Built by Grok 4.6 High

Reports Thursday, October 22, 2026. Prepared September 24, 2026. This is the Q3 2026 print, quarter ending September 27. It has not been reported. The hour is not on Intel's IR calendar. Last reported quarter is Q2 2026, released after the close on Thursday, July 23, 2026. Intel's fiscal year is the calendar year.

INTC | Earnings Preview

Q3 2026 has not printed. The baseline is Q2 2026 versus Q2 2025, not versus Q1. Net revenue was $16,128 million, +25.4% versus $12,859 million. That rate stepped up from +7.2% in Q1 2026 ($13,577 million versus $12,667 million), about +1,820 bps of acceleration. Non-GAAP gross margin was $41.8%, +1,210 bps versus $29.7%. Non-GAAP diluted EPS was $0.42, from a loss of $(0.10). GAAP diluted EPS of $(2.16) is the wrong operating compare: the Q2 10-Q records a $12.5 billion fair-value loss on escrowed U.S. government shares. That figure is the filing, not a Daloopa fundamental.

The growth mix is not the company average. DCAI was $6,262 million, +59.0% versus $3,939 million, and DCAI operating income was $2,474 million versus $633 million, about +2,340 bps of operating margin. Intel Foundry was $5,765 million, +30.5%, but almost all of that is internal wafers. External foundry was $293 million. The foundry operating loss was still $(2,089) million. CCPG was $8,877 million, +12.8%, and the call said that was ASP and mix, not units. CCPG operating income of $2,343 million versus $2,053 million is only about +30 bps of margin. Revenue re-accelerated. Client profit did not.

The annual context is a five-year decline that has only just turned. Net revenue went from $79,024 million in FY2021 to $52,853 million in FY2025, −0.5% in the last year after −20.2%, −14.0%, and −2.1%. Non-GAAP gross margin compressed from 57.7% to 36.7%, then ticked +70 bps in FY2025. Non-GAAP EPS went from $5.47 to a loss of $(0.13) and back to $0.42. Q1 and Q2 2026 are the first hard evidence the revenue decline itself reversed. The Q3 guide does not ask the year to keep the Q2 rate.

What the street is underwriting. The July 23 guide, not revised since, is revenue $15.8 billion to $16.8 billion, midpoint $16.3 billion, non-GAAP gross margin 42.0%, non-GAAP EPS $0.38. Versus Q3 2025 revenue of $13,653 million and non-GAAP EPS of $0.23, that midpoint is +19.4% revenue and +65% EPS, and +200 bps of gross margin versus 40.0%. FMP, pulled September 24, is $16.403 billion and $0.394. That is +0.6% and +3.7% above the guide, inside the range, and +20.1% revenue versus Q3 2025. Versus Q2's +25.4%, the FMP bar is about 530 bps slower. The guide midpoint is about 600 bps slower. The deceleration is the base case. It is a tougher compare and a late-quarter supply calendar, not a demand cut in servers.

On the day it was issued, Reuters citing LSEG had the Street near $15.1 billion and $0.27. The guide was aggressive versus that panel. Eight weeks later FMP has walked up onto the midpoint. The sandbag versus the Street is gone. The sandbag versus Intel's own 2026 prints is not: the last two revenue actuals beat the prior midpoint by +11.3% and +12.8% and cleared the high end, and non-GAAP EPS beat by 29 cents and 22 cents. A repeat of that residual would clear both the guide and today's consensus. Management's reason for not underwriting it is supply timing.

