UBS raises SMTC target price to $165 and remains positive on the expansion of AI data center connectivity demand
AI summary card
UBS raises SMTC target price to $165 and remains positive on the expansion of AI data center connectivity demand
The report argues that SMTC will benefit from ACC customer expansion, TPU-related shipments, 1.6T/LPO optical-module upgrades, and IoT Systems optimization, and reiterates Buy on the basis of stronger EPS expectations.
- SMTC's target price is raised from $105 to $165, with the valuation base rolling from CY27E EPS of $3.55 to the CY27/CY28E average EPS of $5.13, and the target multiple lifted from 30x to 32x.
- UBS expects SMTC FQ1:27 revenue/EPS of $287MM/$0.46 and Jul Q revenue/EPS of $312MM/$0.52, both slightly above consensus.
- The enterprise AI survey shows the share of production deployments rose to about 19%, NVDA's mention rate for inference chips is about 75%, and the share of CPU infrastructure used for AI inference rose to 30%.
- The quarterly WFE framework shows C1Q26 at about $31.5B may be the low point for the year, UBS keeps 2026 WFE at about $142B, and sees potential for 2027 to advance toward about $200B.
Report interpretation
Overview
This report is UBS's SemiBytes update on the U.S. semiconductor and semiconductor equipment sector, covering NVDA H200 shipment approvals to China, the enterprise AI survey, the quarterly WFE framework, and SMTC earnings preview. At the single-stock level, the focus is SMTC. UBS believes its revenue opportunities in AI data center connectivity, ACC, 1.6T, and optical-module upgrades are clearer, so it raises earnings estimates and the target price.
Core views
The key views are: first, U.S. government approval for NVDA H200 shipments to about 10 Chinese customers is a positive threshold event, but actual delivery still depends on approvals and execution progress on the China side; second, enterprise AI deployment continues to expand, NVDA remains dominant at the inference layer, and rising CPU inference usage reinforces the view that AI infrastructure spending will continue; third, C1Q26 WFE at about $31.5B may be the annual low, with memory demand and cleanroom capacity constraints supporting later equipment spending; fourth, SMTC benefits from TPU-related ACC demand, potential new customers, and LPO/NPO upgrades, improving earnings leverage.
Analysis framework
The report analyzes company earnings previews, peer and supply-chain commentary, UBS's enterprise AI survey, reported results from semiconductor equipment companies, the quarterly WFE framework, earnings estimate revisions, and relative valuation multiples. SMTC's target price is derived by multiplying the CY27/CY28E average non-GAAP EPS by the target P/E multiple, with upside, base, and downside cases provided.
Methodology notes
CY27/CY28E average EPS multiplied by the target multiple
SMTC's base target price is based on the CY27/CY28E average EPS of $5.13 and a 32x multiple, with the multiple benchmarked against the median levels of traditional networking/high-performance analog peers and AI-exposed networking peers.
Target price range under different assumptions for revenue growth, gross margin, and valuation multiples
The upside case is $260, assuming F28/F29E revenue growth of 42%/40%, F29E gross margin rising to 63%, EPS of $7.35, and a 35x multiple; the base case is $165; the downside case is $90, assuming slower growth, limited margin improvement, EPS of $2.90, and a 30x multiple.
Quarterly wafer fab equipment spending run rate
UBS updates the WFE framework using revenue structures from the top five SPE companies such as ASML, AMAT, LRCX, KLAC, and TEL, estimates C1Q26 WFE at about $31.5B, and judges that quarter may be the annual low point.
AI deployment stage, chip selection, and CPU inference usage
UBS's semiannual enterprise AI survey covers 139 respondents, skewed toward senior IT decision-makers, and is used to track AI production deployment, NVDA inference chip share, and changes in the use of CPU infrastructure for AI inference.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- SEMTECH CORP / US.SMTCPrimary covered stock
- Strengths
- Benefits from ACC, TPU-related demand, 1.6T/LPO upgrades, higher optical-module content, and IoT Systems portfolio optimization; UBS raises longer-term EPS and increases the target price.
- Weaknesses
- Near-term results still need to confirm customer expansion, design wins, and revenue conversion speed, and part of the growth depends on continued AI data center capex.
- Comparison
- The 32x target multiple references the median multiple of traditional networking/high-performance analog peers and AI-exposed networking peers.
- Risks
- Pricing pressure from competition, design wins below expectations, gross margin improvement weaker than expected, and a slowdown in AI data center demand.
