Built by Grok 4.6 HighMDB | Earnings Review — FQ2 FY2027
Verdict: ACCELERATING on the top line and on operating leverage; Atlas is stable-high, not accelerating. FQ2 revenue $771.8 million (+30.5% YoY vs $591.4 million) is the fastest quarter since FY2024. Atlas held 29% for a fifth straight print on a record +$126.9 million YoY dollar-add. EA and other $181.2 million jumped +35.9% — file 36%, not the CFO’s same-call ~30%. Non-GAAP operating margin expanded +940 bps YoY to 24.1%; non-GAAP EPS $1.90 (+90% vs $1.00).
Clean double beat, and a re-widening after two compressed prints. vs FMP $735.0M / $1.62 = +$36.8M / +5.0% revenue and +$0.28 / +17.3% NG EPS. Cleared the company high ($734 million / $1.61) by +5.1% / +18.0%. L12Q is 12/12 on both lines vs FMP and vs the high of the box. Street sat $1 million / 1¢ through the box and they still cleared it — a true operating beat, not a sleepy Street.
Guidance: second consecutive FY raise. Revenue mid $3.01B ($2.99–$3.03 billion, +$70M / +2.4% vs May), Atlas FY ~27% (+300 bps), EA ~11% (from mid-single-digit), NG OI mid $626.3M, Rule of 44 at the high end. FQ3 $756–$761 million / $1.57–$1.61 sits at or below post-print FMP $762.8M / $1.61. That is the sandbag, not a demand cut. Implied FQ4 revenue ~+14% YoY is prudence they keep repeating.
Tone: same playbook, louder on Atlas durability. Berry named the 200–300 bp Atlas beat framework on the call and reused it for Q3. Desai: “I have never felt better.” Street spent the hour trying to get a Q4 Atlas number; management gave the fifth-straight-29% speech and “and not or,” and left implied ~24% Q4 Atlas unbridged.
Do not underwrite +30.5%. The +525 bps of sequential acceleration is EA duration on a stable ~29% Atlas. FQ3 guide +20.7% is −~980 bps off a 30% Atlas compare. Fade the FQ3 low; fade implied FQ4 +14%; do not fade the Atlas 29% run-rate until a quarter prints inside the pin.
| Revenue | $771.8M (+30.5% YoY, +5.0% vs FMP) | Atlas / mix | $565.9M +29% · mix 73.3% (EA outgrew) |
| EA and other | $181.2M +35.9% · ARR ~+11% (file 36%, not 30%) | Non-GAAP EPS | $1.90 (+90% YoY, +17.3% vs FMP) |
| NG OI / OM | $185.9M / 24.1% (+940 bps YoY) | NG gross margin | 75.9% (+210 bps YoY) |
| NRR / $100k+ | 122% (+300 bps) · 2,999 (+17.0%) | RPO / cRPO | $1,519.2M (+91.3%) · $797.3M |
| FCF / H1 FCF | $137.6M (+97%) · H1 $335.2M | FY27 raise | $2.99–$3.03B · Atlas ~27% · EA ~11% · Rule of 44 |
| FQ3 box vs FMP | $756–$761M vs FMP $762.8M (Street above the high) | L12Q beat rate | Rev 12/12 · EPS 12/12 vs FMP and vs co. high |
Revenue drivers sit above the consolidated total. YoY is same-fiscal-quarter (FQ2 vs FQ2), never sequential across seasons. Atlas is the installed-base engine; EA is the swing factor this print.
