UBS Preview of Earnings for Four U.S. Chipmakers, Comprehensive Target Price Raising

Overview

This UBS report provides an earnings preview for the U.S. semiconductor sector, covering CEVA, Ambiq Micro (AMBQ), Applied Materials (AMAT), and Marvell Technology (MRVL). The report focuses on analyzing each company’s upcoming quarterly earnings forecasts, core business drivers, and valuation changes. Overall, UBS holds an optimistic view on these four companies, with its core logic centered around the surge in AI computing power demand, the semiconductor equipment capital expenditure (WFE) supercycle, and the penetration of Apple’s self-developed chips. As a result, UBS has comprehensively raised the target prices for these companies.

Core views

CEVA: Driven by Apple’s self-developed modem and automotive AI dual engines. UBS maintains CEVA’s earnings forecast unchanged but adopts a more positive outlook. Although UBS lowered its global smartphone shipment growth forecast for 2026 to -7.5%, surveys indicate that Apple’s self-developed modem (C1x) penetration may be more aggressive than market expectations (retaining approximately 20% of Qualcomm’s share). C1x has already expanded to the new iPad Air and entry-level iPhone 17e, which is expected to boost performance in the second half of 2026. Additionally, Renesas’ mass production of CEVA-based AI DSPs on Toyota’s RAV4 platform (annual sales of about 1.1–1.2 million units) validates CEVA’s traction in the automotive MCU space and is expected to contribute to performance in the second half of 2026 and 2027. Based on a 49x P/E multiple for 2027 (previously 32x), the target price was raised from $27 to $42, maintaining ‘Buy’.

Analysis framework

UBS’s analysis follows a methodology combining bottom-up and cross-validation across the industry chain. Supply Chain Surveys and Terminal Cross-Validation: The report repeatedly references ‘checks’ (supply chain surveys) to verify or refute market baseline assumptions. For example, surveys found that Apple’s self-developed modem penetration could be more aggressive than the baseline assumption of retaining 20% of Qualcomm’s share; through AMAT’s supply chain survey, negative forecasts from a certain PVD valve supplier were ruled out as a drag on AMAT. By tracking terminal wearable device shipments such as Garmin’s, AMBQ’s chip demand is retroactively inferred. Capacity and Capital Expenditure Cycle Transmission: For AMAT, UBS deeply links the company’s performance with the global semiconductor equipment (WFE) capital expenditure cycle. By forecasting year-on-year growth rates for DRAM WFE and the industry’s annualized run rate (exceeding $180 billion), it derives AMAT’s system shipment growth (up 65% year-on-year) and upside potential for revenue from a top-down perspective. Large Customer Project Breakdown and Volume-Price Estimation: For MRVL, UBS breaks down the custom ASIC business by major customers (Amazon, Microsoft, Google), assessing unit shipment assumptions for each project (such as Microsoft’s Maia 3 potentially reaching nearly 500,000 units) and translating them into revenue elasticity. Meanwhile, by identifying structural drivers such as NVDA-MRVL collaboration and AWS adoption of NVLink, MRVL’s addressable market expansion in high-speed switches is derived.

Methodology notes

• Valuation MethodPE/PEG valuation

  P/E (Price-to-Earnings) Valuation

  Multiplying projected earnings per share (EPS) by a reasonable P/E multiple to arrive at the target price. In this report, UBS raised CEVA’s 2027 P/E multiple from 32x to 49x based on its increasing strategic importance in mobile and automotive terminals; for AMAT, a 22x NTM P/E (a premium of one round over its three-year average) was adopted, reflecting the favorable impact of the DRAM WFE supercycle.

• Valuation MethodPS valuation

  EV/Sales (Enterprise Value/Sales) Valuation

  For growth companies that have yet to achieve stable profitability, valuations often anchor on revenue rather than profit. For AMBQ, given its difficulty in achieving profitability before 2028, UBS adopted a 6.0x EV/Sales multiple (up from the previous 4.2x) and referenced the median of AI semiconductor peers for pricing.

• Industry/Industrial Analysis FrameworkUpstream-Downstream Industry Chain Transmission

  Terminal Demand Transmission to Upstream Industry Chain

  When analyzing the prospects of chip companies, starting from the prosperity of downstream terminal products (such as Toyota RAV4 annual sales, wearable device shipments, cloud vendor capital expenditures), we can infer the order and revenue elasticity of upstream chip design or equipment companies.

• Competition and Strategic FrameworkMoat / competitive advantage

  Supply Chain Diversification and Second Supplier Opportunities

  In the fields of custom ASICs and network switches, cloud giants actively introduce second suppliers to avoid over-reliance on single vendors (such as AVGO/CSCO). This diversification demand forms the core growth logic for second suppliers like MRVL.

• Cycle and Prosperity FrameworkCapacity/Equipment Cycle (Juglar)

  WFE (Wafer Fab Equipment) Capital Expenditure Supercycle

  Capital expenditures in semiconductor manufacturing exhibit clear cyclical patterns. The report points out that we are currently in a WFE supercycle driven by DRAM demand, with the industry’s run rate expected to reach $200 billion by 2027. This prosperity turning point is the core basis for being bullish on AMAT.

• Industry/Industrial Analysis FrameworkSupply-demand framework

  Capacity Expansion Under Supply-Demand Framework

  By forecasting equipment spending growth rates in segments such as DRAM and the industry’s annualized run rate, we can judge the upside potential for equipment manufacturers’ future orders and visibility of profitability.

