Built by Grok 4.6 High

AMAT | Earnings Review — FQ3 FY2026

Verdict: ACCELERATING. FQ3 broke the FY21–FY25 +3–4% revenue grind. Net sales $9,115 million (+24.8% YoY vs $7,302 million) and non-GAAP EPS $3.50 (+41.1% vs $2.48) are the second consecutive ~1,350 bp acceleration after a two-quarter China-mix air pocket. Non-GAAP GM 50.4% (+150 bps, 13th consecutive YoY expansion). Non-GAAP operating margin 34.0% (+330 bps) is a record.

Print (modest double beat): Revenue vs FMP $8.995B is +$120M / +1.3%; vs company midpoint $8.95B ± $500M is +$165M / +1.8%, inside the band. NG EPS vs FMP $3.40 is +$0.10 / +2.9%; vs $3.36 ± $0.20 is +$0.14. Street had already moved onto the May raise. Beat rate is intact (rev L12Q 11/12, NG EPS 12/12); magnitude is compressing.

Quality of earnings: Semi Systems $7,040 million (+26.5% restated) is 81% of the dollar add. DRAM $ (calc.) +49.5% to ~$1.83B (mgmt +52% incl. HBM packaging). China $2,506 million is −1.6% YoY / 28% of sales — the acceleration is rest-of-world foundry/logic + DRAM, not a China restock. GAAP EPS $3.17 sits below NG (FQ2 GAAP was above NG on investment marks). FCF $2,330 million recovered from FQ2's $210 million WC hole; YTD FCF is still slightly down because capex is running $707 million.

Guidance: FQ4 first look $10.25B ± $500M / NG EPS $4.02 ± $0.20 (+51% / +85% YoY off the FQ4'25 trough). Semi Systems ~$7.9B (+62%). AGS ~$1.84B (+22%). NG GM 50.4% — flat vs FQ3. CY26 Systems raised a third time above >30%; they refused 40%, did not guide FQ1, and parked 2027 dollars until 13 Oct SEMICON. Post-print FMP FY26E $34.35B / $12.81 already implies FQ4 ~$10.31B / ~$4.07 — in line with the midpoint. FY27 FMP $46.74B / $18.67 is a Street blank, not a company number.

Tone: More confident on demand (8-quarter forecasts, CSPs “already generating positive returns,” ICAPS digestion “expiring”). Downgraded on sequential GM (flat on a ~$1.1B step-up; ramp costs + display mix). Same playbook on the ceiling: raise the floor, refuse the number.

Contradictions (4, two high): Q2’s 30% Systems year vs a linear path that is ~40% is still unresolved; 2H volume flipped from a GM tailwind to a ramp-cost headwind; ICAPS digestion was retired in 90 days; AGS “a little above mid-teens” was already running +22%.

Catalysts into 2026-11-12: SEMICON West investor breakfast 2026-10-13 (convert “stronger than >30%” into a number); FQ4 print vs $7.9B Systems and DRAM 2H mix; any FQ1 sequential-growth confirmation.

Semi Systems is ~77% of sales and the entire earnings engine. DRAM (incl. HBM) is the incremental dollar; foundry/logic is the base; China is no longer the growth engine — and that is the quality of this print.

