Built by Grok 4.6 High

TTM Technologies reported fiscal third-quarter 2026 results on Thursday, October 22, 2026. This review was prepared Monday, October 26, 2026. Quarter ended September 26, 2026 on a late-December fiscal year that Daloopa maps as calendar-aligned, so the slug is 2026Q3, not FY2026Q3. Next print: Q4 2026 on February 3, 2027.

TTMI | Earnings Review — Q3 2026

Verdict: accelerating, and the acceleration got steeper. Net sales of $1,203.9 million grew +59.9% YoY versus $752.7 million. That is up from +37.4% in Q2 2026 and +22.1% in Q3 2025 — the YoY rate itself rose +2,251 bps versus last quarter. Non-GAAP diluted EPS of $1.46 grew +117.9% versus $0.67, a +4,722 bp step in the YoY rate. Adjusted EBITDA of $242.3 million doubled (+100.4%), and the margin was 20.1%, +400 bps versus 16.1%. Non-GAAP gross margin of 24.8% expanded +330 bps versus 21.5%. Margin expansion did not come along in Q3 2025 (gross margin −50 bps). It arrived this quarter, with the data-center mix.

Double beat, and the widest one in twelve quarters — but not a data-center beat. FMP had revenue at $1,130.4 million and non-GAAP EPS at $1.25. Sales beat by +$73.4 million / +6.5%. EPS beat by +$0.21 / +16.8%. Both also cleared the high end of the August guide ($1.14 billion and $1.27). The pattern is a consistent beater whose magnitude faded through 2025 and has reversed. The quality check is the other EPS: GAAP diluted EPS of $0.79 grew only +58% versus $0.50, a slowdown from Q2's +92.5%. The non-GAAP bridge widened from $0.17 a year ago to $0.67. Epic closed September 30 and is not a meaningful Q3 sales contributor. The add-backs (M&A, restructuring, amortization, stock compensation, FX) are the series to watch as more deals land.

The Q4 sales guide holds the number August already implied, and the EPS beat was not kept. Q4 sales are $1.42 billion–$1.44 billion, about +85% YoY versus $774.3 million. Q4 non-GAAP EPS is $1.77–$1.83, about +157% at the $1.80 mid versus $0.70. That is a further step-up in the YoY rate. It is not a higher Q4 than the old $4.40 billion full-year point already embedded. The full-year sales mid rose about $80 million, roughly the size of the Q3 sales beat. The full-year EPS mid is still $5.00. Boehle's bridge is about $20 million of extra quarterly interest from Epic debt and Ophir financing. FMP, four days after the print, sits on the Q4 midpoint ($1,430 million / $1.78). The street is not fighting Q4.

What the street has not underwritten is the product the multiple depends on. Data center and networking was guided at 49% of Q3 sales and printed 41% — about 10% light in dollars — while the company total beat. Roks called it customer timing, not yield: one of three N+M programs moved right and ramps in the second half of Q4. In the same hour he described N+M as both slightly more than $200 million / slightly under $400 million and about 30% of data center. Those are different numbers. FY2027 is a first look at $6,625 million and $8.50, about +48% sales growth off the new 2026 mid, not a further acceleration. Boehle's "more than 200 basis points" of 2027 EBITDA-margin expansion is Q&A, not a range, and FMP's EBITDA math is only about +140 bps. The call's "48%" description of 2026 growth is light: $4.48 billion on FY2025 sales of $2,906.3 million is +54%.

Tone. October is more specific and less elastic than August. They issued the first 2027 numbers and refused the end-market split. The data-center miss was owned as timing. Neither sounded worried about demand. Both are now managing a balance sheet and a slipped program inside a $20 million Q4 range. SharePoint, OneNote, Outlook, Excel, Visible Alpha, and Bloomberg were unavailable. Consensus is FMP as of October 26, post-print, four analysts.