Watch items

• Guidance, unchanged. No release between July 23 and September 24 moved the $15.8–$16.8 billion, 42%, $0.38 framework. Segments are Not Guided. Full-year revenue and EPS are Not Guided. What is still in force is non-GAAP opex of $16.5 billion, gross capex of more than $20.0 billion, and NCI of about $250 million in Q3 and again in Q4. The August equity deal does not change the revenue math. It can make the $0.38 stale, because that EPS point was set on the pre-deal share count. Do not haircut it by an invented diluted count.
• Memory and substrates, not a new tariff line. The guide already assumes PC consumption subseasonal in the second half and down low double digits for 2026, on memory price and availability. TrendForce's post-guide DRAM and NAND path still splits the same way: AI-server pricing stays tight, consumer gets smaller further increases. That confirms the assumption. It does not break it. The tariff item in the record is historical. On the Q4 2024 call, part of the revenue upside was described as customers hedging tariffs. Nothing in this window is a new tariff bridge into the Q3 P&L.
• Supply linearity is the quarter. Wafer outs were ahead of the plan from 90 days earlier, and 18A yields were ahead of the March targets, but supply growth was described as skewed to late Q3 and into Q4, especially servers. Client revenue was guided flattish, CCPG up a bit on edge, and Intel still short of demand in Q4. A miss explained as substrates or memory timing fits the caveat. A miss that is client demand below the already-reduced plan, or server shipments that do not show the intra-quarter ramp, does not.
• 14A is a credibility test, not a revenue line. PDK 0.9 was on track for October. At Deutsche Bank on August 26, David Zinsner said external talks had moved to "how much capacity" and that Intel has conviction it will land external 14A customers. He did not name one, and he said Intel will not be the one to announce a foundry customer. External foundry of $293 million cannot move a $16.4 billion consensus. Silence is the base case. A name would be upside to how expectations were set.
• Who speaks first. Only two company-scheduled prints sit ahead of Intel: ASML on Wednesday, October 14 (in ASML's own interim-report calendar) and TSMC on Thursday, October 15, before the U.S. open, call at 2:00 a.m. Eastern, on TSMC's own site. Micron's September 30 date is an FMP and pattern estimate, not a release read this session. Lam's October 21 FMP date disagrees with a Zacks estimate of October 28. AMD does not print first (FMP November 3). AMD's August 4 quarter is already known: data-center revenue +107% and a claim of x86 server revenue-share gains. That raises the share question. It does not clear Intel's wafer calendar.

The beat habit is intact. Revenue beat FMP in 10 of 12 quarters from Q3 2023 through Q2 2026, and in 4 of 4 recently, with the surprise widening to +11.7% in Q2. Non-GAAP EPS was 9 of 12 and 4 of 4. The base case is another beat of a bar that now equals the guide, not another +9% to +12% surprise off a stale estimate. Gross margin is the tight line versus the quarter just printed: 42% is only about 20 bps above the 41.8% just printed, with the Q2 inventory reserve going away and early Panther Lake and Granite Rapids giving it back. Versus Q3 2025 the same 42% point is still +200 bps. The 20 bps is a level compare, not the growth rate.

The Q2 beat did not hold on the tape. Daloopa's split-adjusted close was $100.23 on July 23, the pre-print close, and $92.32 on July 24, −7.9%. Those closes are not a live price, a market cap, or a multiple, and they are not src-linked. The after-hours pop is not the move. No options feed was available, so the implied move is n/a.

The outlook was spoken on July 23 and stored by Daloopa on the issuance quarter, calendar 2026Q2. It applies to Q3. Nothing since has revised it. Classification: in line with FMP on September 24, a hair under the average, and still conservative versus Intel's own 2026 beat pattern. Operating margin and EBITDA are Not Guided. CCPG, DCAI, Foundry, and external foundry are Not Guided as numbers.

FMP revenue low, average, and high are $16.252 billion, $16.403 billion, and $16.643 billion (23 analysts). EPS is $0.374, $0.394, and $0.414 (24 analysts). The whole street revenue range sits inside Intel's range, in the upper half. No FMP revenue estimate is down at the $15.8 billion floor. The company is still carrying downside the Street has abandoned. That is the remains of a conservative range, not a conservative midpoint.

The same midpoint was well above the Street on July 23. The revision since then is the Street marking the Q2 beat and the new guide. FY2026 FMP revenue moved from $60.43 billion on the July 26 snapshot to $63.078 billion, and FY2026 EPS from $1.272 to $1.536. Intel never guided those full-year revenue and EPS figures. H1 2026 revenue of $13,577 million plus $16,128 million is $29.705 billion. Adding FMP Q3 and Q4 revenue foots to the FY2026 average. The street's year is the Q3 midpoint plus a higher Q4, not a haircut.

Why conservative versus Intel, not versus FMP: the last four guides were beaten on non-GAAP EPS and non-GAAP gross margin every time, and revenue beat the high in three of four.