Segment drivers

| Driver | FQ4'24 | FQ1'25 | FQ2'25 | FQ3'25 | FQ4'25 | FQ1'26 | FQ2'26 | FQ3'26 | |---|---:|---:|---:|---:|---:|---:|---:|---:| | Semi Systems ($M) | 5,177 | 5,356 | 5,401* | 5,564* | 4,760 | 5,141 | 5,965 | 7,040 | | YoY % | +6.0 | +9.1 | +10.2 | +13.0 | −8.1 | −4.0 | +10.4 | +26.5 | | SS NG op. profit ($M) | 1,834 | 1,998 | 1,781* | 1,849* | 1,538 | 1,691 | 2,102 | 2,673 | | SS NG op. margin | 35.4% | 37.3% | 33.0%* | 33.2%* | 32.3% | 32.9% | 35.2% | 38.0% | | margin YoY (bps) | −150 | +160 | −190 | −180 | −310 | −440 | +220 | +480 | | SS NG GM % | — | 53.5% | 53.6% | 53.5% | — | 54.5% | 54.8% | 55.4% | | AGS ($M) | 1,639 | 1,594 | 1,420* | 1,463* | 1,625 | 1,559 | 1,665 | 1,781 | | YoY % | +11.4 | +8.0 | restated | restated | −0.9 | −2.2 | +17.3 | +21.7 | | AGS NG op. margin | 30.0% | 28.0% | 26.6%* | 27.3%* | 27.9% | 28.1% | 29.2% | 30.1% | | DRAM % of SS | 23% | 28% | 27% | 22% | 29% | 34% | 29% | 26% | | DRAM $ (calc., $M) | 1,191 | 1,500 | 1,458 | 1,224 | 1,380 | 1,748 | 1,730 | 1,830 | | DRAM $ YoY % | −9.7 | −10.1 | −7.0 | +3.6 | +15.9 | +16.6 | +18.6 | +49.5 | | Foundry/logic % of SS | 73% | 68% | 66% | 69% | 65% | 62% | 67% | 67% | | China ($M, 8-K) | 2,136 | 2,243 | 1,774 | 2,548 | 1,964 | 2,095 | 2,087 | 2,506 | | China % of total | 30% | 31% | 25% | 35% | 29% | 30% | 27% | 28% | | China $ YoY % | −28 | −25 | −37 | +18 | −8 | −7 | +18 | −1.6 |

*FQ2'25 and FQ3'25 SS/AGS are Daloopa restated (200 mm equipment moved from AGS into Systems beginning FQ1'26). DRAM $ = mix × SS net sales.

Driver read. Four things at once; only one is a problem.

1. Semi Systems re-accelerated through the China air-pocket. Same-quarter SS stack +5.3% → +13.0% → +26.5%. FQ4'25/FQ1'26 (−8.1% / −4.0%) was license destock, not a WFE cycle break. SS NG op profit $2,673 million (+44.6%) with op margin 38.0% (+480 bps) and GM 55.4% (+190 bps).
2. DRAM is the incremental dollar; foundry/logic is the base. DRAM $ +49.5% after four mid-teen quarters. Foundry/logic $ +22.9% to ~$4.72B and still 67% of SS. Flash is noise (7% of SS).
3. AGS is compounding in the mid-teens-plus, with margin recovery. $1,781 million +21.7%, NG op margin 30.1% (+280 bps). Hill's caveat: utilization/spares “you can only grow to 100% utilization once.”
4. China is no longer the growth engine — and that is quality. China −1.6% YoY, mix 35% → 28%. Ex-China revenue (calc.) ~+39%. Concentration is still well above 10%; the trajectory of that risk is improving.

Consolidated P&L

| Metric | FQ4'24 | FQ1'25 | FQ2'25 | FQ3'25 | FQ4'25 | FQ1'26 | FQ2'26 | FQ3'26 | |---|---:|---:|---:|---:|---:|---:|---:|---:| | Net sales ($M) | 7,045 | 7,166 | 7,100 | 7,302 | 6,800 | 7,012 | 7,910 | 9,115 | | Revenue YoY % | +4.8 | +6.8 | +6.8 | +7.7 | −3.5 | −2.1 | +11.4 | +24.8 | | NG GM % | 47.5% | 48.9% | 49.2% | 48.9% | 48.1% | 49.1% | 50.0% | 50.4% | | GM YoY (bps) | +20 | +100 | +170 | +150 | +60 | +20 | +80 | +150 | | NG op. income ($M) | 2,063 | 2,190 | 2,180 | 2,245 | 1,947 | 2,107 | 2,536 | 3,096 | | NG op. margin | 29.3% | 30.6% | 30.7% | 30.7% | 28.6% | 30.0% | 32.1% | 34.0% | | OM YoY (bps) | −20 | +110 | +170 | +190 | −70 | −60 | +140 | +330 | | NG diluted EPS ($) | 2.32 | 2.38 | 2.39 | 2.48 | 2.17 | 2.38 | 2.86 | 3.50 | | NG EPS YoY % | +9.4 | +11.7 | +14.4 | +17.0 | −6.5 | 0.0 | +19.7 | +41.1 | | GAAP diluted EPS ($) | 2.09 | 1.45 | 2.63 | 2.22 | 2.38 | 2.54 | 3.51 | 3.17 | | FCF ($M) | 2,168 | 544 | 1,061 | 2,050 | 2,043 | 1,040 | 210 | 2,330 | | Capex ($M) | −407 | −381 | −510 | −584 | −785 | −646 | −635 | −707 |

GAAP is not the operating EPS series. FQ1'25 GAAP $1.45 vs NG $2.38 is a tax/investment hole; FQ2'26 GAAP $3.51 vs NG $2.86 is a mark-up; FQ3 flips the other way (−$0.33). Use NG.

