MongoDB MDB

Reference table

1. What they actually sell — the segments, with real numbers

MongoDB's disclosure is clean and useful, which is a welcome change in this batch. FY26 (ended 2026-01-31) revenue was $2,463.8M, +22.8%.

Three observations.

1. Atlas is the company now. At 73.4% of revenue growing 28.7%, it determines everything. The self-managed Enterprise Advanced line grew just +7.3% and is shrinking as a share of the mix — from 31.0% (FY24) to 23.5% (FY26). The consolidated 22.8% growth rate understates Atlas and overstates the business; the correct way to model MongoDB is as a 29% grower with a 7% drag attached that shrinks each year. That is a structurally improving arithmetic and it is the strongest argument for the current growth acceleration continuing.

2. Atlas is consumption-priced, and that is the exponential mechanism. Customers pay by usage rather than by seat. If applications built on MongoDB process more data — which is exactly what all_in's "10x the data stored in 3 years" claim predicts — revenue rises without any new sale, contract or renegotiation. This is the specific structural reason the exponential score is 7.

3. Geography is broadening evenly. EMEA at +23.1% is nearly matching the Americas at +23.4%, which suggests the product is winning on merit rather than on home-market distribution.

2. The acceleration — the single most important fact

Infrastructure software companies at $2.5B of revenue are supposed to decelerate. MongoDB is doing the opposite.

Read the growth column. 21.8%, 23.7%, 18.7%, 26.8%, 25.2%. Excluding the Q3 dip, the trend is clearly upward, and the two most recent quarters average +26.0% against the two oldest at +22.8%. For a company that grew 19.2% in FY25, this is a genuine reacceleration.

Read the operating income column. From -$53.6M and -$65.3M to breakeven in Q4'FY26 and -$24.8M in Q1'FY27. Note that Q1 is seasonally the weakest quarter — Q1'FY26 was -$53.6M, so the year-over-year improvement is $28.8M. The trajectory is right; the destination has not arrived.

Annual view of the operating leverage:

Operating leverage is arriving through sales and marketing efficiency, not research. SG&A has fallen from 66.9% to 48.2% of revenue — nearly nineteen points — while R&D has held roughly flat as a percentage. That is exactly the right shape: the company is spending the same proportion on the product and progressively less on selling it, which is what happens when a product starts pulling itself through.

The one deterioration is gross margin: 74.8% (FY24) to 71.7% (FY26), down 310 basis points. This is the cost of the Atlas mix shift — running databases in the cloud carries infrastructure expense that shipping software does not. It is a structural feature of the mix improvement, not a sign of pricing pressure, but it does cap the ultimate margin ceiling. A 71-72% gross margin business cannot reach the operating margins a 80%+ gross margin licence business can.

3. The stock-compensation problem — why 58.5x is really something else

This is the reason the verdict is Hold and it needs to be stated precisely.

Four things follow, and they are all consequences of the same fact.

1. Stock-based compensation of $550.5M exceeds free cash flow of $500.2M. The consensus non-GAAP EPS of $6.12 for FY27, on which the 58.5x multiple is calculated, adds back a cost that is larger than the cash the business generates. This is not an accounting quibble — it is the difference between a company that makes money and one that does not. On GAAP, MongoDB lost $71.2M in FY26.

2. The direction is genuinely improving. Stock compensation has fallen from 29.7% to 22.3% of revenue over four years while the absolute dollar figure rose 44%. Revenue is outgrowing the grant pool. This is the mechanism by which the GAAP business eventually turns profitable, and it is working — just slowly.

3. The $400.3M buyback in FY26 did not stop dilution. Diluted shares still rose from 74.6M (FY25) to 81.2M (FY26), an 8.8% increase. MongoDB spent $400M and the share count went up. More recently the count has stabilised — 81.1M (Q1'FY26) to 81.6M (Q1'FY27), essentially flat — which is a genuine improvement, but shareholders should understand that the buyback is offsetting dilution rather than shrinking the company.

