Open in interactive viewer → charts, metric popovers & call review

Texas Instruments Breaks Its Own Peak — and the Stock Says Not So Fast

Industrial hit a four-year high and data center nearly doubled, but a flat Q4 guide and a 14% drawdown expose the market’s durability doubts.
TXN · Earnings Call · 2026-10-21

A real cycle, finally

Texas Instruments did not just grow in the September quarter; it closed a loop that had been open since the last cycle. Revenue was $6.1 billion, up 11% sequentially and 28% year over year, with Analog up 32% and Embedded Processing up 21%. More important, industrial revenue rose more than 40% year over year and about 10% sequentially, finally exceeding its prior 2022 peak. That is the difference between a bounce and a new cycle: “only now in Q3 2026, it reached the same level of revenue we had in Q3 22. So this is a first time we created a new peak in industrial after, you know, 4 years.” — Haviv Ilan · 2026-10-21 Management called it a broadening cyclical recovery, with industrial, automotive, and data center all contributing. Data center was up roughly 95% year over year and is now about 13% of revenue; automotive grew mid-teens year over year. That is not a single-end-market story. It is the broad content story TI has promised through the downturn. The fundamentals had already begun to show it: revenue was $5.5 billion in the June quarter, up 23% year over year, and the September quarter added another leg. Gross margin reached 64% in the quarter, up 280 basis points sequentially, and operating margin was 47%. Yet the stock sits 9.2% lower over the last 90 days and 14.3% below its June 22 peak of $332.28. The tape is not disputing the quarter. It is disputing the durability.

The keyword rotation: from AI to industrial

The company’s own keyword history tells the story of what changed in the narrative. In prior quarters, data center was the headline: it ranked first or second in the keyword trajectory through 2025 and early 2026. This quarter, the top company keywords are cyclical recovery, embedded business, clean room, COVID cycle, gating item, and previous peak. That is not because data center disappeared — it grew 95% — but because TI is folding it into a broader industrial recovery. Haviv Ilan framed the data center business as a breadth opportunity: roughly 10,000 generic part numbers sold into the market, with power density, telemetry, and cooling driving more analog and embedded content per rack. The nuance matters: when asked how much of industrial is really data center, Mike Beckman said only two of the nine industrial sectors — energy infrastructure and test and measurement — are directly data-center-specific. The rest is genuine industrial recovery, aerospace and defense, robotics, and automation. That is a healthier mix than a pure AI-infrastructure supplier, and it helps explain why TI’s data center growth can coexist with a global AI data center tape that has been hit over the last 90 days. The market is de-rating AI infrastructure exposure broadly; TI’s exposure is not GPU-adjacent hype, it is power and signal chain. But the stock is still caught in the same sentiment wash.

Pricing is the new swing factor

The other fresh keyword is annual pricing negotiations. That is new language for TI. For most of the downturn, management insisted pricing was flat to down low single digits and explicitly not a growth driver. In April, Haviv said pricing was stable in the first half and that if demand stayed strong, “it is likely that prices may go up in the second half of the year.” — Haviv Ilan · 2026-07-22 Three months later, that has happened — but only modestly. “We did see some price movement, I would say, mainly on the analog side, in the second half. I would just characterize it as the second half of 26. After a flat pricing environment, year over year in the first half of 26. So we are seeing a little bit of an uptick, not very significant for this year.” — Haviv Ilan · 2026-10-21 For 2027, TI is now negotiating supply commitments with a pricing element, but management deferred details to the January call. That is the tension: the market can see the cycle, but it cannot yet underwrite the pricing tailwind. A flat-to-modest pricing recovery is not enough to justify a 44.8x trailing earnings multiple on its own. It needs volume and mix to do the work.

The COVID echo they keep denying

The most revealing exchange came when William Stein asked about customer inventory and kitting — the “golden screw” problem of the last upcycle. Haviv’s answer was a careful denial:

I would not compare the situations. I think also we are earlier right now in terms of where the cycle is. So we. We will have to wait and see. But, again, from a TI perspective, we are much better prepared in terms of. We have. We can grow our capacity every quarter we have built inventory during the down cycle.

Haviv Ilan · 2026-10-21
He conceded that automotive is the one place where customers have not kept enough inventory, and that some industrial and data center customers have asked for help. That is not a crisis, but it is the first echo of the last cycle. Management’s scar tissue is visible in the prior calls. In April, Haviv warned that 2025 looked strong early and then “it was, whatever you want to call it, a head fake, a false start, or whatever.” — Haviv Ilan, Chief Executive Officer · 2026-04-22 That is why the Q4 guide matters. Revenue is guided to $5.8–$6.3 billion, midpoint $6.05 billion, slightly below the September quarter. EPS is guided to $2.45–$2.79, midpoint $2.62, essentially flat sequentially. A flat Q4 after a 28% year-over-year quarter is not a disaster, but it is not the acceleration the tape wanted. Mike Beckman said lead times have ticked up and backlog is consistent with demand, while Haviv argued there is no gating item like clean-room availability — only tactical tightness in back-end tools. The capacity build is real, but the market is asking whether the demand signal is as real as the supply build.

What would make the stock believe it

TI’s balance sheet and cash engine are not the problem. Trailing-twelve-month free cash flow was $8.4 billion, up from $2.4 billion a year earlier, including $1.6 billion of CHIPS Act incentives. The dividend rose 7% to $1.52 per share, the 23rd consecutive increase. But the valuation is still a debate: 44.8x trailing net income and 13.9x revenue. The P/FCF multiple has collapsed from its 2024 peak because free cash flow exploded, not because the stock is cheap: 54.9x free cash flow, down from 275x at the 2024 peak. That is the setup. If industrial holds above its old peak, data center keeps growing, and 2027 pricing negotiations produce a real tailwind, the earnings power is higher than the trailing multiple suggests. If Q4 is the start of another false start, the stock already voted. The quarter was good enough to prove the cycle is real. The guide was cautious enough to keep the doubt alive.