YTD FY26 (FQ1–FQ3) net sales $24,037M vs $21,568M (+11.4%). A FQ4 at the $10.25B midpoint puts FY26 near ~$34.3B, +21% vs FY25 $28,368 million — matching FMP, not a sandbag.

Absolute net sales — 12 quarters ($B)

Revenue and NG EPS YoY — 12 quarters

Annual arc (FY2021–FY2025)

| Metric | FY2021 | FY2022 | FY2023 | FY2024 | FY2025 | |---|---:|---:|---:|---:|---:| | Net sales ($M) | 23,063 | 25,785 | 26,517 | 27,176 | 28,368 | | Revenue YoY % | +34.1 | +11.8 | +2.8 | +2.5 | +4.4 | | Semi Systems ($M) | 16,286 | 18,797 | 19,698 | 19,911 | 20,798 | | AGS ($M) | 5,013 | 5,543 | 5,732 | 6,225 | 6,385 | | NG GM % | 47.5% | 46.6% | 46.8% | 47.6% | 48.8% | | NG operating margin | 31.7% | 30.5% | 29.1% | 29.2% | 30.2% | | NG diluted EPS ($) | 6.84 | 7.70 | 8.05 | 8.65 | 9.42 | | NG EPS YoY % | +64.0 | +12.6 | +4.5 | +7.5 | +8.9 | | FCF ($M) | — | — | 7,594 | 7,487 | 5,698 |

FY2021 was the last real growth year. FY2023–FY2025 was a three-year grind. If FQ4 prints to guide, FY26 is the first +20% revenue year since FY21. FMP: FY2026E $34.35B / $12.81; FY2027E $46.74B / $18.67.

FQ3 was a modest double beat vs the pre-print FMP bar, not a blowout. Street was already above the company midpoint ($8.995B / $3.40 vs $8.95B / $3.36). Clearing Street is a residual overshoot, not just beating a sandbagged guide.

Pattern: consistent beater, compressing magnitude. L12Q NG EPS 12/12. L12Q revenue 11/12 (only miss: FY25Q2 −0.4%, and that quarter printed exactly at the company sales mid). This print’s +1.3% / +2.9% is smaller than FQ2’s +3.0% / +6.7% because Street caught the May raise.

Heatmap — last 8 quarters

This quarter vs consensus and own guide

| Metric | Consensus / guide | Actual (Daloopa) | Variance | Result | |---|---:|---:|---:|---| | Net sales vs FMP | $8.995B | $9,115M | +$120M / +1.3% | BEAT | | Net sales vs co. mid | $8.95B ± $500M | $9,115M | +$165M / +1.8% | Above mid / in range | | NG EPS vs FMP | $3.40 | $3.50 | +$0.10 / +2.9% | BEAT | | NG EPS vs co. mid | $3.36 ± $0.20 | $3.50 | +$0.14 / +4.2% | Above mid / in range | | NG gross margin | co. ~50.1% | 50.4% | +30 bps | Above guide | | Semi Systems | co. ~$6.9B | $7,040M | +$140M / +2.0% | Above co. mid | | AGS | co. ~$1.75B | $1,781M | +$31M / +1.8% | Above co. mid | | Corporate & Other | co. ~$300M | $294M | −$6M / −2.0% | In line |

Mgmt variance story: customers pulled tools in as they found clean-room space (Dickerson: highest sequential revenue growth in company history). Semi Systems is the dollar beat — record foundry/logic plus DRAM/HBM +52% YoY. Value-based pricing plus opex leverage converted a +1.3% sales beat into a +2.9% EPS beat (NG OI +37.9% on +24.8% sales). No one-time P&L item of size.

| Window | Revenue vs FMP | NG EPS vs FMP | Sales above guide mid | |---|---|---|---| | L12Q | 11/12 = 92% | 12/12 = 100% | 10/11 (91%) | | L4Q | 4/4 = 100% | 4/4 = 100% | 4/4 | | L4Q surprise path | +1.7 → +2.1 → +3.0 → +1.3 | +2.8 → +7.7 → +6.7 → +2.9 | Magnitude compressing this print |

AMAT does not issue a fiscal-year revenue/EPS range. It guides one quarter ahead and overlays calendar-2026 directional color (FQ2'26 through FQ1'27, a 14-week quarter). FY2027 is unguided in dollars.

