Long-term positive, near-term cautious

研报英文原文证据摘录

Long-term positive, near-term cautious

oted, all metrics are based on Morgan Stanley ModelWare

trading at 36x our CY27 estimates, and we think our optimism around Intel capex is framework

** = Based on consensus methodology

somewhat understood by the market. Our understanding is that the stock is trading § = Consensus data is provided by Refinitiv Estimates

e = Morgan Stanley Research estimates

on 2028 forecasts, but KLA is unlikely to comment on conditions more than 24

Quarterly EPS ($)

months out. The company has talked toward 2027 y/y growth being higher than 2026e 2026e 2027e 2027e

Quarter 2025 Prior Current Prior Current

2026, but acceleration should be the minimum expectation given the undergrowth

Q1 0.73 - 0.88a - 1.19

vs WFE in 2026, with MSe WFE at +32% vs KLA at +23%. Once again, we are Q2 0.82 - 0.88a - 1.30

positive on KLA’s long-term story, but we do not think earnings will provide reasons Q3 0.84 - 0.94a - 1.40

Q4 0.94 - 0.99 - 1.52

to become materially more positive. As a result, we are more cautious than positive e = Morgan Stanley Research estimates, a = Actual Company reported data

into the print.

1D and 1.4nm will be tailwinds in 2027. DRAM and logic mix has been a significant

headwind for process control intensity in 2026. DRAM WFE in 2026 has been

focused on greenfield and brownfield capacity for 1C/Gamma, which was most

process-control intensive in 2025. Similarly, TSMC’s WFE has been focused on 2nm

and 3nm, with 2nm process control intensity peaking in 2024/2025 and 3nm before

that. Our analysis of process control intensity at TSMC by node, as well as industry

DRAM wafers, suggests that KLA’s leading-edge logic revenue at TSMC should

accelerate in 2027, while DRAM growth rates should sustain despite a deceleration