4. Free cash flow of $500.2M is real and it is a genuine inflection — up 315% from $120.6M — driven by the operating loss narrowing and a working capital line that was essentially neutral (-$0.7M) rather than a swing. Against our corrected enterprise value of $26.44B, that is a 1.9% free-cash-flow yield. Attractive on trajectory; expensive on level.

4. Valuation — priced in or room?

At $357.97 (market cap $28.79B, our corrected EV $26.44B), trailing revenue is $2.602B and trailing non-GAAP EPS is $5.29.

Note the FY29E dispersion: $3.21 to $17.85, a 5.5x spread across five analysts. This is the widest estimate range in the entire batch and it tells you something important — nobody covering this stock knows what the operating leverage looks like at scale. The FY27E and FY28E ranges are tight ($6.00-$6.27 and $7.11-$7.73 on 20 analysts each) because they are near-term extrapolations. Beyond that, the model breaks down. All valuation work here uses FY27E and FY28E; FY29E is used only for the bull case and FY30E is excluded entirely.

6a. What today's price assumes (the inversion)

At $357.97 — 58.5x FY27E consensus non-GAAP EPS and 10.2x trailing EV/sales — the price embeds roughly:

• Revenue compounds at ~19% annually from the trailing $2.60B to $3.50B by FY28, i.e. MongoDB adds $894M of annual revenue in under two years. (Source: consensus, 26-27 analysts — deep coverage.) Falsifier: any quarter below +18% year-over-year growth. The recent trend is +25-27%, so consensus is actually modelling deceleration from the current run rate.
• Non-GAAP EPS grows ~20% annually to $7.32 by FY28, requiring operating leverage to continue at roughly the current pace. (Consensus.) Falsifier: any quarter where the non-GAAP EPS beat streak breaks — six consecutive beats averaging 32% is a high bar to maintain.
• The market keeps paying roughly 50-58x forward. At an unchanged price the multiple rolls to 48.9x FY28E and 39.3x FY29E. This price requires the market to keep paying a near-60x multiple for a company that does not yet make GAAP money. That is the assumption, stated plainly.
• Stock compensation continues declining as a share of revenue, from 22.3% toward the mid-teens, so that GAAP eventually converges toward non-GAAP. (Our computation from the reported trend: 29.7% → 27.1% → 24.6% → 22.3%.) Falsifier: SBC rising above 24% of revenue in any year.
• THE MOST FRAGILE ASSUMPTION: that the multiple holds. Unlike LPLA or ZM, where the base case works on earnings growth alone, MongoDB at 58.5x has essentially no margin for multiple compression. A move from 58.5x to 40x — still a very full multiple for a GAAP-lossmaking business — takes the stock to $245 on FY27E earnings even if every operating assumption is met. The equity is a leveraged bet on sentiment toward high-multiple infrastructure software, not on MongoDB's execution. The 52-week range of $201.00 to $440.60 — a 2.2x swing in twelve months on a business that never grew slower than 18% — is the empirical proof of that.

6b. The return bridge (why the multiple moves)

Expected return ≈ EPS growth + multiple drift + shareholder yield.

Our base case of $400 is built as approximately 55x FY28E consensus non-GAAP EPS of $7.32 = $403. Equivalently, that is holding roughly today's forward multiple (58.5x on FY27E) with a modest 3-4 turn compression, and rolling it forward one year to the FY28E number.

Decomposing the ~12% base-case return: non-GAAP EPS grows 19.6% from FY27E to FY28E; the multiple compresses from 58.5x to ~55x (-6%); shareholder yield is negative, because the $400.3M buyback did not prevent diluted shares rising 8.8% in FY26. So the return is: earnings growth, less a small multiple give-back, less dilution — arriving at roughly 12%.

This is a bridge with very little slack in it. If the multiple compresses to 45x instead of 55x — a level still far above the market — the fair value drops to $329, below today's price. The entire base case depends on a rich multiple staying rich. That is not a robust foundation, and it is why the verdict is Hold despite genuinely liking the business.

