Citigroup’s interpretation: The device bull market will reach $250 billion, and the real test will come in 2027.

TL;DR

• According to market reports, Citigroup expects the global WFE bull-case scenario to rise to $250 billion by 2028.
• TSMC, Samsung, and Intel account for 55% of global WFE spending in 2025; earnings guidance will determine how much additional upside can be raised.
• Equipment stocks that benefit still depend on the durability of AI demand, whether Samsung’s investments land as planned, and Intel’s foundry progress.

TSMC, Intel, and Samsung will successively disclose their Q2 results in mid-to-late July, and semiconductor equipment stocks are also facing another test of capital expenditure expectations. According to market reports, ahead of earnings season, Citigroup remains bullish on wafer-fab equipment spending and believes AI/HPC demand is boosting investment in advanced process nodes, memory, and foundry operations. For investors, WFE refers to wafer-fab equipment spending, covering key equipment purchases such as lithography, etching, deposition, and testing, which directly affects orders and revenue for equipment companies including Applied Materials, Lam Research, and Teradyne.

The focus of the equipment cycle shifts from 2026 to the following two years

In the earlier phase, semiconductor equipment stocks rose mainly due to investment expectations driven by AI servers, advanced packaging, HBM, and ramp-ups of advanced logic process nodes. Now the market is asking whether capital expenditures can be revised upward from 2026 and continue to expand in 2027 and 2028.

According to market reports, Citigroup’s WFE bull-case scenario is approximately $145 billion in 2026, approximately $200 billion in 2027, and approximately $250 billion in 2028. The broker report also shows that TSMC, Samsung, and Intel together account for about 55% of global WFE spending in 2025. If the three companies maintain or raise their medium- to long-term capital expenditures, there will be room for the equipment cycle to rise further.

The next few earnings releases will provide more direct clues. TSMC will release earnings on July 16; Intel will disclose results after market close on July 23. Samsung already issued Q2 earnings guidance on July 7, and will hold an earnings call at 10:00 KST on July 30. The market will be watching not only for quarterly revenue and profit, but also for capital expenditure guidance, demand for advanced process nodes, the pacing of memory investment, and management’s statements on AI demand over the next three years.

The transmission chain for equipment companies is relatively clear. When wafer fabs increase capital expenditures, equipment companies’ orders and shipments benefit first. If demand remains tight, equipment makers also have opportunities to support gross margins through improvements in product mix and higher capacity utilization. The equipment-related companies mentioned in Citigroup’s report include Applied Materials, Lam Research, Teradyne, and AEIS, but the upside for these individual stocks still needs to track customers’ procurement cadence.

TSMC is the strongest anchor, and the 2027 assumption is clearly above consensus

TSMC remains the most critical driver in this round of the AI capital expenditure cycle. During the April earnings call, TSMC confirmed its 2026 capital expenditure guidance of $52 billion to $56 billion and said spending is trending toward the high end of the range. The market expects the company will most likely keep its 2026 guidance in the upcoming earnings and continue to emphasize demand for advanced process nodes and advanced packaging.

The bigger upside is in the next two years. Citigroup’s model shows TSMC’s capital expenditures at $75 billion in 2027 and $80 billion in 2028, implying year-over-year growth rates of 36% and 7%, respectively. This assumption is above market consensus, with an especially larger gap for 2027.

The core rationale behind this view is that AI/HPC demand continues to drive expansion of advanced process capacity. TSMC captures AI chip demand from Nvidia, AMD, and Broadcom, and also benefits from advanced packaging—such as CoWoS—and migration to higher-end process nodes. As long as AI chip orders remain strong, TSMC has the motivation to continue purchasing more front-end and back-end equipment.

However, higher capital expenditures do not automatically mean the equipment cycle is already locked in. Whether 2027 and 2028 can reach the optimistic model depends on whether AI orders remain durable, whether customers’ in-house chip development pace keeps up, whether advanced packaging bottlenecks ease, and whether equipment delivery cycles can keep pace.

Samsung and Intel bring incremental upside—along with uncertainty

If TSMC provides the base for the equipment cycle, Samsung and Intel determine the upside potential.