- NVIDIA / NVDAKey readout for industry demand and AI infrastructure
- Strengths
- The enterprise AI survey shows about a 75% mention rate for NVDA at the inference-chip layer, and U.S. approval for H200 shipments to Chinese customers creates potential revenue upside.
- Weaknesses
- Actual H200 shipments have not yet been executed and still depend on approvals and the delivery timeline on the China side.
- Comparison
- NVDA is clearly ahead at the enterprise inference-chip layer, and older-generation GPUs still have a long useful life.
- Risks
- Cross-border regulation, export-license execution, delivery delays, and limited visibility into China demand.
- WFE / Semiconductor equipment chainIndustry cycle backdrop
- Strengths
- C1Q26 at about $31.5B may mark the annual low, with rising memory demand and cleanroom capacity constraints supporting subsequent quarterly run rates.
- Weaknesses
- Quarterly equipment spending may be constrained by cleanroom availability, supply-chain timing, and customer capex cycles.
- Comparison
- Among the top five SPE companies, memory-related revenue increased by about $2.1B qoq, while foundry/logic was about $13.4B, down $1.2B qoq.
- Risks
- If full-year WFE does not reach the low-to-mid $140B range, or if system shipment growth falls below supply-chain checks, the path toward $200B may be delayed.
Key data
- Report date2026-05-18UBS Global Research publication date.
- SMTC target price$165Raised from $105 to $165, and Buy reiterated.
- SMTC FQ1:27 revenue/EPS forecast$287MM / $0.46Slightly above the guidance midpoint and consensus of $284MM / $0.45.
- SMTC Jul Q revenue/EPS forecast$312MM / $0.52Above the consensus estimate of $300MM / $0.51.
- CY27E revenue/EPS forecast$1,303MM / $2.27Raised from $1,292MM / $2.20.
- CY28E revenue/EPS forecast$1,747MM / $3.92Raised from $1,651MM / $3.55.
- CY29E revenue/EPS forecast$2,270MM / $6.33Raised from $1,932MM / $5.14.
- Enterprise AI production deployment rateabout 19%About 17% in the last survey, around 6% at the end of 2023.
- NVDA inference chip mention rateabout 75%Broadly unchanged from the October 2025 survey.
- CPU share used for AI inference30%Above the 23% in the October 2025 survey.
- C1Q26 WFE estimateabout $31.5BUBS believes this may be the annual quarterly low.
- 2026 WFE estimateabout $142BThe report says that if the full-year figure lands in the low-to-mid $140B range, the view that C1Q26 was the low point becomes more credible.
Impact & implications
For SMTC, the report extends the investment case from a single customer or short-term earnings beat to a broader AI data center connectivity upgrade cycle: TPU/ACC, 1.6T, LPO/NPO, and optical-module content gains may simultaneously drive revenue growth and gross margin improvement. For the semiconductor equipment chain, the view that WFE bottoms in the quarter and continues to rise in 2027 supports medium-term demand resilience for equipment companies. For NVDA, H200 approval in China creates potential upside, but delivery timing still faces regulatory and cross-border execution uncertainty.
Risks
- Technology industry investment risks are high, including low sales visibility, rapid innovation and technological change, intense competition, frequent M&A, and low barriers to entry in some markets.
- SMTC's upside thesis depends on growth in ACC, CopperEdge, 1.6T/LPO, and optical-module content; if design wins or customer expansion fall short, revenue and EPS upgrades may not materialize.
- Gross margin improvement depends on a richer data center product mix and efficiency gains; if competition weakens pricing power, the downside-case target price may be only $90.
- Although NVDA's H200 has made progress on the U.S. approval side, China-side regulation and actual delivery remain uncertain.
- The WFE uptrend view depends on memory demand, cleanroom availability, and system shipment growth; if capex or supply chains weaken, industry run rates could come in below expectations.
What to watch
- SMTC's after-hours earnings release on May 26 and management's commentary on Jul Q guidance.
- CopperEdge socket progress beyond the first hyperscaler customer, including additional hyperscaler ASICs, AMD, or non-AI data center use cases.
- The adoption pace of 1.6T full LPO relative to LRO/FRO, and the future NPO upgrade path.
- The timing of potential portfolio optimization or IoT Systems rationalization and its impact on gross margin and revenue mix.
- SMTC's R&D investment direction in optical modules and other signal-integrity content.
- The actual shipment timing of NVDA H200 to Chinese customers, remaining approval steps, and delivery visibility.
- Whether subsequent quarterly WFE run rates rebound from the C1Q26 low, and whether 2027 system shipments approach the high growth implied by supply-chain checks.