Shareholder yield is negative. No dividend; $400.3M of buybacks in FY26 offset by $550.5M of stock compensation, producing net dilution. Do not attribute any part of the return to capital return.

6c. Variant perception (where we differ, what would surprise)

• We have NO variant perception on price, and we say so plainly. Our base fair value of $400 sits within 0.05% of the street's $400.18 consensus target. The ratings profile is the strongest in this batch — 36 buy, 6 hold, 2 sell — and we agree with the direction. Per house discipline, a name where our view equals consensus defaults toward Hold. That is what this is. Manufacturing a divergence to justify a more exciting verdict would be dishonest.
• Where we do differ is on the risk characterisation, and that is the value this dive adds. The street's $400 target and 36 buy ratings are constructed on non-GAAP earnings that exclude $550.5M of stock compensation — more than the entire $500.2M of free cash flow. We think the market is under-weighting the dilution: MongoDB spent $400.3M on buybacks in FY26 and the diluted share count still rose 8.8%. The consensus is right about the business and insufficiently precise about who owns the returns.
• Where we are MORE positive than the modelled consensus: the growth trajectory. Consensus models FY27E revenue growth of +20.3% and FY28E of +17.9% — i.e. deceleration — while the last four quarters have run +18.7%, +26.8%, +25.2% and are accelerating, with Atlas at 73.4% of the mix and growing 28.7%. If the acceleration is real and sustains, consensus revenue is too low. That is a genuine, if modest, positive variant, and it is the reason the bull case reaches $515.
• Positive surprise that would force a repricing: GAAP operating income turning sustainably positive, which on the current trajectory could happen within two to three quarters. The moment MongoDB stops losing money on reported accounting, the entire "the multiple is on fake earnings" objection dissolves and a 55-60x multiple becomes far more defensible. This is the single cleanest resolution of our objection. Watchable at: 2026-08-26 and the following two prints.
• Negative surprise: revenue growth falling below 18%, or gross margin continuing to slide below 71%. Either would break the operating-leverage arithmetic at a price with no valuation support beneath it, and the $230 bear case becomes live. Watchable at: 2026-08-26.

Synthos fair values

• Bear ~$230 (-36%). AI-native workloads land on relational databases that have added document and vector capability rather than on MongoDB, growth reverts toward 15%, and the market re-rates a GAAP-lossmaking business to 31x FY28E non-GAAP EPS of $7.32 = $227. Rounds to $230. Note this is 14% above the $201.00 52-week low, so the market visited a worse scenario within the last year.
• Base ~$400 (+12%). ~55x FY28E consensus non-GAAP EPS of $7.32. Assumes modest multiple compression from today's 58.5x and consensus earnings arriving. Lands within half a percent of the street's $400.18.
• Bull ~$515 (+44%). The acceleration proves durable, GAAP turns positive, and the market pays 57x FY29E non-GAAP EPS of $9.10 = $519. Rounds to $515, matching the street's high target. Note the FY29E estimate carries a $3.21-$17.85 dispersion, so this case rests on the least reliable number in the file — flagged rather than hidden.

5. Knowledge base — the strongest lane in this batch, on the business

4 claims across 3 tracked voices. Thin on breadth, but the quality and recency are the best in this batch.

no_priors (2026-01-22, conviction 85, bullish, entity-tagged MongoDB) — the single highest-conviction current single-name claim in this batch:

> "MongoDB is the only truly disruptive database force past $2B, large durable TAM, mission-critical sticky apps, well-positioned for AI-native unstructured data."

This is a dense claim and every element of it is testable against the data. "Past $2B" — FY26 revenue was $2.464B. "Mission-critical sticky apps" — the Enterprise Advanced line still generates $578.1M and grows, seven years into the Atlas transition, because customers do not migrate off production databases. "AI-native unstructured data" — the structural argument for the document model. This is a claim that has aged well in the six months since it was made: revenue growth has accelerated from 18.7% to 25.2% over that period.

all_in (2026-06-08, conviction 75, bullish, entity-tagged MDB):

> "Infrastructure software (Databricks, Snowflake, MongoDB, Oracle) is undervalued — enterprises will need 10x the data stored in 3 years, so core storage/data infra is needed."