Segment drivers
| Driver | FQ3'25 | FQ4'25 | FQ1'26 | FQ2'26 | FQ3'26 | FQ4'26 | FQ1'27 | FQ2'27 |
|---|---|---|---|---|---|---|---|---|
| Atlas ($M) | 362.6 | 389.0 | 395.9 | 439.0 | 470.4 | 502.6 | 512.5 | 565.9 |
| Atlas YoY | +26.4% | +24.1% | +26.1% | +29% | +30% | +29.2% | +29% | +29% |
| Atlas $ add YoY | +75.7 | +75.6 | +82.0 | +99.3 | +107.8 | +113.6 | +116.6 | +126.9 |
| EA and other ($M) | 149.6 | 141.9 | 135.6 | 133.4 | 138.7 | 170.5 | 153.7 | 181.2 |
| EA YoY | +13.8% | +8.0% | +10.2% | +7.5% | −7.3% | +20.1% | +13.4% | +35.9% |
| Atlas mix | 68.5% | 70.9% | 72.1% | 74.2% | 74.9% | 72.3% | 74.5% | 73.3% |
| NRR | 120% | 118% | 119% | 119% | 120% | 121% | 121% | 122% |
| $100k+ customers | 2,314 | 2,396 | 2,506 | 2,564 | 2,694 | 2,799 | 2,895 | 2,999 |
| $100k+ YoY | +17.3% | +16.8% | +17.3% | +17.1% | +16.4% | +16.8% | +15.5% | +17.0% |
| RPO ($M) | 728 | 748 | 774 | 794 | 994 | 1,473 | 1,459 | 1,519 |
| Customers (k) | 52.6 | 54.5 | 57.1 | 59.9 | 62.5 | 65.2 | 67.7 | 70.6 |
Driver read. Three things; only EA is a problem to annualize.
- Atlas is a 29% machine adding more dollars on a larger base. Five quarters in a 28.9–29.7% band. The dollar-add line is the honest tell: +$99 → +$108 → +$114 → +$117 → +$127 million — sixth straight rising YoY add, now a ~$2.26B run-rate. Mix ticked down 90 bps because EA exploded, not because Atlas slowed.
- EA is the second derivative, and they will not guide it. +35.9% vs a ~20% intra-quarter pin. ARR ~+11% for a third straight double-digit quarter is the clean number. Search + vector search GA June 30; no large bundled deals this quarter (Berry, vs last year’s Q4 mix). H2 EA still ~flat; Q3 mid-single-digit. Do not annualize 36%.
- Leading indicators corroborate duration. NRR back to 122% from an 118% trough. $100k+ re-accelerated to +17.0% at 2,999. RPO +$91.3% YoY after the FQ4 bulge; first sequential add since. Record +2,900 customer adds; Atlas customers 69,300.
Consolidated P&L
| Metric | FQ3'25 | FQ4'25 | FQ1'26 | FQ2'26 | FQ3'26 | FQ4'26 | FQ1'27 | FQ2'27 |
|---|---|---|---|---|---|---|---|---|
| Revenue ($M) | 529.4 | 548.4 | 549.0 | 591.4 | 628.3 | 695.1 | 687.6 | 771.8 |
| Rev YoY | +22.3% | +19.7% | +21.9% | +23.7% | +18.7% | +26.7% | +25.2% | +30.5% |
| NG GM (calc.) | 76.6% | 75.1% | 74.1% | 73.8% | 74.2% | 75.5% | 74.5% | 75.9% |
| GM YoY bps | −81 | −213 | −93 | −167 | −244 | +41 | +44 | +210 |
| NG OI ($M) | 101.5 | 112.5 | 87.4 | 86.8 | 123.1 | 158.8 | 123.2 | 185.9 |
| NG OM | 19.2% | 20.5% | 15.9% | 14.7% | 19.6% | 22.8% | 17.9% | 24.1% |
| OM YoY bps | +104 | +541 | +865 | +370 | +42 | +234 | +199 | +940 |
| NG EPS | 1.16 | 1.28 | 1.00 | 1.00 | 1.32 | 1.65 | 1.32 | 1.90 |
| EPS YoY | +20.8% | +48.8% | +96.1% | +42.9% | +13.8% | +28.9% | +32.0% | +90.0% |
| GAAP EPS | −0.13 | 0.19 | −0.46 | −0.58 | −0.02 | 0.18 | 0.05 | 0.50 |
| FCF ($M) | 34.6 | 22.9 | 105.9 | 69.9 | 140.1 | 176.7 | 197.5 | 137.6 |
Margin trajectory. Gross-margin trough is in. Non-GAAP GM compressed from a 77.4% FQ3 FY24 peak to 73.8% in FQ2 FY26, then recovered +210 bps YoY — first clean YoY expansion since FY24. Operating leverage is the bigger signal: NG OI more than doubled to $185.9 million on +30.5% revenue. Third consecutive GAAP EPS-profitable quarter. FCF $137.6 million (+97%); H1 $335.2M. Do not annualize 24% OM — FQ3 is guided ~20.5% because they will invest.