Asset mapping & comparison

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

• CEVA Inc (CEVA.US)

  Benefiting from deepening Apple’s self-developed modem penetration and ramp-up of automotive AI DSPs (NPU)

  Strengths

  Strategically relevant in mobile connectivity IP and edge AI sectors, with continued validation from Apple and automotive clients

  Comparison

  Valuation multiple raised from 32x to 49x, reflecting stronger confidence in revenue sustainability compared to peers in IP and edge computing

  Risks

  Slower-than-expected adoption of AI DSP/NPU, lower-than-expected proportion of Apple’s self-developed modems in iPhones

• Ambiq Micro Inc (AMBQ.US)

  Benefiting from improved wearable device demand, but with increased short-term investment and Atomiq platform still far from mass production

  Strengths

  Apollo series products show strong demand in the wearable space; January equity offering improves funding path toward 2028 profitability

  Weaknesses

  Not yet profitable; Atomiq platform is more of a long-term roadmap than a near-term revenue driver

  Comparison

  Valuation multiple (EV/Sales 6.0x) already reflects some long-term opportunities; compared to peers, short-term validation of its diversification and product mix is still early

  Risks

  Delays in edge AI and new customer mass production schedules, persistently high operating expenses

• Applied Materials Inc (AMAT.US)

  Core beneficiary of the DRAM WFE capital expenditure supercycle

  Strengths

  Supply chain surveys show system shipments growing strongly year-on-year, holding a high share in DRAM equipment spending, and expected to regain WFE market share

  Risks

  Storage industry cuts capital expenditures, global economic downturn weakens terminal demand, showing business capital expenditures below expectations

• Marvell Technology Inc (MRVL.US)

  Benefiting from higher-than-expected demand for custom ASICs from cloud customers such as Amazon and Microsoft, and expansion of the high-speed switch market

  Strengths

  Microsoft’s XPU projects (Maia 3/4) offer huge revenue elasticity; collaboration with NVDA and AWS adoption of NVLink expand the addressable market for switches

  Comparison

  Valuation uses weighted average P/E of peers instead of median, better reflecting the weight of large-cap peers like NVDA/AVGO

  Risks

  Cloud customers seeking alternative suppliers leads to market share loss; SSD suppliers’ internal controller R&D capabilities underestimated

Key data

• CEVA Target Price$42Raised from $27, based on a 49x 2027 P/E multiple (previously 32x)
• CEVA FQ1:26 (March) Revenue Forecast$26.2MMGross margin forecast at 87.0%, roughly in line with Street consensus
• AMBQ Target Price$43Raised from $32, based on a 6.0x EV/Sales multiple (previously 4.2x)
• AMBQ FQ1:26 (March) Revenue Forecast$21.5MMASP $2.94 (+4% Q/Q), non-GAAP gross margin at 44.5%
• AMAT Target Price$480Raised from $430, based on a 22x C2027E Non-GAAP EPS of $21.28
• AMAT FQ2:26 (April) Revenue Forecast~$7.71BAbove the midpoint of guidance; SSG $5.83B, AGS $1.63B
• AMAT FQ3:26 (July) Revenue Forecast~$8.52BOver $400MM above Street consensus, with system shipments expected to grow by about 65% year-on-year
• MRVL Target Price$195Significantly raised from $120, based on a 23x C2028E EPS of $8.35
• MRVL FQ1:27 (April) Revenue Forecast$2.44BSlightly above Street’s $2.40B; data center revenue expected to rise by 12% Q/Q to $1,857MM
• MRVL FQ2:27 (July) Revenue Forecast$2.66BAbove Street’s $2.59B; data center revenue expected to rise by 11% Q/Q to $2,061MM

Impact & implications

UBS’s view suggests that the AI and semiconductor cycle is spreading from mere computing power demand to broader customized chips and equipment capital expenditures. For CEVA, deepening Apple’s self-developed modem penetration and the implementation of automotive AI mean that the IP licensing model’s long-term growth potential has been underestimated by the market. For AMAT, the DRAM WFE supercycle signals that equipment manufacturers’ profits will continue to exceed expectations, and the industry’s annualized equipment spending approaching $200 billion reflects the rigidity of demand. For MRVL, the timing of cloud giants’ self-developed chip ramp-ups and new product cycles represent significant valuation reshaping opportunities, especially the Microsoft Maia project and the option value of its high-speed switch business.

Risks

• CEVA: Lower-than-expected growth in AI DSP and NPU revenues, slower-than-expected adoption or proportion of Apple’s self-developed modems in iPhones
• AMBQ: Delays in edge AI and new customer mass production schedules, persistently higher-than-expected operating expenses
• AMAT: Deep economic recession weakens terminal demand, storage industry cuts capital expenditures, showing China/LCD business growth below expectations or market share loss
• MRVL: Cloud customers introducing other suppliers leads to market share loss; SSD suppliers’ internal controller R&D capabilities underestimated, resulting in lower-than-expected revenue
• Industry-wide: The tech industry features low sales visibility, rapid innovation, intense competition, and frequent M&A activities, making overall risks above average

What to watch

• CEVA: Connectivity licensing trends (shifting toward Wi-Fi 6/7 and multi-protocol designs), edge AI demand and NPU licensing progress, content and timing of Apple’s self-developed modem IP
• AMBQ: Mass production progress of new wearable projects, Atomiq/Apollo 6 tape-out and testing progress, expansion into medical/industrial markets
• AMAT: 2026 WFE spending outlook (especially advanced logic and DRAM), impact of export control rules (BIS Affiliates Rule), domestic WFE demand in China
• MRVL: Data center/cloud demand and capital expenditure guidance, core enterprise networking demand, long-term gross margin outlook from custom ASIC mass production, AI-related revenue expectations