The 2026/27 WFE bull case becomes the base case, shifting the industry's focus to end-market compute demand

Overview

The report reassesses the global semiconductor wafer fab equipment cycle and concludes that both demand intensity and the pace of capacity expansion have exceeded prior expectations, turning the 2026/27 bull case into the base case. Growth is driven mainly by advanced logic, DRAM, and advanced packaging, while the research focus is shifting from WFE growth itself to whether compute deployment can support such substantial equipment investment.

Core views

Morgan Stanley raises its 2026/27/28 WFE market forecasts from US$155bn, US$201bn, and US$227bn to US$163bn, US$223bn, and US$254bn, respectively, revising the corresponding year-over-year growth rates from 32%/31%/13% to 39%/37%/14%. Since December 1, 2025, its 2026 and 2027 forecasts have been raised by a cumulative 26% and 54%, respectively, marking the fifth upgrade this year. The report expects the industry's annualized run rate to reach approximately US$200bn in the December quarter of 2026, versus approximately US$125bn in 4Q25, implying an increase of roughly US$75bn over four quarters. The previous view that the WFE cycle would be sustainable was based on the expectation that cleanroom and equipment supply-chain constraints would limit near-term capacity expansion but extend the medium-term opportunity. The report now believes the industry is breaking through both constraints: manufacturers are releasing cleanroom space, accelerating equipment installation, and bringing forward node migrations to meet demand. The increase of more than US$100bn in WFE from 2025 to 2027 broadly matches the approximately 30GW of equipment demand calculated using the report's WFE-per-unit-of-compute methodology; its internet team forecasts industry compute deployment of 29GW and 36GW in 2027/28, respectively. Consequently, the key question is no longer simply how quickly equipment investment can grow, but whether compute demand can absorb the additional capacity. Advanced logic is the first major growth pillar. The report raises its 2026/27 logic WFE forecasts from US$89bn/US$114bn to US$95bn/US$128bn and believes the base of spenders is broadening. The detailed segment forecast projects advanced logic WFE of US$51bn in 2026, up 65%; US$76bn in 2027, up 50%; and US$90bn in 2028, up 17%. Another summary model shows advanced logic growth of 72%/50% in 2026/27. Mature logic is significantly weaker: the detailed section forecasts growth of -2%/+18% in 2026/27, while the summary model shows -3%/+15%, reflecting broadly flat spending from 2023 to 2026 and only a moderate recovery in 2027. The advanced logic/mature logic mix is expected to shift to 63%/37% by 2028, versus 35%/65% in 2024; elsewhere, the report uses 39%/61% in 2025 as the comparison starting point. Process migration is central to advanced logic investment. 3nm wafer capacity is expected to increase from 180kwpm in 2025 to 265kwpm in 2026, with limited additions from 2027 to 2029. Meanwhile, TSMC, Intel, and Samsung are jointly expanding 2nm capacity: the detailed wafer model shows capacity of 180/335kwpm in 2027/28, while the report overview separately shows 335/500kwpm. 1.4nm capacity is expected to reach 60/130kwpm in 2027/28, led by TSMC and Intel. 2nm equipment spending is projected at approximately US$50bn in 2027, with TSMC expected to spend US$43.3bn and Intel US$20.5bn, making Intel the second-largest spender after TSMC. For 2028, the report conservatively assumes slower spending growth at TSMC and Intel, partly offset by a recovery at Samsung, and does not include a potentially significant US$2bn contribution from SpaceX. DRAM is the second major growth pillar and the segment receiving the largest forecast upgrades. The 2026/27 DRAM WFE forecasts are raised from US$49bn/US$62bn to US$53bn/US$70bn, representing growth of 70% and 33%, respectively, and more than US$120bn combined over the two years. Average annual DRAM WFE from 2026 to 2028 is expected to be US$67bn, far above the US$27bn average from 2023 to 2025. Suppliers generally indicate that DRAM will continue to outperform overall WFE in 2027, implying more than US$75bn and approximately 40% growth. Morgan Stanley currently retains its