Columns are quarters. Rows are metrics. Growth is same-quarter YoY, never Q3 dollars versus Q2 dollars. Non-GAAP gross margin, adjusted EBITDA, and non-GAAP EPS are what management guides. The gold rule marks Q3 2026. Italic rows are the change in the YoY rate versus the prior quarter's YoY rate. That is acceleration. It is not sequential growth. Non-GAAP EPS here is the as-reported series the street scored. Later restated tags on three 2024 quarters ($0.28, $0.55, $0.49) are not used.

The dollar bars have stepped up every quarter since Q2 2025. That is not the tell. The tell is the YoY line: a small dip in Q4 2025 (+18.9% after +22.1%) that did not persist, then three widening steps (+1,147 bps, +701 bps, +2,251 bps). Non-GAAP EPS YoY cooled through Q3 2025 even while revenue YoY was still rising, then re-accelerated every quarter from Q4 2025. EPS YoY is now almost double revenue YoY because incremental margin rose with the mix. GAAP did the opposite this quarter.

Why the rate moved — segments, not a one-market story

Commercial is the rate and almost all of the dollars. Of the $451 million year-over-year sales increase, Commercial contributed about $379 million and the Aerospace and Defense segment about $80 million.

Data center plus networking, on a same-quarter basis, has only gone one direction on the recent tape: +23.6% (Q1 2025), +30.9%, +41.4%, +51.9%, about +60% implied in Q1 2026, about +91% implied in Q2, then management's ~+115% in Q3. The 2026Q1–Q2 dollar residuals are calculated (reported sales minus the three other end markets), not a single fundamental. Do not back into Q3 dollars from the rounded mix. Automotive's rounded 7% would misstate the spoken +11%. A&D end-market growth was a steady mid-teens story for years; the operating-segment series is the cleaner recent step-up. Penang is still only expected to approach breakeven in Q4, so the margin jump is volume and mix, not a plant that has already flipped.

Revenue and non-GAAP EPS YoY

Five completed years

FY2026 is not closed. Annual revenue went +11.0%, −10.5%, +9.4%, +19.0%. The trough was FY2023, when networking collapsed. The 2026 quarterly tape is a break above that slope, not a continuation of +19%. Gross margin never stopped expanding on the annual series, but the big step was FY2022. EBITDA margin dipped in FY2023 and then widened.

Green is a beat versus FMP. Red is a miss. There is no in-line band, so Q3 2025 revenue at −0.03% is a miss. The gold rule is this quarter. Street EPS is non-GAAP diluted EPS, not GAAP of $0.79. Scoring GAAP against $1.25 would be the wrong comparison. The heatmap uses the original print, because that is what FMP recorded against the consensus then in force.

Pattern: consistent beater, magnitude improving after a 2025 fade. Last 12 quarters (Q4 2023–Q3 2026): non-GAAP EPS 11/12, with Q3 2024 an exact in-line and zero misses. Revenue 9/12. The three revenue misses are small: −0.5% in Q4 2023, −2.2% in Q3 2024 (the only material one), and −0.03% in Q3 2025. The last four quarters are clean double beats.

EPS surprise in dollars, oldest to newest across those twelve quarters, shrank from +$0.11 in Q1 2025 (+28.2%) to +$0.02 in Q4 2025 (+2.9%), then widened. This quarter's +$0.21 / +16.8% is the largest absolute EPS beat in the window. It is not the largest percentage beat. This quarter's +$73 million is the largest dollar sales beat. Average revenue surprise in the last four quarters is about +5.2%, versus about +3.3% in the prior four. The EPS beat is running ahead of the sales beat because gross margin, operating margin, and adjusted EBITDA margin are all higher year on year. Non-GAAP operating income of $209.0 million was +125.8% versus $92.5 million. Operating margin of 17.4% is about +510 bps versus the 12.3% the CFO cited. There is no FMP consensus on EBITDA, so that line is not scored.