Five-year annual (FYE 31 Jan)
| Metric | FY2022 | FY2023 | FY2024 | FY2025 | FY2026 |
|---|---|---|---|---|---|
| Revenue ($M) | 873.8 | 1,284.0 | 1,683.0 | 2,006.4 | 2,463.8 |
| YoY | +48.0% | +47.0% | +31.1% | +19.2% | +22.8% |
| Atlas ($M) | 492.3 | 808.3 | 1,105.4 | 1,405.2 | 1,807.9 |
| Atlas YoY / mix | +81.8% / 56.3% | +64.2% / 62.9% | +36.8% / 65.7% | +27.1% / 70.0% | +28.7% / 73.4% |
| NG OM | 1.4% | 4.8% | 16.1% | 14.9% | 18.5% |
| NG EPS | −0.02 | 0.81 | 3.33 | 3.66 | 4.97 |
| FCF ($M) | −6.7 | −24.7 | 109.9 | 114.5 | 492.6 |
FY2025 was the trough (+19.2% revenue, −115 bps OM). FY2026 re-accelerated. H1 FY2027 is running +28.0% ($1,459.4M vs $1,140.4M), so FY2027 prints a faster year even after the guided FQ3 step-down. FMP FY2027E $3.026B / $6.56 sits inside the new box — Street is not modeling a continuation of FQ2’s 30%.
FQ2 FY2027 vs FMP print-time estimate. Company guide issued on the FQ1 call applies to this print (+1Q offset). Daloopa publishes no consensus series for MDB.
| Metric | Consensus (FMP) | Actual | Variance | Beat/Miss |
|---|---|---|---|---|
| Revenue | $735.0M | $771.8M | +$36.8M / +5.0% | BEAT |
| NG diluted EPS | $1.62 | $1.90 | +$0.28 / +17.3% | BEAT |
| Atlas growth | co. pin ~26% | +29% / $565.9M | ~+300 bps | BEAT vs guide |
| EA and other | co. pin ~20% | $181.2M +35.9% | ~+1,590 bps | BEAT |
| NG operating income | co. $152–$156M | $185.9M | +$29.9M / +19.2% vs high | BEAT |
| NG operating margin | ~21% at guide high | 24% | +900 bps YoY vs 15% | BEAT |
Why they beat. Atlas consumption, third straight 200–300 bp beat vs the initial Atlas box, driven by North America and the $100k+ cohort — not a one-off logo. EA duration / search-and-vector launch is the other half of the overshoot; Berry: “broad based … rather than any single transaction,” and “there were no large bundled deals in the quarter.” Almost all of the extra revenue dropped to the operating line. AI is early / small (CFO to Raimo) — treat it as a 2027 option, not the FQ2 variance bridge.
| Metric | FQ3'25 | FQ4'25 | FQ1'26 | FQ2'26 | FQ3'26 | FQ4'26 | FQ1'27 | FQ2'27 |
|---|---|---|---|---|---|---|---|---|
| Revenue vs FMP | +6.5% | +5.3% | +4.0% | +6.7% | +5.8% | +3.7% | +3.5% | +5.0% |
| EPS vs FMP | +71% | +94% | +50% | +49% | +64% | +11% | +11% | +17% |
Green = beat vs FMP. Yellow = structurally smaller EPS surprise after Street marked onto the box (FQ4 FY26 onward). Blue outline = this print. No miss cells in the 8-quarter window. Source: Daloopa actuals, FMP Street.
