J.P. Morgan raises global WFE forecasts to a 28% CAGR for 2025–28 as AI investment broadens semiconductor-equipment demand.

Overview

J.P. Morgan’s global semiconductor-equipment update raises its WFE outlook on stronger AI-driven cloud, memory, foundry and advanced-packaging investment. The institution expects the spending cycle to remain unusually strong through 2028, while emphasizing customer mix, production capacity and pricing power as the principal stock-selection factors.

Core views

J.P. Morgan raises its bottom-up global wafer fab equipment forecast to 31% year-on-year growth in 2026, equivalent to $163 billion, from 28% previously; to 38% in 2027, or $225 billion, from 29%; and to 17% in 2028, or $263 billion, from 16%. This implies a 28% CAGR for 2025–28. The revisions are led by DRAM and TSMC, but the institution also lifts assumptions for NAND and logic. The central mechanism is accelerating cloud-service-provider investment in response to AI demand: the top four US CSPs are forecast to lift investment at a 58% CAGR in 2025–28, prompting chipmakers to add leading-edge capacity and purchase more technically complex equipment. The report argues that equipment demand should be reinforced by tight supply conditions rather than merely a broad cyclical recovery. DRAM supply is expected to lag demand through at least 2028 despite capacity growth, as HBM consumes three to four times as much die capacity and becomes a larger share of output. The institution estimates 2027 supply at only 70–80% of demand and says an additional 300,000 WSPM would be needed to balance the market in 2028, which it doubts is achievable. It raises its 2026–28 memory TAM by 4–8% and its 2027 HBM price forecast to 42% year-on-year. Long-term agreements, with reported 20–25% prepayments and 50–70% volume coverage, are cited as evidence of memory makers’ bargaining power. TSMC is the other major demand engine. J.P. Morgan expects N2, N3 and N5 utilization to remain above 100% through 2028 as AI accelerators and agentic-AI CPUs drive demand. It raises TSMC capex forecasts to $62 billion, $81 billion and $90 billion for 2026, 2027 and 2028, respectively, from $56 billion, $65 billion and $72 billion. Leading-edge capacity at N5 and below is forecast to grow at a 24% CAGR in 2026–28, while TSMC’s CoWoS capacity forecast is raised to 115,000, 190,000 and 225,000 wafers per month at end-2026, 2027 and 2028. The report also expects OSAT CoWoS-like capacity to reach 15,000, 50,000 and 85,000 wafers per month, linking advanced packaging growth to future NVIDIA, AMD, Google, AWS and other AI-product ramps. For stock selection, the institution favors suppliers with substantial DRAM and TSMC exposure, enough component and manufacturing capacity to serve demand, and the ability to improve margins through price revisions. It sees equipment pricing entering a new phase: value-based pricing remains intact, while suppliers such as Tokyo Electron are beginning to pass through cost increases. From 2027, J.P. Morgan expects cleanroom start-ups and ramps, together with price revisions, to become increasingly important growth drivers. Among US semiconductor-capital-equipment companies, the report identifies KLA as its top pick because it has lagged Lam Research and Applied Materials year to date despite an expected rotation of WFE growth toward foundry and logic in 2027. KLA’s process-control tools are seen as particularly well placed for new-node tape-outs, and higher Intel capex focused on yield and cost could support upside to consensus 2027 revenue estimates. In Europe, ASML is the top pick: its longer lead times shift more of its growth into 2027–28, and possible Samsung High-NA EUV production adoption in 2028 adds a driver beyond the DRAM cycle. Its 100% EUV market share is expected to make it a key beneficiary. For Japan, J.P. Morgan continues to favor Tokyo Electron in front-end equipment and Advantest in back-end test. Tokyo Electron is preferred for its high TSMC and DRAM exposure, planned capacity expansion at the Miyagi Production Innovation Center by summer 2027, and expected price-revision benefits from the second half of fiscal 2026, primarily the fourth quarter. Advantest is expected to deliver a 47% operating-profit CAGR in fiscal 2025–28, supported by its exposure to GPUs, ASICs and CPUs and the rising importance of testing as chips become more complex. The report also identifies widening AI effects across China equipment, foundry, packaging and servers. It raises China fab capex forecasts by 10% for 2026 and 9% for 2027 and newly estimates $54.7 billion for 2028, implying a 7% CAGR in 2025–28. Domestic advanced logic and memory investment is forecast to grow at a 12% CAGR, equipment localization to rise from about 25% in 2025 to 40%, and domestic-vendor TAM to grow at a 25% CAGR. Meanwhile, AI accelerator demand is forecast to grow at a 50% CAGR in 2025–28, supporting server demand even as higher memory costs leave PC demand fragile. The report notes that lower HBM or SOCAMM specifications are intended to manage supply constraints rather than signal weakening AI demand, although memory demand-destruction concerns could limit further valuation gains.