The company-level beat happened even though data center missed TTM's own prior mix guide. William Stein (Truist) said the end market finished about 10% below the Q3 guide. Roks did not dispute the figure. He said customer program timing, not supply or yield. Two of three N+M programs are running. The third ramps in the second half of Q4. The offset was the rest of the book.

The Q4 sales band sits on the level already implied by the August full-year point. The full-year sales raise is the Q3 beat, not a higher Q4. The Q3 EPS beat was not kept. No gross, operating, or EBITDA margin range was issued. FY2027 is a first look at high-40s sales growth, maintained, not a further acceleration. Quarterly YoY is still accelerating into Q4. Both statements are true.

Waterfall — the sales raise and the EPS giveback

August did not publish a Q4 range. It published a Q3 range and a full-year point of approximately $4.4 billion and non-GAAP EPS approaching $5.00. Implied Q4 sales were about $1.43 billion. Implied Q4 EPS, adding the quarters, was about $2.02. The new Q4 sales mid is the same $1.43 billion. The new EPS mid is $1.80. Closed deals help EBITDA. Debt raised for Epic and to finance Ophir adds about $20 million of interest a quarter, which "drags on our EPS." Up from the deals, offset by the debt.

Nine-month sales are $3,053.9 million, the sum of $846.0 million, $1,004.1 million, and $1,203.9 million. Adding the Q4 band reproduces the full-year range. The year is not new information about the nine months already reported. Q4 end-market mix, as a share of the sales guide: data center 49% (about $700 million), aerospace and defense 31%, medical/industrial/instrumentation 14%, automotive 6%. The data-center mix target is the same 49% that just printed at 41%. The $20 million company-wide range leaves little room if the third N+M program moves right again.

Other Q4 plugs from the prepared remarks, not their own series: SG&A about 6.7% of sales (the speech-to-text file says 0.7%; the indexed transcript's 6.7% matches the cost structure), R&D about 1%, interest expense about $33.9 million, tax rate 15–19% (the Q3 guide was 13–17%), stock-based compensation about $18.7 million. The standing FY2026 capex range was not updated: $345 million to $365 million. An additional $400 million of N+M investment starts in Q4 and runs into early 2028, aimed at 2028 production. Some of that can hit Q4 on top of a range that was not reopened.

Pending deals, including Ophir, are out of the revenue and earnings ranges. In Q&A, Roks also said acquisitions are about $500 million of the $6.50–$6.75 billion, with Ophir only a half year if it closes at the end of Q2 2027. Closed-versus-pending scope is not clean. Treat the release as the guide.

Net-sales YoY has accelerated for three straight quarters, from +18.9% in Q4 2025 to +59.9%. Non-GAAP EPS YoY has accelerated with it to +117.9%. GAAP EPS YoY slowed to +58%. The same-quarter Q3 stack is the clean read through seasonality: revenue YoY went +7.7% (2024), +22.1% (2025), +59.9% (2026). Non-GAAP EPS YoY went +27.9%, +21.8%, +117.9%. EBITDA-margin YoY went +160 bps, −20 bps, +400 bps. 2025 was a growth year whose margin did not expand in the third quarter. 2026 is the year the incremental margin showed up.

Inflection points, in order:

1. FY2023 trough. Revenue −10.5%. Networking −45.6%. The rebuild in FY2024–FY2025 was real (+32.9%, then +37.7% on data center plus networking summed) and was not yet the 2026 quarterly rate.
2. Q4 2025 air pocket. Revenue YoY dipped 314 bps to +18.9%. Non-GAAP EPS YoY collapsed to +16.7% (−4,675 bps). It did not persist.
3. Q1–Q2 2026 re-acceleration. Revenue YoY +30.4%, then +37.4%. EPS YoY +50%, then +70.7%. Gross margin was still grinding inside 21.5–22.3%.
4. Q3 2026 break. Revenue +2,251 bps, EPS +4,722 bps, gross margin +330 bps in one print, EBITDA margin through 20% for the first time in the window. Commercial +92.7%. This is the inflection, not another 20–40 bp margin step.
5. The quality split, same quarter. GAAP EPS YoY decelerated 3,450 bps to +58%. The non-GAAP bridge is doing more of the work. Q4 2024 GAAP of $0.05 makes older GAAP YoY rates unusable. The relevant comparison is 2026.
6. Forward, not yet an inflection. The Q4 guide extends the quarterly acceleration (sales about +85%, EPS about +157%). The FY2027 range then flattens annual sales growth to about +48% and slows annual EPS growth from about +103% to about +70%. Near-term acceleration, medium-term plateau.