FQ4 FY2026 — first issuance

| Item | Low | High | Mid | Stated YoY | FQ4'25 actual | |---|---:|---:|---:|---:|---:| | Revenue | $9.75B | $10.75B | $10.25B ± $500M | +51% | $6,800M → +50.7% | | NG EPS | $3.82 | $4.22 | $4.02 ± $0.20 | +85% | $2.17 → +85.3% | | Semi Systems | — | — | ~$7.9B | +62% | $4,760M | | AGS | — | — | ~$1.84B | +22% | $1,625M | | Other (display) | — | — | ~$510M | — | $415M | | NG GM | — | — | ~50.4% | +230 bps | 48.1% | | NG opex | — | — | ~$1.58B | — | $1,325M | | NG tax | — | — | ~11% | 2027 ~13% | — | | NG OM (derived) | — | — | ~35.0% | +640 bps | 28.6% |

GM is guided flat vs FQ3 50.4% despite a ~$1.1B sequential revenue step. Hill: display mix + CSE/manufacturing ramp costs. Derived FQ4 OM ~35% is +100 bps vs FQ3 via opex leverage (opex +5.3% vs revenue +12.5% seq at the mid). Believe the GM pause.

The FQ4 ±$500M / ±$0.20 band has not tightened as the run-rate doubled. Revenue has cleared the midpoint in each of the last three printed quarters (avg +2.5%); EPS by more (avg +7.2%). Treat the low as non-informative; the mid is the number; history says the print lands above it.

Calendar-2026 color (the raise that matters)

| KPI (CY26 = FQ2'26–FQ1'27) | Feb FQ1 | May FQ2 | Aug FQ3 | Signal | |---|---|---|---|---| | Semi Systems growth | >20% (Daloopa) | >30% (Daloopa) | “greater than that”; refused 40% | Third raise; still a floor | | Advanced packaging | — | >50% | >70% | +20 ppt | | AGS | double-digit | Mid-teens LT; higher this year | >20% CY; mid-teens LT held | Raise of the year | | PDC | — | “fastest-growing” | >50% | New quantified KPI | | China (SS+AGS) | 27%; China −7% | normalizing | Mix 26%; China $ up in CY26 (28 nm) | Qualitative raise | | Other / Display | ~$250M/qtr | ~$300M | FQ4 ~$510M; ~$400M/qtr through 2027 | New multi-year assumption |

Waterfall — CY26 Systems floor and FQ4 vs residual

| Metric | New guide mid | Post-print FMP residual | vs FMP | |---|---:|---:|---:| | FQ4 revenue | $10.25B | ~$10.31B (FY26E $34.351B − 3Q $24.037B) | −$0.06B / −0.6% (in-line) | | FQ4 NG EPS | $4.02 | ~$4.07 ($12.81 − $8.74 YTD NG) | −$0.05 | | Implied FY26 rev | $34.287B (3Q + mid) | $34.351B | −$64M / −0.2% | | Implied FY26 NG EPS | $12.76 | $12.81 | −$0.05 | | FY27 (no co. $) | “another strong, record year” | $46.74B / $18.67 | Street +36% / +46% with no company floor |

FMP was pulled 2026-08-22 — after the raise. Do not read the residual as the bar the print beat. Pre-print LSEG FQ4 was $9.54B / $3.69 (Reuters, 2026-08-13); the company midpoint is +$710M / +$0.33 above that stale bar.

FQ1 FY27: sequential revenue growth expected, not guided. 14-week quarter → “higher-than-average step-up” in opex. Tax ~11% in FQ4, ~13% in 2027 (Pillar Two) — a ~200 bps earnings-rate headwind.