The most recent claim in the lane, and a category call rather than a single-name one. The "10x data" mechanism maps directly onto Atlas's consumption pricing and is the specific reason the exponential score is 7.

all_in (2025-10-09, conviction 80, bullish):

> "Founder-led software winners have 10x'd as public companies in the last decade, validating holding great companies well past the IPO." MongoDB is named alongside Palo Alto Networks, ServiceNow and HubSpot. A category-membership claim rather than a specific one; weighted moderately. Note that MongoDB's current CEO per the profile is Chirantan Jitendra Desai, so the founder-led framing may no longer strictly apply — flagged for accuracy.

arthur_hayes (2022-12-10, conviction 80, BEARISH, entity-tagged MongoDB and MDB):

> "Software names still trading at ~20x revenue are insane; down 75% already and should fall another 75%."

This claim is dated December 2022 and has been contradicted by subsequent price action — MDB trades at $357.97 today. It is included for completeness and honesty rather than excluded for inconvenience. Two things are worth noting: the specific prediction was wrong, but the underlying observation — that ~20x revenue for infrastructure software is a demanding multiple — is precisely the objection this dive is making at 10.2x EV/sales and 58.5x forward earnings. The claim was wrong on timing and direction; the discipline behind it is the same discipline producing our Hold.

Read. Breadth 3, net conviction positive-high, and the recency profile is excellent — the three bullish claims are dated 2025-10-09, 2026-01-22 and 2026-06-08, i.e. all within the last ten months, and the single bearish claim is three and a half years old and demonstrably wrong. This lane strongly underwrites owning MongoDB the business.

And it is worth being explicit about what the lane does NOT say. No tracked voice has commented on MongoDB's valuation at current levels. no_priors' claim is about disruptive position and market size; all_in's is about category undervaluation across four names. The lane is a business endorsement, not a price endorsement, and this dive does not convert one into the other. That distinction is the whole difference between our Hold and a Buy.

6. Technicals — constructive but flat, and mid-range

• Price $357.97 (+6.07% today), -18.8% from the $440.60 52-week high, +78.1% above the $201.00 low. Maximum drawdown from peak: -38.8%.
• Above both moving averages — 50-DMA $335.85 (+6.6%) and 200-DMA $333.60 (+7.3%) — but the two averages sit within 0.7% of each other. That is a flat trend, not a strong one: the intermediate and long-term averages have converged, which typically means a period of directionless consolidation.
• RSI 55.2 — neutral. MACD -0.81, marginally negative despite today's +6% move.
• Relative performance is strong: +50.5% over twelve months against SPY +19.9% and QQQ +23.9%; +35.9% over three months against SPY +5.1%. But six-month return is -3.6% — meaning essentially all of the twelve-month gain came in the last quarter after a poor prior six months. The 52-week range of $201.00 to $440.60 is a 2.2x swing on a business that grew every quarter — this stock is priced by sentiment, and its own price history proves it.
• Beta 1.545.

The technical read is neutral-to-mildly-constructive and offers no entry edge. Mid-range price, flat moving averages, neutral RSI. Nothing here argues for urgency in either direction, which is consistent with the Hold.

7. Moat & competitive position

MongoDB's moat is the developer-facing standard for the document data model, protected by database switching costs. Both halves matter.

Switching costs are the strongest form of enterprise lock-in. Applications are written against a database's query semantics, data model and operational behaviour. Migrating means rewriting application code, revalidating correctness, and accepting downtime risk on production systems. The proof is in the data: the self-managed Enterprise Advanced line still generated $578.1M in FY26 and grew +7.3%, seven years into MongoDB's own push toward Atlas. Customers do not leave databases; the company cannot even fully migrate them to its own better product. This is exactly what no_priors means by "mission-critical sticky apps."