Backlog led the revenue shape. 90-day backlog YoY went from −4.5% in Q3 2025 ($610 million versus $638.9 million the year before that) to +54.6%, +81.2%, and +113.1%. A&D program backlog was flat from $1.49 billion in Q3 2024 to $1.46 billion in Q3 2025, then $1.9 billion. Management said some N+M volume was pulled into Q3. The Q4 guide is still a large step up from Q3, so the quarter is not only a pull-forward. Cash is the lagging series: operating cash flow of $154.7 million was only +9.1% versus $141.8 million, while free cash flow of $74.8 million rose +75.7% because capex fell to $79.9 million from $99.2 million. Earnings are running ahead of cash, into a year when the $400 million program starts.

Consensus has already moved onto the Q4 range management just printed. FMP is $1,430 million / $1.78, against $1.42–$1.44 billion and $1.77–$1.83. That is not a stale bar. The argument is inside the range. Needham now reads Q4 data-center growth at about 150% YoY, not the 195% it had been modeling, because one program slipped. For 2027, FMP revenue sits inside $6.50–$6.75 billion, and FMP's EBITDA margin only expands about 140 bps, short of Boehle's more than 200. Ophir ($650 million, close targeted Q2 2027) is excluded from every one of those numbers.

ILFA and Merlin are aimed at a Q4 close and are excluded from the guide. They are option value, not the thesis. The $400 million capacity program does not change Q4 EPS. It does change 2027 cash. Q3 free cash flow was only $74.8 million. A cash-funded program plus debt paydown requires the "positive free cash flow next year" line to be real. Ophir close is guided under 2x last-twelve-month adjusted EBITDA. The revolver is for the acquisition, not for the $400 million. Both statements are in the same call and can both be true.

Five analysts, twelve distinct asks. Seven were well answered (58%). Five were deflected (42%). All five deflections sit on one argument: management says the N+M ramp is a customer-schedule issue, not a yield or supply issue, and that 2027 simply repeats this year's growth. The street asked for the dollars, the catch-up bridge, the customer commitment, and the 2027 mix. Those numbers were not delivered in a form that ties out.

What was usable. Fox got 30% of data center today, about 50% next year, three programs, one slipped for a non-yield reason, and five to ten programs over time. Boehle got a pause on large M&A, positive free cash flow next year, and a quantified EPS offset of about $20 million of interest a quarter. Stein got a clean rejection of the yield and supply hypothesis, and a timeline: the third program ramps in the second half of Q4 and is fully in production in Q1. Crawford got lead times of 12 weeks, sometimes 24, sometimes longer, and a clear claim that customers take the CCL and T-glass increases. Boehle then said N+M gross margin is significantly above the corporate average and that gross and EBITDA margins increase by more than 200 basis points next year. Ricchiuti got "about half" of A&D already up the chain, with Epic as processing and Ophir as optics. Stein got that Ophir's 2027 is largely booked (he said over 85%, then 80% — model "largely booked," not a point).