Pattern: consistent beater. L12Q revenue 12/12 (100%), EPS 12/12 (100%), same vs the high of the company box. Average L12 surprise +4.8% rev / +52.4% EPS. L4Q average +4.5% / +25.9% — EPS surprise compressed ~4,000 bps once Street sat on the high end starting FQ4 FY26. This print breaks the two-quarter compression (+3.5% / +10.9% → +5.0% / +17.3%) without going back to the 50–100% EPS-sandbag era.
The live analog into FQ3 is this print: +5% revenue, +17% EPS, still above a Street number that already sat on/through the high, then a raise. Read the new FQ3 box with the same 200–300 bp Atlas cushion they just admitted to, plus an EA duration tail they still will not forecast.
stable/earnings at each print. Company quarterly guide series 2051645 / 2051649, +1Q offset.MongoDB guides one quarter ahead plus a full-year range. EBITDA and gross margin are not guided. FCF is guided only as conversion vs NG operating income.
New box (issued 2026-09-01)
| Metric | FQ3 FY2027 | FY2027 | vs prior FY mid (May) |
|---|---|---|---|
| Revenue | $756–$761M (mid $758.5M, +20.7%) | $2.99–$3.03B (mid $3.01B, +22.2%) | +$70M / +2.4% |
| Atlas growth | ~26% (toughest compare; LY +30%) | ~27% | +300 bps |
| EA and other | Q3 mid-single-digit; H2 ~flat | ~11% | MSD → first double-digit year in 3 yrs |
| NG OI | $152–$156M (~20.5% OM at high) | $616.3–$636.3M (~21% at high) | +$45.3M / +7.8% |
| NG EPS | $1.57–$1.61 | $6.39–$6.58 | +$0.44 / +7.3% |
| Diluted shares | 87.1M | 86.4M | −0.3M vs prior 86.7M |
| Rule of X | — | Rule of 44 at high end (23% + 21%) | was Rule of 40 |
| FCF conversion | — | Upper end of 80–100% of NG OI | pinned |
FQ2 conversion (credibility). Guide issued in FQ1 applies to FQ2 actuals. Revenue high $734M → actual $771.8M = +$37.8M / +5.1%. They have cleared the high end of the next-quarter revenue box in each of the last six printed quarters. A 100% above-high beat rate with a static ~$5M revenue band is sandbagging, not precision. Treat the FQ3 low as non-informative; the high is the number they expect to clear.
vs FMP (post-print vintage 2026-09-21)
| New guide mid | FMP | vs Street | |
|---|---|---|---|
| FQ3 revenue | $758.5M | $762.8M (21 analysts) | −$4.3M / −0.6%; high still $1.8M under Street |
| FQ3 NG EPS | $1.59 | $1.61 (22) | −$0.02; Street at the high |
| FY27 revenue | $3.010B | $3.026B (27) | −$16M / −0.5%; Street at the new high |
| FY27 NG EPS | $6.485 | $6.56 (20) | −$0.07; Street at the new high |
| FY28 (unguided) | none — Investor Day Sep 29 | $3.584B / $7.82 | Street fades to ~+18% |
Implied 2H math the Street is skipping. H1 already printed $687.6M + $771.8M = $1,459.4M. FY mid leaves H2 $1,550.6M (+17.2% YoY). FQ3 mid already embeds +20.7%, so implied FQ4 at FY mid = $792.1M vs $695.1M = +14.0%. Implied FQ4 Atlas at a 27% FY (H1 Atlas $1,078.4M; Q3 ~$593M) is ~$625M / ~+24% vs $502.6M. That 500 bp H1-to-Q4 Atlas step-down is Zukin’s question. It is EA H2-flat + Atlas compare + prudence, not Atlas breaking.
Profitability was raised faster than revenue (+7.8% OI / +7.3% EPS vs +2.4% sales). That is operating leverage, not a quality-of-earnings tell. FQ3 OM steps down ~350 bps sequential from 24% to ~20.5% because they will hire quota-carrying reps and fund EA AI features. Do not annualize FQ2’s 24%.