The developer-standard position is the second layer. MongoDB won by being what developers chose when nobody made them choose — a bottom-up adoption pattern that produces genuine mindshare rather than purchased distribution. Once the applications exist, the enterprise contract follows.

Three real competitive constraints, stated honestly:

1. Relational incumbents are adding document and vector capability. PostgreSQL in particular has absorbed JSON handling, and the major cloud providers all offer managed database services adjacent to MongoDB's. This is the primary long-run threat and it is a real one — the moat is around the model and the migration cost, not around any capability that cannot be copied.

2. Gross margin at 71.7% and falling caps the ceiling. Atlas carries cloud infrastructure cost, so MongoDB will never earn the margins a pure licence business earns. A 71-72% gross margin business with 29% R&D intensity has a structurally limited operating margin.

3. Cloud providers are simultaneously partners and competitors. Atlas runs on their infrastructure; they sell competing services on the same infrastructure.

Verdict on the moat: strong, developer-driven, and protected by the highest switching costs in enterprise software — with a margin structure that limits how much of it converts to profit.

8. Data integrity — what we rejected and why

Vendor data corruption checking is standard for these dives. For MDB:

• REJECTED: the consensus EBITDA and EBIT series. The est block reports FY26E EBIT of $678.0M against a reported GAAP operating loss of -$137.0M — a gap of $815M. Stock compensation of $550.5M accounts for only two-thirds of it; the series does not reconcile to GAAP even after that adjustment. The FY25E figure shows EBIT of $550.8M against reported -$216.1M, the same pattern. These are non-GAAP figures on an undisclosed basis and are excluded from all analysis. Only the revenue, net income and EPS lines from est are used, and the EPS line is explicitly labelled non-GAAP throughout.
• REJECTED: FY30E and FY31E consensus EPS. The series runs $6.12 (FY27E), $7.32 (FY28E), $9.10 (FY29E), $15.41 (FY30E), then $12.77 (FY31E) — a non-monotonic decline, with FY30E carrying 1 analyst and FY31E carrying 2. Excluded entirely.
• FLAGGED, used with caution: FY29E EPS dispersion of $3.21 to $17.85 across 5 analysts — a 5.5x spread, the widest in this batch. Used only for the bull case, with the caveat stated inline.
• CORRECTED: enterprise value. The vendor reports $27,786M, netting only the $1,084M cash line and ignoring $1,304M of short-term investments. Correct net cash is $2,354M ($2,387M cash and short-term investments less $32.9M debt), giving an enterprise value of $26,438M. This changes EV/Sales from 10.7x to 10.2x. All EV figures use the corrected number.
• FLAGGED: km_ttm.evToEBITDATTM = 1,173x and enterpriseValueMultipleTTM = 1,173x. Artifacts of a trailing EBITDA near zero ($23.7M by our arithmetic). Meaningless, ignored.
• FLAGGED: ratios_ttm internal inconsistency. ebitMarginTTM (-0.43%) differs from operatingProfitMarginTTM (-4.16%) and equals pretaxProfitMarginTTM (-0.52%) closely — the vendor's "EBIT" field is populated with pre-tax income, the same systematic error seen across this feed. Do not read the EBIT fields as operating income. Related artifacts: priceToEarningsRatioTTM of -995.7x, taxBurdenTTM of 2.14, effectiveTaxRateTTM of -114%. All are arithmetic noise from a near-zero denominator.
• NOTE, not a rejection: consensus EPS is on a NON-GAAP basis and the gap is very large. FY26E consensus EPS was $4.816 against reported GAAP diluted EPS of -$0.88 — the consensus series is not merely above GAAP, it has the opposite sign. Every forward P/E in this dive is a non-GAAP multiple, labelled as such at every use, and section 3 quantifies exactly what is being excluded.
• GAP: no customer metrics. No customer count, no net revenue retention, no Atlas customer cohort data, no consumption growth rate. For a consumption-priced business, net revenue retention is the single most informative metric and it is absent.
• NOTE on insider activity — one ambiguity flagged rather than interpreted. Director and co-founder Dwight Merriman sold approximately 10,000 shares across 2026-07-14 at $316.52-$324.05. The securitiesOwned field runs 980,500 down to 973,000 across those transactions, and then a separate 2026-07-16 entry for 835 shares at $324.12 shows holdings of 464,100 — a drop of roughly 509,000 shares not accounted for by any sale in the file. This is most plausibly a transfer between direct and indirect holdings or between trusts rather than a market sale, but the data does not say. We flag it and decline to interpret it. The disclosed open-market sales themselves are small relative to the position and show no open-market purchases.