One-line verdict: Revenue YoY inflected from a −3.5% FQ4'25 trough to +24.8% — two consecutive ~1,350 bp accelerations — with NG EPS at +41.1%. FQ4 guide +50.7% / +85% is a third step against an easy base.

| Metric | FQ4'24 | FQ1'25 | FQ2'25 | FQ3'25 | FQ4'25 | FQ1'26 | FQ2'26 | FQ3'26 | |---|---:|---:|---:|---:|---:|---:|---:|---:| | Revenue YoY % | +4.8 | +6.8 | +6.8 | +7.7 | −3.5 | −2.1 | +11.4 | +24.8 | | Rev Accel (bps) | −70 | +205 | −1 | +90 | −1,121 | +133 | +1,356 | +1,342 | | NG EPS YoY % | +9.4 | +11.7 | +14.4 | +17.0 | −6.5 | 0.0 | +19.7 | +41.1 | | NG EPS Accel (bps) | −215 | +230 | +262 | +263 | −2,345 | +647 | +1,967 | +2,146 |

Accel = change in the YoY rate vs the prior quarter (second derivative — never sequential revenue as growth).

| # | Quarter | What happened | Durable? | |---|---|---|---| | 1 | FQ4'25 trough | Revenue YoY flipped +7.7% → −3.5% (−1,121 bps). SS $4,760M −8.1%. NG EPS −6.5%. | Temporary. Next two prints reversed it. | | 2 | FQ1'26 still negative | Revenue −2.1% (+133 bps — noise). NG EPS flat at $2.38. Inflection had not started. | Bridge quarter. | | 3 | FQ2'26 turn | First re-acceleration: +1,356 bps to +11.4%. Record $7,910M. NG EPS +19.7%. Q3 guide already implied another step. | Held, and built. | | 4 | FQ3'26 second acceleration | +1,342 bps to +24.8%. Highest sequential revenue growth in company history. NG EPS +41.1%. FQ4 guide is a third acceleration against the trough base. | AI foundry / DRAM / packaging looks durable (8-quarter forecasts, >10 new fab projects). FQ4 YoY is flattered by the $6.80B trough — sequential FQ4 vs FQ3 is +12.5% at the mid, still strong, not a +51% run-rate. |

Trajectory over absolutes: a 25% grower that just added 1,342 bps of rate, with the next guide at +51%, is a better setup than a 25% grower rolling over. The quality of the acceleration is high on mix, mixed on base effects. Incremental FQ3 sales vs FQ3'25 = +$1,813M; Systems delivered 81% of it.

Next scheduled print: FQ4 FY2026, company-projected 2026-11-12. The stock-moving debate is not whether AI WFE is growing — FQ3 already printed record $9,115 million. It is whether CY26 Systems growth is “stronger than more than 30%” or the ~40%+ the Street needs to outgrow a ~38% WFE tape, and whether FQ1 FY27 prints sequential growth.