Credibility verdict. Guidance is informative as a floor, performative as a midpoint. Six straight quarters clearing the high of a $5M-wide band, with Atlas pinned 200–300 bps light of the live run-rate, is a policy. The September raise is the honest part: they rolled a $38M revenue beat and a $30M OI beat into FY and still left FQ3 under post-print Street.
| Metric | FQ3'25 | FQ4'25 | FQ1'26 | FQ2'26 | FQ3'26 | FQ4'26 | FQ1'27 | FQ2'27 |
|---|---|---|---|---|---|---|---|---|
| Revenue YoY % | +22.3 | +19.7 | +21.9 | +23.7 | +18.7 | +26.7 | +25.2 | +30.5 |
| Rev accel (bps) | +946 | −254 | +211 | +184 | −501 | +806 | −150 | +525 |
| EPS YoY % (n/G) | +20.8 | +48.8 | +96.1 | +42.9 | +13.8 | +28.9 | +32.0 | +90.0 |
| EPS accel (bps) | +4,556 | +2,800 | +4,724 | −5,322 | −2,906 | +1,511 | +309 | +5,800 |
Revenue is accelerating, but it is not an Atlas re-acceleration. Total YoY climbed from a +18.7% trough in FQ3 FY2026 (EA −7.3%) to +30.5% — fastest since FY2024, exactly as Desai framed it. Across those same five Atlas prints the consumption engine did not speed up: 29% → 30% → +29.2% → 29% → 29%. Treat +30.5% as a mix print, not a new Atlas run-rate.
Four inflections. (1) FQ3'26 EA-duration trough — Atlas actually peaked ~30%. (2) FQ4'26 multiyear re-acceleration (+806 bps). (3) FQ2'27 EA+Atlas peak. (4) FQ3'27E guide cliff (−980 bps to +20.7%). Atlas (yellow on the chart) is the flat 29% band. Total revenue only leaves that band when EA duration spikes or goes negative.
EPS is operating leverage on a still-easy $1.00 base. +90% / +5,800 bps on 24% OM vs 15% year-ago is real — and not repeatable at +90%. FQ3 guide $1.57–$1.61 vs $1.32 is ~+20.5% YoY, a −6,950 bps collapse in the EPS growth rate as the $1.00 FQ2'26 base rolls off. FY NG EPS $6.39–$6.58 vs $4.97 is +29% to +32% — healthy, not explosive. H1 EPS already +61%; the FY box implies H2 EPS growth of only +7% to +13%.
The 8-quarter shape in one sentence: Atlas found a 29% floor in FQ2 FY2026 and has not left it; total revenue and EPS only look like a breakout when EA duration and operating leverage stack on that floor; management’s own FQ3/FY box takes both stacks back off.
The debate into Investor Day (Sep 29) and the December print is not whether FQ2 beat — it did. It is whether a fifth straight ~29% Atlas print is a plateau the Street will not pay for, and whether the FQ3 Atlas ~26% pin is Berry’s usual 200–300 bp box or the first real step-down. FMP’s post-print FQ3 average sits $1.8M / ~30 bps above the high end. That is a miss-setup on the printed number even if Atlas reprints high-20s.