Eight of the last nine quarters beat the revenue midpoint. The only miss was Q2 2024, and it was still inside the range. Four of the last five cleared the high end. The average surprise stepped up from about +4.6% over Q3 2024–Q2 2025 to about +7.8% over Q3 2025–Q2 2026. The regime change is the 2026 pair. Gross-margin misses in that longer window are the charge quarters, Q3 2024 at 18.0% and Q2 2025 at 29.7%. Clean quarters have been beating by a few hundred basis points. A flat 42% guide after that record is the conservative-but-explained case, not another 300 bp sandbag. Lip-Bu Tan's "seventh consecutive quarter" is fair as a statement about the set of financial expectations. It is not a seven-for-seven EPS record. Q2 2025 missed the EPS guide. Revenue did not.

Periods are columns. Growth is year over year, never the sequential rate across different quarters. Q3 2026 is a guide, not a print. Internal estimates are n/a on every row.

Current quarter — Q3 2026

% diff is (consensus − guide mid) / guide mid. Positive means the street is above the guide. The Q3 low of $15.8 billion is $328 million under Q2 revenue of $16,128 million. The midpoint is $172 million above it. Management can call the quarter supply-constrained and still print a high-teens YoY rate because the year-ago base was $13,653 million, not the Q2 run-rate. Q3 2025 itself was only +2.8% versus $13,284 million. Guided growth is faster than that quarter and slower than Q2.

What the unguided drivers were doing into this print:

DCAI went from +22.9% to +59.0%. Foundry went from −2.7% to +30.5%. CCPG decelerated from +16.4% to +12.8%. Holding the Q2 DCAI dollar level into Q3 would still be about +52% versus Q3 2025's $4,117 million. There is no street segment number. External foundry also printed $222 million in Q4 2025 and $174 million in Q1 2026. First-half 2026 external revenue is $174 million plus $293 million, or $467 million, already above full-year 2025 external revenue of $307 million. The Q2 print is 5% of that quarter's foundry revenue ($5,765 million), not 5% of the half, and it is not guided.

FQ+1 guide — Q4 2026

Intel guides one quarter out. Q4 revenue, EPS, gross margin, operating margin, and EBITDA are Not Guided. Anything on those lines on October 22 will be new. The one Q4 number already spoken is NCI. Do not back into a Q4 revenue guide. There is no full-year revenue guide to subtract from.

FMP Q1 2027, which is not a guide either, is $16.798 billion of revenue and $0.401 of EPS. The street path after Q3 does not roll over.

FY+1 guide — FY2027 is mostly Not Guided; FY2026 is what is still in force

First-half 2026 gross capex was $4,963 million plus $2,652 million, or $7.615 billion. The year-ago first half was $6,203 million plus $4,492 million, or $10.695 billion. That is −28.8% year over year, about −29%, against a full-year floor above $20 billion. Reaching $20 billion requires more than $12.4 billion in the second half, versus about $7.0 billion in the second half of 2025. The capex raise is a commitment, not yet a cash outflow. Q3 is the first quarter that can show tooling orders turning into spend. Zinsner also said 2026 tooling spend is up about 40% versus 2025. That 40% is transcript only.

FY2025 non-GAAP gross margin of 36.7% and non-GAAP operating margin of 5.5% are the full-year base. The Q3 gross-margin point of 42% is +530 bps versus that full-year rate. It is not a full-year guide.

Five-year annual trajectory

The decline got less bad before it turned: −20.2%, −14.0%, −2.1%, −0.5%, then a Q2 print at +25.4%. Gross margin bottomed in FY2024 and ticked +70 bps in FY2025. Operating margin compressed for three years and then recovered +600 bps. EPS YoY is a rate only while it stays positive. The loss year is a swing, not a percent. The Q3 question is whether the quarterly re-acceleration survives a supply-timed pause, not whether the five-year hole has already been filled. Non-GAAP figures above are the as-originally-reported tags. A later recast of FY2021 non-GAAP exists and does not change the direction.

Eight reported quarters

Company revenue YoY went −7.4%, −0.4%, +0.2%, +2.8%, −4.1%, +7.2%, +25.4%. The Q4 2025 dip was a false start, not a new downtrend. Into this print, DCAI's YoY rate stepped from +22.4% to +59.0%. Foundry's YoY rate has risen for three straight quarters, from −2.7% to +30.5%. CCPG stepped from +1.3% to +12.8% and is still below Q3 2025's +16.4%. Q3 2024 gross margin and EPS are the Intel 7 impairment, not a demand print, so the Q3 2025 margin compare stays labeled charge base instead of +2,200 bps. The clean margin step in this window is Q1 2026 at +180 bps. Q2 2026's +1,210 bps is versus the other charge quarter. Non-GAAP EPS YoY is +15.4% in Q4 2025 and +123.1% in Q1 2026 off a $0.13 base. Q3 2025 and Q2 2026 are swings off losses, not percents. Segment YoY before Q3 2025 is left blank where the year-ago pair was not in this table.