US$70bn forecast but believes risks are skewed to the upside. DRAM expansion is being achieved through both greenfield and brownfield investment. Forecast greenfield capacity additions for 2026/27 are raised from 268/395kwpm to 283/493kwpm. The 2027 addition is nearly twice that of 2026 and approximately three times the 165kwpm added in 2025. Average annual additions from 2026 to 2028 are projected at 427kwpm, above the 170kwpm average from 2023 to 2025. Following Micron's acquisition of PSMC's Tongluo P5 fab, accelerated equipment installation is expected to contribute an incremental US$4bn–US$5bn in 2026. Micron's migration to 1Gamma drives 48% growth in 2026 brownfield WFE. Samsung is allocating more P4 cleanroom space to DRAM, with capacity expected to reach 85kwpm by the end of 2026. Because some node migrations have been brought forward to 2026, the report expects brownfield investment to fall from US$25bn in 2026 to US$15.5bn in 2027, although it acknowledges that it may be underestimating 2027 brownfield spending. This capacity expansion will materially increase DRAM supply. The report raises its 2026/27 DRAM bit supply growth forecasts from 29%/33% to 32%/38%, with the 38% forecast for 2027 significantly above the 24% recorded in 2025. HBM bit supply is expected to grow 55%/57%, while conventional DRAM is expected to grow 30%/36%. Beginning in 2027, incremental non-HBM wafers are expected to exceed incremental HBM wafers. DRAM wafer starts are projected to rise from 2,169kwpm in 3Q26 to more than 3,000kwpm in 2Q28, indicating that equipment investment is translating into substantial supply capacity. Adjustments to NAND are relatively limited. The 2026 NAND WFE forecast is lowered from US$15.6bn to US$13.7bn, while the 2027 forecast remains at US$24bn. The greenfield forecast is broadly unchanged, but 2026 brownfield investment is slightly lower. Accordingly, 2026/27 NAND bit supply growth is revised from 24%/31% to 24%/26%, remaining in the mid-20% range overall. YMTC is expected to account for approximately half of new greenfield capacity in 2026/27, and wafer expansion is expected to become an important source of bit supply growth beginning in 2027. Enterprise SSDs' share of the NAND demand mix is expected to rise from 29% in 2025 to 65% in 2028. The report continues to expect NAND, from a lower base, to be the fastest-growing WFE segment in 2027. Advanced packaging constitutes the third growth pillar. Applied Materials, Lam Research, and KLA have all guided to advanced packaging business growth of more than 70% in 2026. The report believes growth will be driven mainly by CoWoS capacity expansion in 2026 and will shift to HBM and SoIC in 2027. Equipment demand related to CoWoS, HBM, and SoIC is expected to grow 59%/56% in 2026/27, with advanced packaging continuing to outgrow overall WFE. Advanced packaging revenue associated with front-end processes is also continuing to outperform the back end. The report expects advanced packaging revenue at Applied Materials, Lam Research, KLA, and Tokyo Electron to grow by approximately 70% each. Although the report is not bearish on the industry and states that the strength of growth has been a positive surprise, the trajectory does not imply continuous acceleration. The US$254bn WFE forecast for 2028 implies growth slowing from 37% in 2027 to 14%. To sustain the current growth rate, the 2028 market would need to exceed US$300bn. As a result, market discussion will gradually shift from equipment forecast upgrades to whether compute deployment, wafer capacity, and memory bit demand can absorb two consecutive years of large-scale investment.

Analysis framework

The report first updates the overall WFE market size and then breaks down its forecasts into advanced logic, mature logic, DRAM, NAND, and advanced packaging. It subsequently validates each segment through process nodes, monthly wafer capacity, greenfield and brownfield investment, bit supply, and capital expenditure by major manufacturers. Finally, it uses WFE investment per unit of compute to connect equipment investment with the internet team's GW deployment forecasts and cross-checks equipment supplier guidance against its proprietary model to identify upside potential and the expected growth slowdown in 2028.