Analysis framework

J.P. Morgan builds a bottom-up WFE forecast from semiconductor makers’ capex plans and estimated equipment intensity. It connects AI-related CSP capex to chip demand, then evaluates memory and foundry capacity, utilization, pricing, packaging and end-market shipment assumptions to identify the equipment suppliers best positioned for the resulting spending mix.

Methodology notes

• Industry AnalysisSupply-demand framework

  Bottom-up semiconductor equipment, memory and capacity supply-demand analysis

  The report compares forecast demand with wafer capacity, utilization and equipment spending to explain why DRAM, foundry and WFE demand may remain strong.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  AI CSP capex-to-chipmaker-to-equipment-supplier transmission

  It traces how higher cloud investment increases demand for AI chips, leading-edge foundry, memory, advanced packaging and ultimately semiconductor equipment.

• Industry AnalysisVolume-price decomposition

  Equipment demand and supplier margin effects from volume, mix and price revisions

  The report distinguishes rising equipment volumes and changes in customer mix from potential pricing actions that can improve supplier margins.

Asset mapping & comparison

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

• KLA Corporation (KLAC)

  Top US semiconductor-capital-equipment pick; expected beneficiary of a 2027 shift toward foundry and logic WFE spending.

  Strengths

  Underappreciated foundry/logic exposure and potential upside to 2027 revenue consensus as Intel capex rises.

  Weaknesses

  Its tools benefited less than peers from the recent memory-led WFE pickup.

  Comparison

  Shares rose about 50% year to date, versus about 85% for Lam Research, 82% for Applied Materials and 68% for the SOX.

  Risks

  The expected rotation toward foundry and logic spending may not materialize as forecast.

• ASML (ASML)

  Top European SPE pick and expected key beneficiary of WFE growth in 2027–28.

  Strengths

  100% EUV market share and a potential High-NA EUV production driver from Samsung in 2028.

  Weaknesses

  Longer lead times make shipment timing more weighted to fiscal 2027 and 2028.

  Comparison

  Up 62.5% in FY2026 versus the AEX up 16%, but underperforming US peers.

  Risks

  The anticipated timing of shipments and High-NA EUV adoption could change.

• Tokyo Electron (8035)

  Preferred Japanese front-end equipment supplier.

  Strengths

  High exposure to TSMC and DRAM, planned production-capacity expansion and expected price-revision benefits.

  Risks

  Benefits depend on sustained TSMC/DRAM demand, supply execution and price-revision realization.

• Advantest (6857)

  Preferred Japanese back-end semiconductor test supplier.

  Strengths

  Strong GPU, ASIC and CPU exposure; J.P. Morgan forecasts a 47% FY2025–28 operating-profit CAGR.

  Risks

  Growth depends on continued chip-complexity and AI test-demand expansion.

Key data

• Global WFE market growth, 202631% YoY; $163 billionRaised from 28% YoY.
• Global WFE market growth, 202738% YoY; $225 billionRaised from 29% YoY.
• Global WFE market growth, 202817% YoY; $263 billionRaised from 16% YoY.
• Global WFE CAGR, 2025–2828%J.P. Morgan’s revised forecast.
• Top four US CSP investment CAGR, 2025–2858%Forecast based on aggregate J.P. Morgan analyst estimates.
• TSMC capex forecast, 2026/2027/2028$62 billion / $81 billion / $90 billionRaised from $56 billion / $65 billion / $72 billion.
• 2027 DRAM supply relative to demand70–80%The institution expects the shortage to worsen before easing somewhat in 2028.
• Advantest operating-profit CAGR, FY2025–2847%Supported by GPU, ASIC and CPU test demand.

Impact & implications

The report sees a multi-year equipment upcycle led by AI-related memory, leading-edge foundry and advanced-packaging investment. It argues that the strongest beneficiaries should be suppliers with DRAM and TSMC exposure, operational capacity to fulfill demand and credible pricing or margin expansion, while supplier and end-market differences should drive relative performance.

Risks

• Higher memory prices and rising component costs could weaken PC demand through price elasticity.
• Concerns about demand destruction from lower HBM and SOCAMM specifications could cap additional memory valuation gains.
• The report notes that CSP AI-investment sustainability and funding remain market concerns.
• Equipment suppliers may be constrained by component procurement and production capacity despite strong demand.

What to watch

• CSP capex trajectories through 2027 and whether funding continues through bond and equity markets.
• DRAM supply-demand balance, HBM pricing and the pace of capacity additions through 2028.
• TSMC utilization at N2, N3 and N5, capex execution and potential 2027 price increases.
• CoWoS and OSAT advanced-packaging capacity ramps and new AI-product launches in 2027–28.
• Equipment suppliers’ customer mix, component availability and realization of price revisions.