| # | Catalyst | Timing | Consensus / watch | Implication | |---|---|---|---|---| | 1 | FQ4 print + first look at FQ1 | 2026-11-12 (projected) | Pre-print LSEG $9.54B / $3.69. Co. mid $10.25B / $4.02. Post-print FMP residual ~$10.31B / ~$4.07. | Highest-conviction near-term. A $10.0–10.2B print beats stale LSEG and screens in-line vs revised FMP. Watch SS vs $7.9B, GM 50.4%, any FQ1 qualitative. | | 2 | CY26 Systems growth (unquantified third raise) | SEMICON 10/13 and FQ4 | Daloopa series still >30%. Arcuri: Lam WFE ~38% ⇒ outgrow ≥40%; a 40% path implies seq +18 / +12 / +6% into FQ1. | The thing Street is most likely to get wrong. Hill expects FQ1 sequential growth but will not give the number. | | 3 | DRAM / HBM 2H weighting | FQ4 is the first “very significant increase” quarter | Mix 26% even as dollars hit a record. Watch mix and dollars — mix fell FQ1 34% → FQ3 26% while dollars rose. | If DRAM mix does not rise, the 2H DRAM call is a miss even if SS hits $7.9B on foundry. | | 4 | Packaging >70% (was >50%) | CY2026; panel 2027 | +20 ppt raise in 90 days with no dollar disclosure. NEXX (ASMPT) announced 2026-05-03. | Medium-term re-rating; near-term a promise. SEMICON / FQ4 should convert this into a run-rate. | | 5 | SEMICON West / EPIC | 2026-10-12 EPIC unveiling; 2026-10-13 investor breakfast | Dickerson deferred LT growth/OM at $150–175B WFE to this event. 11 EPIC engagements (Broadcom, SCREEN, UC Berkeley added). | Highest-leverage event before the FQ4 print. | | 6 | Citi / Goldman firesides | 2026-09-08 Hill; 2026-09-09 Dickerson | First post-print podiums. Street will press 40% SS and FQ1 seq. | Tactical. Unlikely to replace SEMICON. | | 7 | AGS >20% / PDC >50% | CY2026 | FQ4 AGS $1.84B = +3.3% seq / +22% YoY — the CY >20% is already in the run-rate. | Duration catalyst. If OM stays ~30%, mid-teens LT is conservative. | | 8 | Capacity: double quarterly systems output by 2028 | Intra-year hiring / Singapore | Hill explicitly rejected a $14B quarterly revenue read. FMP FY2028E $56.21B already rhymes on an annual basis. | Bull: FQ3's $7.0B SS is the low end of a 2028 range. Bear: opex/capex run ahead of demand. | | 9 | ICAPS / China mix | CY2026–27 | China dollars flat-to-down YoY ($2,506M vs $2,548M) while the company grew +25% — the mix de-risking Street wanted. | Sentiment-relief, already printing. Risk is a new control list, not the 26% mix. | | 10 | GM trajectory vs DRAM-heavy 2H | FQ4 GM guided flat | Street wanted another sequential GM print. Guide is unchanged seq. | Flat seq GM into the biggest revenue quarter is the margin-skepticism setup. A 50.6%+ print is the bull confirmation. |

Bull confirmation: SEMICON replaces “stronger than >30%” with ≥~40%, or Hill repeats FQ1 sequential growth with enough color; FQ4 SS ≥ $7.9B with DRAM mix up; NG GM ≥ 50.4%.

Bear confirmation: SEMICON is qualitative-only and FQ1 language is dropped; FQ4 SS misses $7.9B or hits it only on Display/China; GM ≤ 50.0%; a new China control tranche on 28 nm.

Live analyst Q&A only (10 names, 20 scored pairs). IR cut the queue after Srini Pajjuri (RBC).

Street did not debate the quarter. It pressure-tested one idea: management has now raised CY26 Systems three times and is telling the Street 2027 is “another strong year,” but will not put a number on the new growth rate, will not guide FQ1, and will not rank 2027.

| # | Analyst | Topic | Grade | What they got / didn't | |---|---|---|---|---| | 1a | C.J. Muse, Cantor | Systems growth now vs last quarter’s “30-plus”; CY27 | Deflected | “Greater than that.” No new floor. No 2027 rate. | | 1b | C.J. Muse, Cantor | Like-for-like pricing / GM into FY27 | Well answered | ~300 bps over 3 years; tool-by-tool reprice; SS already 55.4%. | | 2a | Vivek Arya, BofA | Visibility beyond 8 quarters | Well answered | 5-year roadmap; 8-quarter SKU; cancellation/expedite charges. | | 2b | Vivek Arya, BofA | Why GMs lag a US peer ~150 bps | Deflected | “Portfolios are different” (display). No 150 bp bridge. | | 3a | Stacy Rasgon, Bernstein | Why FQ4 GM flat on a large step-up | Well answered | Display + CSE/manufacturing ramp. Direct causal. | | 3b | Stacy Rasgon, Bernstein | Ramp-cost duration; still ~10 bps seq? | Deflected | “Slow improvement.” Did not confirm 10 bps. | | 4a | Timothy Arcuri, UBS | Commit to systems +40%? | Deflected | Will outgrow. Will not give the number. FQ1 seq growth expected, unguided. | | 4b | Timothy Arcuri, UBS | Doubling capacity = $14B qtr in CY28? | Well answered | “It is capacity. Not a revenue forecast.” | | 5a | Krish Sankar, TD Cowen | Pricing freedom vs capacity cost; January outlook | Deflected | Environment answered; January explicitly not given. | | 5b | Krish Sankar, TD Cowen | 2030 conversations: tech or capacity? | Well answered | Tech 10 years; capacity beyond 8Q; 5-year fab plan. | | 6a | Harlan Sur, JPM | ICAPS growing this year (was flat-to-up)? | Well answered | Digestion “expiring.” ICAPS grows this year and next. China-led 2026; non-China 2027. | | 6b | Harlan Sur, JPM | AGS GM toward 40%, OM mid-30s? | Deflected | Direction yes; neither 40% GM nor mid-30s OM accepted. | | 7a | Blayne Curtis, Jefferies | NAND doubled in the quarter? | Well answered | Strong % year off a small base; 2027 slower; wafer starts still declining. | | 7b | Blayne Curtis, Jefferies | Capex into 2030 / EPIC | Well answered | 2027 capex still “higher than normal,” declines as % of sales. | | 8a | Jim Schneider, GS | Rank 2027: foundry vs DRAM vs packaging | Well answered | Will not rank; all three strong; ICAPS now also grows. | | 8b | Jim Schneider, GS | Any reason CY27 would not accelerate? | Deflected | Clean-room availability is the governor. No 2027 rate. | | 9a | Mehdi Hosseini, Susquehanna | Targeted growth/OM at $150–175B WFE | Deflected | Parked to 13 Oct breakfast. | | 9b | Mehdi Hosseini, Susquehanna | When does NAND add wafers? | Well answered | Upgrades, not wafer starts, “for the next few years.” China is the exception. | | 10a | Srini Pajjuri, RBC | When is panel packaging mainstream? | Deflected | “Pretty significant growth next year.” No date. | | 10b | Srini Pajjuri, RBC | 14-week FQ1 sequential impact | Well answered | Services ratable; systems not; opex steps up because “everybody wants to be paid.” |