Methodology notes

• Industry/Sector Analysis FrameworkSupply-demand framework

  Linkage among equipment investment, wafer capacity, and bit supply

  The report derives wafer capacity and DRAM and NAND bit supply growth from WFE investment, then assesses whether the additional supply matches compute and storage demand.

• Industry/Sector Analysis FrameworkUpstream-Midstream-Downstream Value Chain Transmission

  Transmission from compute deployment to semiconductor equipment demand

  The report connects downstream compute GW deployment with upstream WFE investment, using deployment forecasts of 29GW/36GW in 2027/28 to test whether equipment investment corresponding to approximately 30GW is reasonable.

• Industry/Sector Analysis Framework

  WFE per GW mapping

  This method estimates how much wafer fab equipment is required per unit of computing capacity and uses it to compare the more than US$100bn increase in WFE from 2025 to 2027 with the scale of compute infrastructure deployment.

• Industry/Sector Analysis FrameworkVolume-price decomposition

  Decomposition by equipment spending, wafer capacity, and bit output

  The report examines not only WFE spending but also tracks kwpm wafer capacity and bit supply growth separately to identify how equipment expenditure translates into actual output.

• Event-Driven Strategy and Behavioral FinanceExpectation Gap/Expectation Management

  Comparison of supplier guidance with proprietary forecasts

  The report compares equipment suppliers' implied 2027 DRAM WFE of more than US$75bn with its own US$70bn forecast to identify upside forecast risk.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• TSMC

  The report expects it to be the largest contributor to advanced logic WFE growth in 2027, primarily driven by 2nm capacity expansion.

  Strengths

  Expected spending of US$43.3bn in 2027, leading the expansion of 2nm and 1.4nm capacity.

  Weaknesses

  The report conservatively expects its spending growth to slow in 2028.

  Comparison

  Its projected 2027 spending is higher than Intel's expected US$20.5bn.

• Intel

  The report expects it to become the second-largest advanced logic WFE spender in 2027, representing one of the important conditions for the bull case becoming the base case.

  Strengths

  Expected to spend US$20.5bn in 2027 and participate in 2nm and 1.4nm capacity expansion.

  Weaknesses

  The report expects its spending growth to slow in 2028.

  Comparison

  Expected to rank second only to TSMC and significantly broaden the base of advanced logic spenders.

• Samsung Electronics

  Participating in both 2nm advanced logic expansion and greenfield DRAM expansion, with more P4 cleanroom space allocated to DRAM.

  Strengths

  The P4 fab is expected to reach DRAM wafer-start capacity of 85kwpm by the end of 2026, while logic spending could recover in 2028.

  Weaknesses

  The report expects its contribution to 2027 growth to continue to lag TSMC and Intel.

  Comparison

  Samsung and SK hynix account for the largest upgrades to the 2027 DRAM greenfield forecast.

• Micron

  Its acquisition of PSMC's Tongluo P5 fab, accelerated equipment installation, and migration to the 1Gamma node are driving stronger-than-expected DRAM WFE in 2026.

  Strengths

  The P5 fab is expected to contribute an incremental US$4bn–US$5bn in 2026, while node migration supports brownfield spending.

  Weaknesses

  Following the pull-forward of node migration, the report expects brownfield investment to slow materially in 2027.

  Comparison

  Together with Samsung and SK hynix, it is a major source of DRAM capacity expansion.

• SK hynix

  It is one of the main participants behind the modest revision to the 2026 DRAM forecast and the substantial upgrade to the 2027 greenfield forecast.

  Strengths

  Benefits from strong growth in DRAM and HBM equipment demand.

