Ultra Clean Printed a Record Quarter and Watched Its Stock Halve

UCTT beat, pulled $4B of capacity forward six months, called $200B+ WFE for 2027 — and sold 2.2 million shares at $78 while the tape collapsed 50%

UCTT · Earnings Call · 2026-10-27

The quarter that should have been a victory lap

Ultra Clean's third quarter was, on the numbers, the best in company history. Revenue of $759 million rose 17.7% sequentially and 48.8% year-over-year. Gross margin expanded 60 basis points to 17.3%, operating margin jumped 200 basis points to 9.0%, and EPS of $1.15 nearly doubled off the prior quarter. Guidance for Q4 of $800–860 million implies another 9.4% step up. “Revenue was $759 million, a record quarter, up 17.7% sequentially and 48.8% year-over-year.” — Michael Keogh, Chief Financial Officer · 2026-10-27 And the stock is down roughly half in 90 days. The tape shows a single down segment of 17 weeks and about −50%, with a peak of $142.59 on June 30 and a drawdown from it of 60.8%. A $2.5 billion market cap against LTM revenue of roughly $2.4 billion puts the shares near one times sales — for a subsystem supplier growing 49% into a capacity-constrained upcycle. That gap, not the beat, is the story.

What actually changed: the capacity clock moved, and 'excursion' became real

Three months ago, Charles Shi asked when the $4 billion run rate would arrive. James Xiao was explicit: “in the first half of 2027, we will hit that $4 billion run rate in capacity.” — James Xiao, CEO · 2026-08-03 On this call the same number arrived with a new date — Q1, six months early. The ramp readiness doctrine that UCT has preached for several quarters is now a physical fact: “we now expect to be at approximately $4 billion of capacity in Q1” — Michael Keogh, Chief Financial Officer · 2026-10-27, with 15–20% burst capacity behind it and $80–100 million of incremental spend through 2027–28 to get past $5 billion. The 26,000 square feet of clean room space in Malaysia came online early — that is the mechanical reason the quarter beat. Next comes doubled weldment capacity in Cebu, more clean room in Singapore and the Czech Republic, and doubled vertical-integration capacity in Malaysia and Thailand. In other words, UCT is spending against a customer forecast, not an order book. Xiao's answer to Tim Arcuri's question about what he gets in exchange is the single most important line in the transcript:

we actually have very open and constructive discussion with our key customers... so that the future forecasts are shared and the future demand are really kind of openly discussed, and that the future capacity we're going to add are going to be fully utilized by our key customers.
James Xiao, Chief Executive Officer · 2026-10-27

That is an assurance, not a contract. And the second genuinely new item is that supply chain excursion has flipped from hypothetical to observed. In August, Ed Yang asked whether UCT had hit component shortages and Xiao said flatly, “So Ed, the answer is no... we really initiated the ramp readiness campaign internally way ahead.” — James Xiao, CEO · 2026-08-03 This week: “excursion will happen. And it did happen.” — James Xiao, Chief Executive Officer · 2026-10-27 The mass flow controller bottleneck, in Xiao's view, persists another quarter or two. UCT's pitch is that it never missed a module — a reliability claim that only matters if competitors did.

The cash engine is running backwards, and the ATM filled the hole

The uncomfortable part is how the growth is being financed. Operating cash flow was negative $42.3 million; inventory rose $93.4 million to $723.3 million as UCT pre-bought critical components and built safety stock. The trailing figures in the filed statements already showed the pattern: free cash flow (less SBC) of −$64 million in the June quarter, a free cash flow margin of −9.9%, against revenue of $645 million, up 24% year-over-year. Reported gross margin of 16.1% at the time has since ticked to 17.3%. The funding answer was equity. UCT sold 432,000 shares at $80.13 in Q3, then another 2.2 million at $78.45 in the first week of October for $169.1 million, using proceeds to clear the revolver. The ATM program has $193 million left of $400 million. Diluted shares go from 46.2 million to about 48.9 million. Net debt stood at −$359 million at the June quarter-end, with liabilities at 62.6% of assets. The cash-conversion math is stark: free cash flow ran at roughly negative five times net income in the June quarter. Selling paper at $78 and watching the stock settle near $52 is the kind of detail that colors a whole narrative. Management raised roughly $200 million into a de-rating.

Concentration got worse while the story said otherwise

In August, Xiao pointed to diversification as evidence of resilience: “the top 2 customers, as presented with revenue actually reduced from the 64 down to the high 50s.” — James Xiao, CEO · 2026-08-03 This quarter, large customers were back to 61% of revenue. The direction reversed. That matters because Xiao explicitly concedes UCT's revenue "follows the same underlying demand as our largest customers, but not always quarter-to-quarter" — timing gaps that have produced guidance disappointments before. The reassurance is that Services revenue (10% of the total, $77.1 million, up 6.7%) lags products and should accelerate as wafer starts build through 2027–28, with services margin already at 30.1% versus 15.9% for products. Services is the annuity; products is the beta. There is one genuinely new commercial thread: Ed Yang noted the large dep customer expanding into optical inspection, and Xiao confirmed the subsystem content is largely transferable — new SAM, same engineering. The Center of Excellence in Hillsboro now carries eight new program wins across etch, ALD, PVD and lithography, with a similar cohort in qualification. Those productize in two to three years, which is the correct time horizon for a share-gain argument and the wrong one for this tape.

What the market is actually voting on

This is not purely a UCT problem. Look at the global theme tape: the whole complex that UCT sells into has been marked down over 90 days — AI data center exposure is six advancers against fifty decliners, High bandwidth memory is down about 7% across twelve names, and data center capacity is off 8%. Those same themes were up 50% to 100% over 360 days. UCT's −50% is that de-rate with leverage to it. The bull case is arithmetic: $4 billion of capacity arriving in Q1 against guided Q4 revenue of $830 million at the midpoint, a 20% gross margin target attached to that $4 billion, a stated $5 billion run-rate tied to a “$250 billion WFE” — James Xiao, Chief Executive Officer · 2026-10-27 in 2028, and a share price at roughly one times sales. The bear case is that current footprint expansion is being funded with equity and inventory ahead of contracts, that concentration rose rather than fell, and that excursion-driven supply tightness cuts both ways. Both are visible in the same transcript. That is why the stock halved while the company beat — and why the next two quarters of cash conversion matter more than the next two quarters of revenue.