| # | Catalyst | Timing | What consensus needs | What we heard |
|---|---|---|---|---|
| 1 | FQ3 print + FQ4/FY close | 2026-12-07 | FMP $762.8M / $1.61 — above the high. FMP FQ4 $806.2M sits on the FY high-end path. | Q3 $756–$761M; Atlas ~26%; EA MSD; OM ~20.5%. “Q3 is our toughest compare of the year for Atlas.” Watch the FQ4 guide more than the FQ3 print. |
| 2 | Atlas 26% guide vs 29% run-rate | Continuous; next read Dec 7 | No FMP Atlas split. Whisper into FQ2 was 30.5–31% (Mizuho/Barron's). Implied Atlas at 26% ≈ $593M. | Fifth straight ~29%; 200–300 bp beat framework named. A print ≥28% with NRR still ≥122% makes 26% the box. A print ≤26% without “consumption a little better” language breaks the streak. |
| 3 | Investor Day New York | 2026-09-29 | FMP FY2028E $3.584B / $7.82 (+18.4% / +19.2%) is the number a new LT frame has to clear or reset. | “You will hear more about our new product innovations.” Financial session for next-year guide. A dated Voyage/MCP ARR or Atlas AI-mix update off FY26 30% is incremental. Qualitative victory lap is a nothing-burger. |
| 4 | MongoDB.local NYC | 2026-09-30 | Not modelled. | “Major launches.” Product, not P&L — matters if Voyage-on-EA or a new core-server release shows up. |
| 5 | EA durability vs duration lottery | FQ3 print | Street will fade +36% as term unless ARR holds double-digit. FY EA ~11% ≈ $642M. | ARR ~+11% third straight; H2 still ~flat. Quality check is EA ARR, not EA revenue. If Atlas mix falls again, EA is stealing the narrative. |
| 6 | AI / Voyage / MCP / Frontier Labs | Intra-quarter; next quantified read is Investor Day | Street is not modelling a discrete AI line. Atlas ARR with ≥1 AI use case last printed 30% at FY26 YE — not updated. | Voyage “nearly doubled” QoQ, 2nd straight. MCP GA 2026-08-13 (Claude, Codex, Grok Build, Devin). “It is still early.” A quantified mix at Investor Day is the bull surprise. |
| 7 | NRR / $100k+ / RPO | Dec 7 | Hold at 122% is table-stakes; 123%+ is unmodelled positive. | NRR up on “strength in both Atlas and EA.” $100k+ reaccelerated. RPO finally added sequentially after the FQ4 bulge. |
| 8 | Margin / Rule of 44 / GAAP | FQ3; FY close | FMP FQ3 EPS $1.61 = the high end. Street models 20.5% Q3 / 21% FY OM, not a beat on it. | +250 bps FY OM (100 bps above prior high). Full-year GAAP EPS profitability now in the guide. Negative if Q3 OM &;20% or EPS &;$1.57. |
Bull confirmation (in order): Investor Day puts a number on AI; FQ3 Atlas ≥+28% with NRR ≥122%; FQ3 sales inside the range with a FQ4 guide that supports FMP’s $806M; EA ARR still double-digit on a mid-single-digit revenue quarter.
Bear confirmation: Atlas prints ≤26% and they drop the 200–300 bp language; printed FQ3 sales miss FMP and the FY high end is not reaffirmed; Investor Day is qualitative; EA revenue MSD and ARR rolls over; NRR back to 119%.
The thing the Street is most likely to get wrong. A $759M / $1.59 / Atlas +28% / NRR 122% quarter would be a fundamental beat that screens as a sales miss. That is the December setup.
Nine names got through; operator cut the queue after Tyler Radke and took Mike Cikos as last. Tally: 9 Well Answered / 3 Deflected. The deflections cluster on one number: implied Q4 Atlas growth, plus AI-lab inference dollars.
| Analyst | Ask | Badge |
|---|---|---|
| Raimo Lenschow, Barclays | Atlas drivers / confidence in the raise | Well answered — 29% duration, $127M add, NRR, $2.3B run-rate; AI “small but we are excited” |
| Alex Zukin, Wolfe | Is EA flipping deals away from Atlas? | Well answered — “not coming at expense of Atlas”; June 30 feature gap; hybrid named accounts |
| Alex Zukin, Wolfe | Why does implied Q4 Atlas decelerate? | Deflected — philosophy, holidays, “no bundled deals this quarter.” No Q4 Atlas %. |
| Matthew Martino, Goldman | Voyage logos converting to Atlas? | Well answered — conversion “still early”; Claude #1 / Codex #2 referrals |
| Karl Keirstead, UBS | Which AI use cases have pull-through? | Well answered — customer-facing agents, not 200-person copilots; kill-criteria given |
| Sanjit Singh, Morgan Stanley | EA → Atlas migration clock; neo-clouds | Well answered — rejected the waiting-room premise; some workloads never move |
| Ryan MacWilliams, Wells | Early renewals / true-ups | Well answered — onesies and twosies, max single digits, not widespread |
| Ryan MacWilliams, Wells | Inputs to implied Q4 Atlas | Deflected — second philosophy speech in 20 minutes |
| Evercore ISI | Voyage / Vector attach speed | Well answered — Vector is weeks; Voyage is a GTM/awareness problem |
| Tyler Radke, Citi | Frontier-lab inference sizing | Deflected — timeline yes (one lab since late-2025, more in Q2); no dollars |
| Tyler Radke, Citi | Why not more H2 EA upside after June 30 GA? | Well answered — multiyear deals; 11% is the floor |
| Mike Cikos, BofA | 3–5 year Atlas vs EA mix | Well answered — Atlas mix still up, “probably not at the rate we thought before” |
Implied Q4 Atlas they would not give: H1 Atlas $1,078M; 27% FY implies ~$2,296M; Q3 at 26% ≈ $593M; residual Q4 ≈ $625M / ~+24% YoY. That 500 bp H1-to-Q4 step-down is sitting in the guide whether they say it or not.