  Comparison

  Alongside Samsung, it receives one of the largest upgrades to the 2027 DRAM greenfield forecast.

• YMTC

  The report expects it to account for approximately half of the increase in NAND greenfield capacity in 2026/27.

  Strengths

  It is one of the largest individual sources of new NAND wafer capacity.

  Weaknesses

  The 2026 NAND WFE forecast has been lowered, while industry bit supply growth falls to the mid-20% range.

  Comparison

  Accounts for approximately 50% of the NAND greenfield additions forecast in the report.

• Applied Materials, Lam Research, KLA, Tokyo Electron

  The report expects advanced packaging revenue at this group of equipment manufacturers to grow by approximately 70% each.

  Strengths

  They benefit from CoWoS capacity expansion in 2026 and HBM and SoIC investment in 2027, while growth in front-end-related advanced packaging revenue outpaces the back end.

  Comparison

  All four companies are expected to deliver broadly comparable high growth in advanced packaging.

Key data

• 2026 WFE ForecastUS$163bn, +39% YoYPreviously US$155bn, +32%
• 2027 WFE ForecastUS$223bn, +37% YoYPreviously US$201bn, +31%
• 2028 WFE ForecastUS$254bn, +14% YoYPreviously US$227bn, +13%; growth is materially lower than in 2027
• Cumulative Forecast Upgrade2026 +26%, 2027 +54%Relative to the forecasts as of December 1, 2025
• Industry Annualized Run RateApproximately US$200bnExpected to be reached in the December quarter of 2026; approximately US$125bn in 4Q25
• 2026/27 Logic WFE ForecastUS$95bn/US$128bnPreviously US$89bn/US$114bn
• 2027 Advanced Logic WFEUS$76bn, +50%TSMC and Intel are the main sources of growth
• 2027 2nm Equipment SpendingApproximately US$50bnDriven by capacity expansion at TSMC, Intel, Samsung, and other manufacturers
• 2026/27 DRAM WFEUS$53bn/US$70bnUp 70%/33%, respectively, and more than US$120bn combined over the two years
• 2026/27 DRAM Bit Supply Growth32%/38%Previously forecast at 29%/33%
• 2026/27 NAND WFEUS$13.7bn/US$24bnPreviously US$15.6bn/US$24bn
• 2026/27 Advanced Packaging Equipment Demand Growth59%/56%Including CoWoS, HBM, and SoIC
• Enterprise SSD ShareRising from 29% in 2025 to 65% in 2028The report's projected change in the NAND demand mix

Impact & implications

The report concludes that both the scale and duration of the semiconductor equipment cycle are stronger than previously expected, with advanced logic, DRAM, and advanced packaging equipment suppliers benefiting from a broader range of capital expenditure sources. However, large-scale capacity expansion will shift the discussion toward whether end-market compute and storage demand can absorb the additional supply; following the high base in 2027, growth is expected to slow materially in 2028.

Risks

• The 2027 DRAM WFE forecast carries upside risk: equipment supplier signals point to more than US$75bn, versus US$70bn in the report's model.
• The report acknowledges that it may be underestimating 2027 DRAM brownfield investment because its model assumes that spending declines from US$25bn in 2026 to US$15.5bn.

What to watch

• Track whether the WFE annualized run rate can reach approximately US$200bn in the December quarter of 2026.
• Monitor whether compute deployment of 29GW/36GW in 2027/28 can support rapidly growing equipment investment.
• Watch whether 2027 DRAM WFE is closer to the report's US$70bn forecast or reaches the more than US$75bn implied by suppliers.
• Track the realization of TSMC's US$43.3bn, Intel's US$20.5bn, and approximately US$50bn of 2nm equipment spending.
• Monitor whether advanced packaging growth shifts from CoWoS to HBM and SoIC in 2027 as expected.
• Watch whether the 2028 WFE market can exceed US$300bn; a level below this would make it difficult to sustain the 2027 growth rate.