 8-K earnings releases.

B. Company communications

The transcript is publicly available via Elastic Investor Relations (ir.elastic.co) and standard transcript services.

C. Quantitative data services

Data-quality note: the ROIC.ai news feed returned no results for ESTC over a 90-day window, so the recent-events timeline in Section 8 was built from 8-K filings and public news sources instead.

D. Third-party and industry sources

E. Peer companies referenced for comparison

Peer metrics cited in Sections 4, 10 and 11 are drawn from each company’s own public filings and earnings releases as of the dates shown. They are directional quality and valuation cross-checks, not synchronized comparables, and have not been re-struck as of 2026-07-31.

F. Analytical frameworks applied

G. Methodological notes and corrections made in this report

1. Enterprise value corrected. Standard screens — including ROIC.ai — net only cash and equivalents ($768.7M) against debt, omitting the $601.5M of short-term marketable securities. This overstates ESTC’s enterprise value by ~$601.5M and every multiple by ~10%. All valuation figures use the corrected ~$6,112M.
2. GAAP P/E and its percentile disregarded. FY2026 GAAP EPS of ~$3.49 is ~100% the product of a $434.9M valuation-allowance release. The AZI P/E percentile of 13.1 is invalid for this name; the P/S (5.4th) and P/B (2.0th) percentiles are the valid own-history reads.
3. FactorsToday returns are annualized at every horizon. The m3_return of +2.667 is an annualized figure; de-annualized it is +38.4% over the quarter, which was cross-checked against the AZI price CSV ($47.53 on 2026-04-30 → $65.79 on 2026-07-31) before being reported.
4. Factor loadings read within a single model. Because the model orthogonalization is hierarchical and level-dependent, the same factor’s beta legitimately differs across the four nested models. Loadings are quoted with their model identified and are never compared across models.
5. Owner free cash flow defined. Throughout this report, “owner free cash flow” means operating cash flow less capital expenditure less stock-based compensation — i.e. reported free cash flow with SBC charged at its cash-equivalent economic cost. FY2026: $326.9M − $5.1M − $298.4M = ~$23.4M.
6. Peer comparisons are asynchronous. Peer multiples and metrics are drawn from prior reports dated 11 June to 17 July 2026 and have not been re-struck as of 2026-07-31. They are used as directional quality and valuation cross-checks only.

H. Disclosures

This report makes no statement, implication or assumption regarding any position held by the author or any affiliate in Elastic N.V. or in any company named herein. No such position should be inferred.

The analysis in Sections 1–15 contains no investment recommendation and no price target. The single, deliberate exception is the clearly-labeled Claude's Take block, which is the author’s own subjective opinion, is general information only, and is not investment advice. Readers should conduct their own research and consult their own advisers before making any investment decision.