What they would answer: mechanism (value pricing, 8-quarter forecasts, ICAPS turn, NAND = upgrades). What they would not: the calendar-2026 Systems number, FQ1 dollars, 2027 acceleration, peer/AGS margin targets, WFE-scenario math, panel timing. Three analysts asked for a January outlook. One non-answer.

Four contradictions, two high-materiality. Applied is consistent on the things it will put on a slide (record print, 8-quarter visibility, leading-edge / DRAM / packaging ≈ 80% of WFE growth, FCF payout 80–100%). The breaks cluster where the CY26 model is most sensitive.

Checked, not flagged: Systems >20% → >30% → higher than 30% (each step spoken as a raise). Packaging >50% → >70% (labeled). AGS long-term low-double-digits → mid-teens (labeled). Tax ~13% → ~11%, then ~13% in 2027 (labeled). Other ~$400M/qtr through 2027 (labeled; FQ4 $510M is a spike).

This is not a rates / CPI / consumer call. AMAT’s macro is AI-capex industrial: hyperscaler spend, wafer-fab utilization, clean-room scarcity as the binding constraint, ICAPS digestion flipping to growth, and China mix stabilizing at 28 nm foundry. Named companies are scarce (AMAT almost never names customers). Broadcom-into-EPIC is the only high-value named-company print this hour.

Headline: Q3 is an AI-WFE acceleration plus ICAPS-digestion-end print. CSP returns claimed; Systems growth raised again above May’s >30%; Broadcom joins EPIC for AI packaging; China 28 nm called up. Lam’s ~38% WFE and a 150 bps GM gap are the scoreboard.