Confidence is highest where the Q3 print will not settle the argument, and only medium where it will. The guide reads conservative against seven beats of Intel's own outlook and against unconstrained server demand, and roughly in line with a street that is only a little above the midpoint. It is not conservative against supply. They have said the constraint is inside the factory calendar and the substrate and memory lines, and that the more bullish statements are about 2027 and 2028.

What they bound the quarter to, on July 23. David Zinsner guided revenue of $15.8 billion to $16.8 billion, and at the midpoint gross margin of 42%, a tax rate of 11%, and EPS of $0.38, all non-GAAP. The bridges under that sentence:

• Supply, not demand, sets Q3 revenue. Wafer outs beat the plan from 90 days earlier. Supply growth is skewed to the end of Q3 and into Q4, especially servers. Company revenue is therefore roughly flat with Q2. They will still be short of demand in Q4. He would not guide Q4.
• The shortage is no longer only a front-end wafer story. Substrates, T-glass, and memory are the chunkier constraints. They will not catch up this year.
• Client is the shock absorber. PC consumption is subseasonal in the second half and down low double digits for 2026. Q3 client revenue about flat, because customers have been running CPU inventory tight and will rebuild some of it. Q4 client revenue turns soft as wafers move to data center. CCPG "likely up a bit" on edge, not on PCs. Q2 client revenue of $7,700 million was +16.7%, and he told Stacy Rasgon the beat was largely ASP.
• Server units were raised again. Strong double-digit industry unit growth this year and next, with the sentence now extended into 2028. In April it stopped at 2027. CPU-to-GPU was described as almost in parity and able to skew further toward CPUs. Granite Rapids on Intel 3 is the tightest product.
• Gross margin is flat for a mix reason. The Q2 client inventory charge for stranded match-sets does not repeat, which is a lift. Panther Lake and Granite becoming a larger share, still below the corporate average, gives it back. Panther Lake primary-SKU cost was said to be down roughly 50% year to date, with another roughly 20% expected this year. The 40% to 60% incremental gross-margin drop-through is still the rule of thumb, and the Q3 guide sits at the low end of that range.
• Foundry is still a drag inside a better P&L. External foundry of $293 million and an operating loss of $(2,089) million. 14A risk production for internal products is the second half of 2027. High-volume manufacturing was committed in Q2 for 2028. PDK 0.5 is done. PDK 0.9 was on track for October. The number Tan put on 18A-P versus 18A is about 5% more performance, not a secondary frequency claim.
• Capex above $20 billion because they believe commitments are real, not because a foundry logo was named. He would not split internal versus external wafers. If the upside case happens, they might still need the capital markets. August then did that.

Tone across the last four calls. Q3 2025 was repair: fourth straight beat, "a long way to go," capex still not to be spent ahead of customers, 18A yields adequate for supply and not yet for margins. Q4 2025 was "we are tight and we are sorry": Q1 was the trough, would be above seasonal if supply existed, 14A still held back. Q1 2026 was the tone break: "a year ago the conversation was whether we could survive; today it is how quickly we can scale supply," demand ahead of supply by an amount that "starts with a B," capex reframed as flat rather than down. Q2 2026 is "we will spend to the shortage": seventh straight beat, strongest revenue growth in more than 15 years, capex through $20 billion, 2027 higher again, 14A high-volume committed for 2028. The caveat moved from "customers may not show up" to "we cannot ship what customers have already asked for, and the shipments we can make are timed late in the quarter." That is a more bullish problem. It is a harder one for a quarterly model, because the swing factor is factory linearity, not a book-to-bill they will number.

What they did not drop: no Q4 number, still short of demand in Q4, PC units a planned decline, 18A products below corporate margin, foundry still loss-making, no named 14A customer, and 2027 free cash flow not a given once back-end prepayments are in.

What changed after the guide. The model did not.