What was dodged, and why it matters. Stein asked for one N+M dollar for Q3 and one for Q4. The sentences do not describe one ramp. Thirty percent of reported data center — 41% of $1,203.9 million, about $494 million — is about $148 million, not slightly more than $200 million. Thirty percent of the $700 million Q4 forecast is about $210 million, not slightly below $400 million. Roy asked how much of that $700 million is new demand versus catch-up of the slip, and whether customers have committed programs against the extra $400 million of tools. He got brownfield mechanics and a repeated 50% mix line. No split, no booked percentage, no take-or-pay. Ricchiuti asked how much of 2027 versus earlier expectations is N+M versus organic aerospace and defense. He got "a substantial part" is N+M and aerospace and defense grows "over 20%." Stein asked for the 2027 end-market split the company usually gives with a revenue guide. Boehle deferred it to the Q4 call. The deferral is dated. It is still not an answer.

The sentence management keeps repeating, and that the questions show the street does not yet accept, is that N+M schedule noise is normal, technical risk is behind them, and 2027 growth roughly repeats 2026. The repetition is the setup. The conflicting dollars are why it is not yet believed. Consolidated guidance accuracy and N+M guidance accuracy are not the same fact. Q3 sales and non-GAAP EPS beat the prior quarterly guide. The end market management had told the street would be 49% of sales printed at 41%.

Four are material enough to change how the quarter should be read. One is a same-sentence arithmetic miss. The October 22 file is a low-quality speech-to-text transcript. A garbled phrase was not treated as a management contradiction unless the conflicting figures are repeated, match a clean earlier transcript, or fail against a reported number. Guidance that was raised, and labeled as a raise, is not a conflict: aerospace and defense from low-to-mid-teens to about 20% including deals, automotive from a mid-single-digit decline to mid-single-digit growth, medical from 35–40% to nearly 70%.

Checked and not contradictions: the acknowledged data-center miss versus the 49% mix guide; book-to-bill of "2.1x" versus 2.13x; adjusted EBITDA margin +400 bps, which matches 20.1% versus 16.1%; Ophir as all-cash consideration funded with cash plus the existing revolver, distinct from "no plans to come to the financial markets"; and the Q3 revenue print of about $1.2 billion and about 60%, which matches +59.9%. Not treated as contradictions because a later sentence repairs a speech-to-text slip: "€1.2 billion" against "over $1.3 billion," SG&A at "0.7%" against a 6.7% structure, and an ungrammatical "roughly 80% on an organic basis" inside a 20% aerospace-and-defense growth rate.

This is an AI-program, defense-content, and materials call. There is no Fed, CPI, consumer-confidence, PMI, or tariff discussion. No hyperscaler, defense prime, laminate vendor, auto OEM, or automated-test vendor was named. Those names are not filled in. "AMD" in the Ophir remarks is aerospace and defense, not Advanced Micro Devices. "Optics" in one sentence is Ophir.

No rates narrative beyond TTMI's own revolver. Zero comments on the Fed. The capacity plan — more than $1.3 billion already invested or committed, plus an additional $400 million from Q4 2026 into early 2028 — is described as funded from operating cash, with no anticipated need to access borrowing facilities. Ophir, $650 million cash, is cash plus the existing revolver. Two different balance sheets. The credit read-through is "draw the revolver, stay inside 2x, harvest 2027 cash," and tell that to a leveraged-finance room on December 1. It is not a sign they need the equity or bond market. Interest expense of $12.8 million was only +23% versus $10.4 million. The larger step is the forward $20 million quarterly drag, which is not in the Q3 print. A realized FX loss of about $4.1 million, blamed on a weaker dollar for the second quarter in a row, is a P&L leak, not a demand comment. It is not large next to $242.3 million of adjusted EBITDA.

Inflation is T-glass and high-end CCL, passed through. Lead times are 12 weeks, sometimes 24, sometimes longer. Costs are up. Roks said customers will take the increases. Q2 already said CCL producers were shifting away from lower-complexity grades, with the pain in automotive, not in the data-center stack. That is better for the high-end laminate and low-CTE glass-cloth category, none named, and worse for anyone still buying the low-complexity grades. N+M average selling price was described as about four times conventional architectures, with price rising faster than cost because of complexity. That is value, not cost-push. The more-than-200-basis-point 2027 margin comment is a management claim, not a printed guide line. It is consistent with gross margin already flipping from −50 bps YoY in Q3 2025 to +330 bps YoY in Q3 2026. It is not yet evidence the November laminate hike will be fully passed through in Q4.