| Name | Relationship | AMAT datapoint | Implication | Confidence | |---|---|---|---|---| | AVGO | Named — new EPIC partner | “Broadcom will join Epic as an innovation partner to accelerate development of advanced chip packaging technologies for next generation AI systems.” | Better for AVGO custom AI ASICs and AMAT packaging. HBM/packaging demand is not NVIDIA-exclusive. | High (named) | | 7735.T SCREEN | Named — new EPIC partner | SCREEN + UC Berkeley; 11 announced EPIC engagements. | Better for SCREEN as wet-process complement. Slightly worse mindshare for TEL if EPIC becomes the default co-dev venue. | High (named) | | TSM | Q2-named EPIC founder; implied foundry | Record foundry/logic; Taiwan geo $2,025M / 22% (+9.9% YoY dollars, −300 bps mix). TSMC 2026 capex raised to $60–64B (Jul print). | Better for TSMC leading-edge capex. Taiwan mix lost share of AMAT’s pie as DRAM/Korea outran it. | High (end-market) | | MU / 005930.KS / 000660.KS | Q2-named EPIC founders | DRAM incl. HBM packaging +52% YoY; Korea geo $1,521M +31.1% YoY. 2H DRAM “very significant increase” as customers expand clean-room. | Better for HBM/DRAM WFE. 2H step-up is wafer-start adds, not only upgrades — opposite of the NAND message. Worse for the “HBM is a 2026 peak” thesis. | High (category) | | LRCX | Analyst-named “Lamb”; implied “US peer” | Arcuri: Lam WFE ~38% so outgrow ⇒ ≥40%. Hill will outgrow, will not commit 40%. Vivek: peer GM ~150 bps higher. AMAT SS NG GM 55.4% is the stack vs Lam systems GM, not 50.4% corporate. | Mixed. Volume/share story is running hotter than the margin-catch-up story. Corporate GM is capped by display + ramp. | High (asked on call) | | KLAC | Implied (PDC / optical) | PDC >50% CY26; new optical inspection to “increase application share.” | Worse on the margin if the optical push is real — that is KLA’s core franchise. E-beam is AMAT’s historical wedge; optical is the invasion. | Moderate (claim, no PDC segment) | | Hyperscalers (MSFT / AMZN / GOOGL / META / ORCL) | Implied CSP capex | “Cloud service providers continue to increase their investments in AI infrastructure. Importantly, many of these companies are already generating positive returns.” | Better for the hyperscaler-capex complex and NVDA/AVGO/TSM upstream. “Positive returns” is the anti-digestion sentence. Management testimony, not CSP-reported ROIC. | Moderate | | NVDA / GPU-HBM stack | Implied | Never named. Proxies: HBM packaging inside DRAM +52%; packaging >70% CY26; hybrid bonding “finally happening.” | Better for HBM/packaging suppliers. Broadcom’s EPIC seat is the reminder that custom ASICs are in the same race. | Moderate | | China 28 nm cohort (SMIC / Hua Hong) | Geography named | China revenue up in CY26 led by 28 nm foundry. Dollars still −1.6% YoY; mix 35% → 28%. | Better for China mature-node WFE; not a leading-edge China AI read. Export-control overlay unmentioned, not gone. | High (geo) | | TXN / ON / STM / IFX / NXPI | Implied ICAPS | Digestion “expiring”; ICAPS grows this year and next; power + photonics bright spots. Non-China ICAPS positive in 2027. | Better for analog/power WFE in 2027. Biggest 90-day narrative flip on the call. Caveat: AGS spares boom is not perpetual. | Moderate | | NAND producers | Implied | Flash mix 7%; strong % year, small base; 2027 slower; wafer starts still declining; new projects only in China. | Worse for a 2027 NAND wafer-start WFE boom. Layer-upgrade cycle is structurally less good for AMAT than for Lam/TEL. | High (category) | | BESI / NEXX | Q2-named; Q3 implied | Acquisitions “add to our strength in packaging”; panel “pretty significant growth next year”; hybrid bonding “only integrated R&D facility.” | Better for AMAT packaging TAM and Besi if hybrid bonding is finally happening. Timing still withheld. | Moderate | | ASML / 8035.T TEL | Implied | Growth framed as materials and packaging, not scanner-count. EPIC+SCREEN is an AMAT-aligned process-of-record. | Neutral-to-slightly worse for litho-heavy WFE mix. TEL competes on etch/dep/clean and coat/develop. | Moderate | | INTC | Implied GAA/FinFET | “Capacity additions in Gate All Around and FinFET drove record foundry logic revenue.” Not named. | Ambiguous. Industry GAA+FinFET, not an Intel booking. Do not treat as an 18A confirmation. | Low |

What management is repeating that the Street still has to underwrite: AMAT is designed into the three fastest WFE pools (leading-edge foundry/logic, DRAM, advanced packaging ≈ 80% of WFE growth in 2026 and 2027); customers are pulling tools earlier as they solve clean-room constraints; 8-quarter forecasts plus 2027 “another strong year” are visible enough to double manufacturing capacity by 2028 (explicitly not a revenue forecast). FQ3 is the first quarter where the revenue second derivative is loud enough that the Street has to engage with that claim rather than treat it as color.

Clean-room-as-governor remains the convenient constraint — it justifies both the 2026 raise (customers found space) and the refusal to guide FQ1 (space still gates 2027). Do not underwrite a $14B quarter.