At its April call, Samsung said AI demand will drive a substantial year-over-year increase in capital expenditures. Citigroup’s model shows that Samsung Semiconductor capital expenditures will still have high growth rates from 2026 through 2028. There are two tracks here: HBM and high-end DRAM demand driving memory investment, while advanced logic and the foundry business determine whether Samsung can continue to catch up to TSMC in higher-end process nodes.

Samsung’s long-term investment plans also expand the imagination for equipment demand. There are differences in the specific scope across public sources. Samsung’s press releases and media reports cover different ranges, including the group’s total domestic investment, Samsung Electronics’ future business plans, and investment by the semiconductor cluster. A relatively safe way to put it is that Samsung’s Korean investments related to semiconductors over the next decade or more are on the order of 2000 trillion KRW or more. This long-term plan spans multiple years; how much of it can translate into equipment purchases in the near term still depends on site construction specifics, equipment arrival schedules, and the cadence of capacity ramp-ups.

Intel’s situation is more complicated. In its Q1 earnings call, the company adjusted its 2026 capital expenditure outlook from the prior “flat to decline” to “flat,” and said tool- and equipment-related spending is expected to grow by about 25% year over year. In Citigroup’s model, Intel’s capital expenditures for 2027 and 2028 still assume upside, with even greater elasticity in 2028.

Whether Intel can deliver this incremental upside hinges on the foundry business. 18A process validation, 14A customer decisions, and potential collaboration with major customers will all affect the intensity of subsequent investment. If progress from advanced-process customers falls short of expectations, it will be difficult for capital expenditures to be released according to the optimistic scenario. If there is substantive progress in the foundry transition, Intel would become an important incremental source to keep driving global WFE higher.

Micron validates storage demand, but it cannot replace the guidance from the three giants

Capital expenditures at memory manufacturers are also providing validation for the equipment cycle. Micron has raised its FY2026 capital expenditure guidance to approximately $27 billion. The company also said that its capital expenditure level for FY2027 is expected to be about $10 billion higher than the level of FY4Q26. If this quarterly level continues, Micron’s full-year FY2027 capital expenditures could exceed $40 billion.

This indicates that AI servers driving HBM, high-end DRAM, and storage demand is not only a story on the logic-process side. Memory capacity expansion will also increase equipment purchases, especially benefiting deposition, etching, testing, and packaging-related segments. According to reports, Micron’s long-term investment plan in the U.S. has also been raised to more than $250 billion, with the timeframe extending to around 2035.

That said, Micron is more of an indirect corroboration of storage demand and cannot replace guidance from TSMC, Samsung, and Intel. The three companies together account for about 55% of global WFE spending in 2025. What truly determines the height of the equipment cycle is still their statements on capital expenditures for 2026 through 2028 over the next few quarters.

The $250 billion assumption gets stuck after 2027

The biggest disagreement in this earnings-season preview lies in Citigroup’s assumptions for 2027 and 2028, which are clearly more optimistic than the market’s. An upward revision for 2026 is relatively easy to understand because AI demand is already reflected in orders and capacity expansion. After 2027, continuing large increases in capital expenditures require more conditions to be fulfilled at the same time.

AI/HPC demand needs to remain strong, and it cannot just be short-term concentrated purchasing by cloud providers. TSMC’s expansion of advanced process nodes and advanced packaging needs to continue to be supported by ongoing customer orders. Samsung’s large-scale long-term investment plans need to translate into concrete equipment spending, rather than remaining at the level of site construction and long-range planning. Intel’s foundry business also must demonstrate that the 18A and 14A nodes have enough customers and promising prospects for mass production.

Equipment delivery cycles, the macro environment, and fluctuations in the semiconductor cycle will also affect actual spending. Capital expenditure plans can be raised, but they could also be delayed due to changes in customer demand, lower capacity utilization, or financing pressures.

The equipment stocks story is still centered on the three major wafer fabs. If earnings continue to signal strong capital expenditure, the global WFE bull-case scenario will gain more support. If management’s stance on 2027 and 2028 turns cautious, the market’s expectation for $250 billion in equipment spending will need to be discounted. The current controversy is not whether AI capital expenditures exist, but whether this expansion can carry beyond 2026.