AI Equipment's "Huge Upside" Is Three Different Bets, Not One Trade

The prevailing pitch for AI semiconductor equipment has an appealing simplicity. Hyperscalers are spending on chips; the chips have to be built, packaged, and tested; so nearly every company selling into that build-out looks like a buy. A widely circulated screen taps into that energy, tagging three names—FormFactor, Amkor, and Ultra Clean—with a top Strong Buy rank and the promise of huge price upside.

That grouping is the mirage. These three companies do not occupy one layer of the chip industry. They sell three different things at three different points of the AI supply chain, and each converts the same wave of capital into shareholder returns by a different mechanism. Ranking them together as "AI equipment" is like ranking a specialty-tool maker, a construction contractor, and a steel distributor as one "construction" trade: the boom underneath is real, but who actually keeps the money depends entirely on which one you own.

The driver is worth stating first because it sets the boundaries. For equipment vendors, unlike memory makers, this is genuinely a demand-and-capex cycle rather than a supply-discipline pricing cycle. SEMI projects 2026 total equipment sales of $165.9 billion, up 23.2%, and three consecutive record years to $229.5 billion by 2028, with the growth concentrated where these three sit: wafer-fab equipment up 23% to $143.9 billion, test equipment up 31% to $15.3 billion, and HBM-related DRAM equipment up 39%. Through the first half the industry posted two consecutive record quarters, with Q2 billings of $40.5 billion up 23% year over year. That is a real, funded build driven by customers' capacity spending, not by anyone's restraint.

It is also the risk made visible in the tape. All three names are up sharply for the year—Ultra Clean roughly +195%, FormFactor +107%, Amkor +31%—yet each is down 11% to 20% over the past month and well off its high. The market is not disputing that the cycle exists. It is starting to price what happens to a capex boom if the AI build slows, the one variable none of these companies controls.

Company	What it sells	Gross margin	Own capex (TTM)	P/E	EV/EBITDA	YTD 2026
FormFactor	Probe cards for HBM and advanced logic test	~43%	~$44M	78x	~57x	+107%
Amkor	Packaging and assembly services	~14%	~$1.37B	23x	~10x	+31%
Ultra Clean	Gas-delivery subsystems and components	~16%	~$47M	n/m (loss)	~27x	+195%

The test-content play prices in its own quality: FormFactor.

FormFactor makes the probe cards used to electrically test and sort chips, notably the stacked high-bandwidth-memory modules that feed AI accelerators, a segment that supplies the bulk of its revenue. Its economics are the best of the three by a wide margin: roughly 43% gross margin, high-teens operating profitability, essentially net cash, and capital-light, with only about $44 million of trailing capex. Test content rises as memory stacks grow, and SEMI expects test equipment to be the fastest-growing equipment segment. Momentum is real—Q2 2026 earnings per share of $0.82 beat the $0.61 consensus by more than a third. The flaw is the price: about 78x trailing earnings and 57x EV/EBITDA. The market has already paid for the quality and the momentum, so strong execution from here mostly earns a multiple it already carries.

The packaging-capacity play buys growth with capex: Amkor.

Amkor is the outsourced assembler and packager that actually plants, stacks, and bakes the chips. Advanced packaging is the industry's current bottleneck, and Amkor holds strategic ground around it: a 10-year agreement with TSMC to build advanced packaging and test in Arizona, and a $1.5 billion multi-year partnership with NVIDIA. The business is growing accordingly, with computing revenue up about 26% and advanced products at $1.56 billion, up about 27%. But the economics invert FormFactor's. Gross margin sits near 14%, operating margin near 8%, and the roughly $1.37 billion of trailing capex poured into new packaging lines pushed free cash flow negative. Amkor looks cheap—about 10x EV/EBITDA and 23x earnings—precisely because the market discounts a business that must keep re-spending its profits on capacity. Its upside is conditional: packaging utilization and pricing must outrun the lines it keeps building. If they do, the discount closes; if not, the capex compounds into thin returns. The stock, down nearly half from its high, is where that doubt shows up.

The subsystems play is a leveraged bet on the front end: Ultra Clean.

Ultra Clean sits the farthest from the finished chip and the closest to the building engine. It supplies the gas-delivery subsystems and precision components used inside wafer-fab equipment and the fabs themselves, which means it holds little pricing power of its own; its revenue is a derivative of front-end capital spending. The cycle is feeding through—Q2 2026 revenue of $644.9 million, up from $533.7 million the prior quarter, and a swing to $8.7 million of GAAP net income from a $17.9 million loss. But trailing operating margin is still negative, and inventory roughly doubled to $629.9 million, absorbing cash. Priced near 27x EV/EBITDA on forward earnings and up almost 195% for the year, this is the most cyclical of the three with the thinnest cushion if front-end spending hiccups.

The ranker's problem is not that it picked the wrong companies. It is that it compressed three unrelated economics into one label. FormFactor sells test content, and its quality is already in the price. Amkor sells packaging capacity, and its discount is the market's price for the capital it must keep spending. Ultra Clean sells the front-end cycle, and it currently has no margin cushion if that cycle stalls. The uniform claim of "huge upside" is not uniform. The real test in each case is whether execution outruns the price already paid—FormFactor's growth beating its 78x multiple, Amkor's utilization beating its capex, Ultra Clean's volume beating its margin gap. That is the difference between a screen and a decision, and it is the number to reconcile before treating a Strong Buy rank as a reason to own the stock.