• August 10–12. Intel priced 210,526,315 shares at $95, upsizing a $15 billion launch to $20 billion. The greenshoe of 31,578,947 shares was exercised in full on August 11. Net proceeds were about $19.7 billion before the option and about $22.6 billion with it. Use of proceeds is general corporate purposes, including capex. Lip-Bu Tan bought shares at the offering price. That removes the "we may have to tap the markets" caveat as a 2026 event. It does not change Q3 revenue or the 42% margin. The prospectus share count goes from about 5,043 million on June 27 to about 5,285 million with the option. The $0.38 guide was set before that count.
• August 26, Deutsche Bank. Zinsner did not change Q3. He said the raise precedes the investment and is the opposite of a defensive balance-sheet deal. 14A defect density is ahead of Intel's own curve and, in his telling, better than any node since 22 nm. Internal product groups are designing on 14A. External engagement "has significantly increased." Conviction, still no name. Analyst Day is not this year. Server undersupply lasts this year and probably next year. He hopes 2028 is when they are in a decent position, and he said they want to more than double Ireland Intel 3 output next year. That quantifies the July 13 €5 billion Leixlip expansion, which is before the guide, not a new one. Gross margin: "comfortably in the 40s. Low 40s." Next targets are the mid-40s, the high 40s, and eventually something that starts with a five. He would not date 50%. Foundry breakeven is an internal end-of-2027 target that can slip to the end of 2028 if customer ramps force more start-up spending. He described the loss as running around $2.5 billion a quarter. That is shorthand. The reported Q2 loss is $(2,089) million. 2027 capex still has no number. He sent the internal ask back. The October call is the next time a digit could appear. He has not promised one.
• September 7, SPIE with ASML. High-NA EUV continues in high-volume manufacturing, with more than one million wafers processed across certification, R&D, and select layers of a subset of Core Ultra Series 3. That supports "18A is in real volume." It is not an external customer, and it does not mention 14A.
• Memory after the guide still fits the guide. TrendForce on July 3, before the call, had Q3 conventional DRAM contract prices up 13–18% quarter to quarter. The September 24 4Q26 outlook still has AI-server demand lifting contract prices, with consumer seeing only small further increases. A European laptop forecast of down 6.4% in Q3 and down 20% in Q4 is one region. It lines up with "client market down in Q3, our own client revenue soft in Q4." No competitor item in this window forced a public guide cut.

The tell on October 22 is whether Panther Lake and Granite are still the margin excuse, or whether the cost-down he quantified is showing up, and whether a midpoint miss is timing or demand. A named external customer would be upside to the way management set expectations, not a condition of the range.

FMP's Q3 range is only about $0.4 billion wide and sits inside the guide. The optimism that was revised lives in the annual file. The catalysts below are being capitalized into 2026–28. Q3 itself is a hold-the-new-run-rate quarter.

Three checks decide whether the setup is right. If DCAI is only a hold of $6,262 million and client is flat, the quarter lands on consensus. Anything well above it means wafers out ran ahead of the guide again. PDK 0.9 without a customer name is what the Street has already chosen not to put in the quarter. First-half capex down about 29% against a full-year guide above $20 billion means the cash has to show up, or the 2027 figure has to. Not in the model, and not treated as Q3 fundamentals: the 2025 U.S. government stake, the 2025 Nvidia and SoftBank investments, and unconfirmed Apple foundry reports.

The read-through that can still change the setup before Intel speaks is TSMC on October 15 (company Q3 guide $44.6–$45.8 billion, gross margin 65–67%) and ASML on October 14 (company guide €11.0–€12.0 billion, gross margin 55–57%). Do not treat FMP's dollar estimate for ASML as that euro guide. Micron's September 30 row is the memory price check if the date holds. It does not clear x86 share or 18A yield.

Window is after the July 23 print through September 24, plus three items from July 13–21 that still set the Q3 debate. The July 23 release, the call, and the July 24 10-Q are excluded. June rumors of a Google TPU volume order, an Nvidia 18A commitment, and an Apple foundry arrangement were not confirmed by those companies or by Intel in this window and are not restated as facts. Daloopa's document index has no 2026Q3 filings. The August equity 8-K is on Intel IR and EDGAR and was not yet in that index.

What did not become news: no CEO or CFO change, no guidance withdrawal, no delayed Q3 date, and no new insider-selling cluster in the coverage checked. Press lists of companies "evaluating" 14A are not in the Q2 filing language, which stays at "potential significant customers," and they are not in Zinsner's August remarks. Treat them as unevidenced.