9. Verdict, kill-criteria & flip conditions

Hold. MongoDB is a genuinely excellent business and this dive says so throughout: revenue growth accelerating from 18.7% to 25.2%, Atlas at 73.4% of revenue growing 28.7%, GAAP operating losses narrowing from -$233.7M to -$137.0M, free cash flow inflecting 315% to $500.2M, $2.35B of net cash, SG&A falling nineteen points as a share of revenue, and the strongest knowledge-base lane in this batch — no_priors at conviction 85, all_in at 75 and 80, with the sole bearish claim three and a half years old and demonstrably wrong.

The price is the entire objection. At 58.5x FY27 consensus non-GAAP earnings and 10.2x EV/sales, with stock compensation of $550.5M exceeding free cash flow of $500.2M and GAAP operating income still negative, there is no valuation support beneath the stock — and the 52-week range of $201.00 to $440.60 on a business that never grew slower than 18% is the empirical proof. Our base fair value of $400 lands within half a percent of the street's $400.18. We have no edge on price, and house discipline says a no-edge name defaults toward Hold.

This is not a criticism of the company. It is a statement about what we are being asked to pay.

Position protocol:

1. If you own it: hold. The business is compounding, the lane is strong, and the base case is modestly positive. There is no reason to sell a category leader at a fair price.

2. If you do not own it: wait for $280-300, which would be roughly 40x FY28E — still a full multiple for a GAAP-lossmaking business, but one where the multiple has room to be wrong. At $300 this becomes a 2-3% Stage-In. The stock traded at $201 within the last twelve months.

3. Or wait for the GAAP crossover. If operating income turns sustainably positive — plausible within two to three quarters on the current trajectory — the central objection in this dive dissolves and a 55x multiple becomes far more defensible at any price. That is an information trigger rather than a price trigger and it may arrive first.

Pre-registered KILL / avoid criteria:

• Revenue growth falling below 18% in any quarter — this breaks both the acceleration narrative and the operating-leverage arithmetic at a price with no support beneath it.
• Gross margin below 71% — the Atlas mix shift is meant to be margin-dilutive-but-worth-it; below 71% the trade stops being worth it.
• Stock-based compensation rising above 24% of revenue, reversing the four-year improvement from 29.7% to 22.3%.
• Diluted share count rising above ~83M despite continued buybacks — evidence that dilution is outrunning capital return.
• The non-GAAP EPS beat streak breaking after six consecutive beats averaging 32%.

Pre-registered FLIP TO STAGE-IN:

• Price at $280-300 with the operating trend intact. This is the primary trigger.
• GAAP operating income sustainably positive for two consecutive quarters — the cleanest possible resolution of our objection, and it would justify raising the base case toward $460.
• Revenue growth sustaining above 26% for two quarters, which would mean consensus FY27-28 revenue is too low and the bull case becomes the base.
• Free cash flow exceeding stock-based compensation for the first time — the point at which the business genuinely funds its own employee costs in cash.

Where MDB fits in the Synthos Framework Portfolio. The data-infrastructure sleeve, at 0-1% today (hold existing positions, add nothing) with a 2-3% target on a pullback below $300. Note the correlation profile: with a 1.545 beta and a demonstrated 2.2x annual price range, MongoDB will dominate sleeve volatility. It is genuinely complementary to the other data-infrastructure names all_in groups it with — a document/operational database rather than an analytical warehouse — so it is not redundant exposure. Logged as a tracked Synthos call (Hold) as of 2026-08-04 at $357.97.