There is no consumer. Auto was 7% of Q3 sales and, on the call, +11% YoY. The full year flipped from an expected mid-single-digit decline to mid-single-digit growth, on higher-value boards, with unnamed competitors reducing their participation. Year-ago auto sales were $79.4 million. A literal 7% of Q3 sales is only about +6% versus that base, so the spoken 7% and +11% do not both tie. Do not build a model line on either figure alone, and do not read it as a vehicle-production signal. The Q2 continuous-glucose-monitoring call-out was not repeated. The live medical/industrial tell is automated test equipment for AI, plus price, at a spoken +97% and an 18% mix. That is better for AI semiconductor test. It is not a hospital-capex story, and no test vendor was named.

One of three AI programs moved right. The cycle did not. Data center and networking at a spoken +115% YoY, commercial book-to-bill of about 2.5x, and a 90-day backlog that went from −4.5% YoY to +113% is the industrial tell. The customers are "seminal AI players," unnamed, and they dual-source. A slip at one program moves TTMI's quarter without saying anything about token demand. Last reported concentration, Q3 2025 only, was two customers at 23% of sales. Do not assume that percentage still holds after a year of this growth. The Q3 miss versus TTMI's own data-center guide is a customer calendar item. It is better for the AI PCB stack as a whole, and it is timing risk for every qualified source, including TTMI.

Defense is a budget and a mix story, not a new prime you can trade off this transcript. Bookings language covered munitions, Golden Dome, maritime radar, restricted programs, and new primes. Golden Dome was named cleanly on both the August and October calls. The Q3 radar proper names are too garbled to assign to a prime. Q2's clean names were APS-153 and the ATP sensor. Aerospace and defense segment sales grew +23.7% to $416.5 million. About 65% of projected 2027 A&D revenue is already visible. About half of A&D is already "up the chain" before Ophir. Programs were called sticky. That is better for U.S. defense content in missile defense, munitions, and sensing, and better for TTMI content per platform if Epic's processing and Ophir's optics actually bundle with the radar franchise. It is not a named-prime print.

About $500 million of the $6.50–$6.75 billion 2027 framework was described as acquisitions. That is under a tenth of the year, and it conflicts with a release that excludes pending deals. Either way, the step-up is still mostly organic AI boards and defense content, not a roll-up of Swiss, German, and UK PCB tuck-ins. China was discussed as a factory method — brownfield halls go up fast — not as a customer, a tariff, or an export-control comment. Penang, mostly medical/industrial with some N+M, is still expected to approach breakeven in Q4.

The business is accelerating, and the acceleration is in the filings: revenue YoY +59.9% after +37.4% and +22.1%, non-GAAP EPS YoY +117.9%, EBITDA margin +400 bps, backlog +113%, and a Q4 guide that extends the YoY rate to about +85% on sales. That is the trajectory, and it is led by Commercial and by data center and networking, not by a one-quarter pull-forward. The company has also been a consistent beater, and the beat got wider after it compressed through 2025.

Hold, because the incremental thesis the street has not underwritten is the part management could not make agree with itself. Data center missed a 49% mix guide by about 10% in dollars inside a quarter that beat the total. N+M was described as both ~30% of that book and as a $200 / $400 million ramp. The EPS beat was absorbed by about $20 million of quarterly interest, so the full-year EPS midpoint did not move. FY2027 is a repeat of high-40s growth, not a further acceleration, and the more-than-200-basis-point margin claim is not in FMP and not in the outlook slide. GAAP EPS YoY slowed to +58% as the add-back bridge widened. The check is near-dated and simple: Q4 data-center mix versus the new 49% guide, one N+M dollar, gross margin through the laminate hike, and whether the third program is actually in production in the first quarter. Until those print, the acceleration is real and the 2027 margin story is not yet a number.