The quarter's news is a financing and roadmap story. Intel chose equity, including a fully exercised greenshoe, to fund capex management had already stepped up, including Ireland Intel 3 tools and the 14A commitment. Named external progress is narrow. Fortinet is an Intel 4 customer with no disclosed economics. Socionext is a development partner and still a TSMC customer. Altera's confidential filing is the cleanest near-term way to mark the 49% stake. The call question is unchanged from July: whether management will name a volume, a node, and a customer, or whether external foundry stays a small base while internal server CPUs carry the growth.

Consistent beater, and the revenue beat is still getting larger. Over the 12 reported quarters from Q3 2023 through Q2 2026, Intel beat FMP revenue 10 of 12 (83%) and FMP non-GAAP EPS 9 of 12 (75%). The last four quarters are 4 of 4 on both lines. The two revenue misses were small, −0.6% and −0.7%, both in the first half of 2024. Rolling four-quarter average revenue surprise went from +1.3% to +3.9% to +6.6%. Q2 2026 was the widest revenue beat in the window at +11.7% ($16,128 million versus a $14,435 million estimate). Non-GAAP EPS surprise in cents went +6.3¢, then −10.4¢, then +19.0¢. The middle block is the impairment window, not a fading operating beat. Q2's +21.0¢ is still large, and it is narrower than Q1's +27.1¢. Q3 2026 is not on the grid.

FMP scores non-GAAP diluted EPS. GAAP diluted EPS is not the street number. In Q2 2026 GAAP was $(2.16) and non-GAAP was $0.42. Scoring GAAP would turn that quarter into a miss. It is not the number FMP stored, and it is not the number management guided. The mark-to-market added back to reach non-GAAP EPS was $2.45 in Q2 2026 and $0.21 in Q1 2026.

Revenue color tracks the percent surprise. EPS color tracks the cent surprise, not the percent. A +1,812% cell on a $0.007 estimate is +12.3 cents and should not look larger than Q1 2026's +27.1 cents. Darker green is a larger beat. Red is a miss. Anchors are #1E8449 and #C0392B.

Read left to right. Two pale revenue misses in the first half of 2024, then eight green cells that get darker. The darkest revenue cells are the last two prints. The EPS row is not a straight darkening: a large 2023 beat, two red cells in 2024, one more red cell in Q2 2025, then four green cells. The darkest recent EPS cell is Q1 2026, not Q2 2026. Q2 is still a beat on both rows.

Revenue surprise magnitude is improving, in percent and in dollars. Inside the last four, Q4 2025 was a pause at +1.8%, then the beat widened again. The level under the beat is also accelerating: revenue YoY went from −4.1% to +7.2% to +25.4%. EPS surprise deteriorated in the charge window and then recovered. It did not widen further in Q2. The estimate itself moved from roughly zero to $0.21, so the same cent beat is a smaller percent. Percent comparisons across those quarters are not the trajectory.

The red EPS cells are not a hidden operating miss dressed up later, except where management said so. Q2 2024 was a gross-margin miss they disclosed as eight cents below the April outlook, actual non-GAAP EPS $0.02. Q3 2024 left an Intel 7 impairment inside non-GAAP, actual $(0.46). Q2 2025 left impairment and period costs inside non-GAAP, actual $(0.10), and Zinsner said the ex-charge figure would have been ahead of the guide. The record does not restate those quarters. From Q3 2025 on, the explanation shifts from charges to supply, mix, and price, and the street was anchored near the company's outlook, so the cent beats got large whenever the guide was breakeven or $0.08 and stayed large after the guide moved to $0.20.

Into October 22, FMP is $16,403 million and $0.394, against a guide the street has now joined. That is a different setup from Q1 2026, when FMP EPS sat at about two cents against a breakeven guide and the company printed $0.29. A repeat of the Q2 dollar revenue beat would be about $18.1 billion, above the $16.8 billion high. Nothing in the Q2 remarks supports using that dollar gap again. The base case is a beat of a bar that equals the guide. On EPS, that is more likely to look like Q4 2025's single-digit cents, or Q2's low-20s cents if mix and price repeat, than like Q1's +27 cents off a one-cent estimate. The habit that has held for seven quarters is beating the company's own outlook. The habit that has held for four quarters is beating FMP on both lines. Neither habit says the beat keeps widening once the street is on the guide. Revenue YoY if the company merely hits the $16.3 billion midpoint would still be about +19% versus $13,653 million. That deceleration is the comp, not evidence that the streak has broken.