Single biggest risk: stock-based compensation of $550.5M — 22.3% of revenue and larger than the $500.2M of free cash flow — sitting underneath a 58.5x non-GAAP multiple. The earnings that justify the price are calculated after excluding a cost bigger than the cash the business produces, and the $400.3M FY26 buyback still left the diluted share count 8.8% higher. The most fragile assumption in today's price is simply that the multiple holds. A compression from 58.5x to 40x — a level that would still be very generous for a company with negative GAAP operating income — takes the stock to $245 on unchanged FY27 earnings. Every operating assumption in this dive can be met and the shareholder can still lose a third of their capital to sentiment alone. That is the definition of a business worth owning at the wrong price.

Provenance & disclosures

• Traceability: 4 KB claims across 3 tracked voices (breadth 3, net conviction positive-high). Bullish: no_priors (2026-01-22, conviction 85 — the highest-conviction current single-name claim in this batch, entity-tagged MongoDB), all_in (2026-06-08, conviction 75, entity-tagged MDB, category call), all_in (2025-10-09, conviction 80, category-membership claim — note the "founder-led" framing may no longer apply given the current CEO). Bearish: arthur_hayes (2022-12-10, conviction 80) — three and a half years old and contradicted by subsequent price action; included for honesty rather than excluded for inconvenience. The lane endorses the business, not the valuation — no tracked voice has commented on MongoDB's price at current levels, and this dive does not convert a business endorsement into a price endorsement.
• Data as-of: fundamentals 2026-04-30 (Q1'FY27, filed 2026-05-29) · estimates 2026-08-04 · prices 2026-08-04 ($357.97, +6.07%; 50-DMA $335.85, 200-DMA $333.60, RSI 55.2, beta 1.545) · KB claims 2026-08-04. Sole data source: scripts/deepdive/vti_data/MDB_data.json. Fiscal year ends 31 January.
• Rejected data: the consensus EBITDA/EBIT series (FY26E EBIT of $678.0M against reported GAAP operating income of -$137.0M, a gap not reconciled even by the $550.5M of stock compensation); FY30E and FY31E consensus EPS (non-monotonic, 1-2 analyst coverage); evToEBITDATTM of 1,173x and the associated near-zero-denominator ratio artifacts.
• Corrected data: vendor enterprise value of $27,786M ignores $1,304M of short-term investments; corrected to $26,438M on net cash of $2,354M, changing EV/Sales from 10.7x to 10.2x.
• Flagged: ratios_ttm populates the EBIT field with pre-tax income (a systematic error across this feed); FY29E EPS dispersion of $3.21-$17.85 across 5 analysts, used only for the bull case with the caveat stated inline; an unexplained ~509,000 share reduction in a director's reported holdings between two July filings, most plausibly a trust or direct/indirect transfer — flagged and deliberately not interpreted.
• Non-GAAP note: every forward P/E in this dive uses the consensus NON-GAAP EPS series, which has the opposite sign to GAAP (FY26E consensus $4.816 against reported GAAP diluted EPS of -$0.88). Section 3 quantifies exactly what is excluded: $550.5M of stock compensation, 22.3% of revenue, exceeding the $500.2M of free cash flow.
• Disclosed gaps: no customer count, no net revenue retention, no Atlas cohort or consumption-growth disclosure — for a consumption-priced business, net revenue retention is the single most informative absent metric.
• Fair-value caveat: the $230 / $400 / $515 anchors are scenario multiples on consensus non-GAAP EPS, labelled at each step. The base case assumes modest multiple compression from 58.5x to ~55x and derives the return from earnings growth less dilution. This is scenario arithmetic, not a discounted-cash-flow model. The base case deliberately lands on the street consensus, and that convergence is itself the verdict input.
• Not investment advice. Independent research, educational and informational only, never personalized. Nothing here is a recommendation to buy or sell any security.
• Version: 2026-08-04-full.