Management vs Street (the only contrarian on the tape): Desai repeated “run anywhere / and not or / two growth drivers.” The Street asked the cannibalization question three ways because it does not yet believe EA strength is additive. The data this quarter supports additive dollars (Atlas still +$127M YoY) and does not yet support additive mix (Atlas mix −90 bps YoY). H2 EA-flat is how they force the mix story back on track.
Two genuine contradictions. File 36% for EA, not 30%. Do not model “EA mix is lifting GM.”
Statement A — Desai: “EA and other had a standout quarter growing 36% year over year.” Berry opening takeaways: same 36%. 8-K EX-99.1: “EA & other revenue up approximately 36%.”
Statement C — Berry, minutes later, same prepared remarks: “Turning to EA and other revenue … growing approximately 30% year over year.”
Filing settles it: EA and other was $181.2 million vs $133.4 million = +35.9%. Total company is the 30% print. Statement C is the company-level growth rate, mis-attached to the EA label. ARR ~11% is a third metric (duration-normalized) and was correctly labeled — it does not bridge 36% to 30%.
How to hold it: model 36%. Pair it with 11% ARR, not with 30%. Ask IR so 30% does not reappear in a sell-side note as a second official print.
Berry: “The increase in subscription gross margin was primarily driven by the higher EA revenue mix in Q2.” Subscription GM 78.3%, +70 bps YoY.
10-Q: “MongoDB Enterprise Advanced revenue represented 21% of our subscription revenue during each of the three and six months ended July 31, 2026 and 2025.” Daloopa series 2051911: 21% this quarter and 21% year-ago.
A YoY mix shift of zero basis points cannot be the primary driver of a ~70 bp GM expansion. EA+other as a share of subscription did rise ~100 bps (24.3% vs 23.3%) — that is the “other” bucket, not what the 10-Q calls Enterprise Advanced. The 75.9% NG GM print is real. Do not build an “EA mix is structurally lifting GM” model off this quarter.
Not contradictions (labeled revisions or object mix-ups): FY raises (rev / Atlas / EA / Rule of 44); “fifth straight ~29%” includes a 30% FQ3'26 print inside a 100 bp rounding band; “Atlas accelerating” vs 29% run-rate (Zukin meant vs the March guide, not vs the live rate); “EA not at the expense of Atlas” vs the disclosed Q4 FY26 bundle (denied for this quarter); FQ3'26 “no FY27 multiyear tailwind” is a dead forecast they have since raised three times.
No classic macro on the call — no rates, CPI, consumer, or PMI. The usable cross-ticker signal is US large-enterprise IT demand plus sovereignty / neo-cloud capacity.
US enterprise is the demand print. Americas $478.3 million vs $364.2 million (+31.3% YoY) vs EMEA +30.7% and APAC +25.7%. Berry: main growth driver was “strength in North America and our largest customers,” especially the $100k+ cohort. Constructive for upmarket database / data-platform peers selling into Fortune 100 / G2000; weaker as a China/APAC IT-spend proxy.
Opposite of 2023-style Atlas optimization. Desai: a few large customers asked to re-look contracts because consumption is growing faster than expected — “onesies and twosies. Maximum single digits. Not widespread.” NRR 122% vs 119%. Better for consumption-database vendors than the 2023–24 “optimize existing apps” tape.
Public-cloud capacity is tight enough to move some estates. Customers run EA in neo-clouds “because of the capacity issues in a public cloud”; some neo-cloud DBaaS sits in the EA line. Split for hyperscalers: better for IaaS dollars (Atlas hosting COGS +$31.4 million YoY) and worse for hyperscaler database SKUs (DocumentDB, Dynamo, Cosmos, AlloyDB) wherever run-anywhere wins. Logos unnamed — do not invent CoreWeave.
10-Q vs call. MD&A still says they “expect to experience macroeconomic headwinds on growth rate for our existing Atlas applications in the short term.” Atlas +29% for a fifth straight quarter and FY Atlas raised 300 bps. Treat the MD&A sentence as legal tail-risk, not as management’s current demand view.
| Relationship | Name | Datapoint | Implication |
|---|---|---|---|
| Partner | Anthropic (Claude) | #1 Voyage referral; first-class Atlas MCP surface | Better for MDB + Anthropic developer attach |
| Partner | OpenAI (Codex) | #2 Voyage referral; MCP list | Better for MDB as default operational DB in Codex |
| Partner | xAI (Grok Build), Cognition (Devin), Cursor | MCP GA 2026-08-13 | Distribution, not a named customer win |
| Competitor | PostgreSQL | Frontier lab left Postgres for Atlas; 10x faster reads; more inference in Q2 | Cleanest competitive win of the print. Worse for Aurora/RDS/Cloud SQL/Neon on chat-memory |
| Competitor | Oracle | 10-Q named set; conference migrator is product-not-people | Long-dated negative, not this quarter’s P&L |
| Competitor + supplier | AWS / Azure / GCP | Hosting COGS +$31.4M; compete with native DB SKUs | IaaS dollars up; database-SKU share down |
| Customer | Financial Times, Fireflies, Ewe, Nationwide UK, ElevenLabs | Production RAG / AI-native / hybrid FS | Voyage-as-funnel is real; conversion to Atlas “still early” |
| Customer | Unnamed major US bank / Fortune 100 bank | EA GenAI attach, charged extra; siloed vector startup lost the bake-off | Worse for Pinecone-class / Elastic-class on operational-data-layer consolidation |
| Customer | Frontier AI labs (unnamed, plural) | Inference + chat memory; “small” to Atlas dollars | Long-duration AI-lab story, not a mix shift. Do not assume OpenAI/Anthropic are the workload customers |
AWS DocumentDB, Snowflake, Databricks, Elastic, Pinecone — absent from FQ2 prepared remarks and Q&A. The spoken competitive frame this quarter is Postgres (labs) + siloed search/vector startups (enterprise) + hyperscaler lock-in vs run-anywhere.
- FQ3 is the test of the box. $756–$761M / Atlas ~26% vs FMP $762.8M. A $759M / Atlas +28% print screens as a sales miss and is a fundamental beat. A 26% Atlas print that is not beaten by 200–300 bps is the first break in five quarters.
- EA duration was the FQ2 kicker. If Q3 EA is truly mid-single-digit and ARR holds ≥10%, the 11% FY pin is intact. If ARR rolls over, +36% was term.
- +90% EPS will not survive the $1.00 lap. Anyone extrapolating FQ2 EPS is using the wrong run-rate. FQ3 $1.57–$1.61 vs $1.32 is the number.
- AI is still a 2028 story in the numbers. Labs, Voyage, Vector, MCP are real; they are not in the $3.01B. Paying for them in FY27 is a sentiment trade, not a guide trade.
- Investor Day (Sep 29) is the only event that can change the multiple without a print. FY28 Street $3.58B / $7.82 is unsupported by the company until then.
HOLD. Own the Atlas dollar compounding and the sandbag. Do not pay for +30.5% as the run-rate.