Analog Chip Performance Exceeds Expectations, Ultra-Large-Scale AI Capital Expenditures Continue to Rise

Overview

In this report, UBS comprehensively assesses the latest performance and market dynamics of the U.S. semiconductor industry. The report notes that the manufacturing PMI remains in a slight expansion range, while analog chip companies—including TXN and STM—reported first-quarter results that significantly exceeded expectations and raised guidance, reflecting broad recovery in downstream demand from industries, automobiles (especially ADAS), and data centers. Meanwhile, earnings reports from the four ultra-large-scale cloud providers further reinforced the ongoing growth trend in AI infrastructure spending, leading to a significant upward revision of industry capital expenditure forecasts for 2027. Additionally, the report quantitatively analyzes changes in semiconductor sector congestion and options Gamma leasing trading opportunities.

Core views

Macroeconomic Conditions and Analog Chip Performance: The S&P PMI came in at 54.5, higher than expected and the previous reading, with new orders rising month-on-month, indicating continued slight expansion in manufacturing. All analog chip companies that have released earnings reports so far (accounting for about half of industry revenue, including TXN, STM, Renesas, and NXP) exceeded expectations and raised guidance. First-quarter revenue and operating profit exceeded expectations by approximately 2.5% and 8%, respectively; second-quarter revenue guidance was raised by about 8.5%; full-year 2026 revenue forecasts were raised by about 8%. From a downstream perspective, although the automotive business was revised to negative, its structure was better than expected, with ADAS becoming a long-term growth driver; industrial demand is broadly recovering across all sectors and regions; data center demand is strong but already fully priced in; distribution channel inventories have fully normalized, though there are no signs yet of restocking. Semiconductor Gamma Leasing and Congestion: UBS’s derivatives team points out that semiconductors are currently the strongest subsector for Gamma leasing strategies. Since ADI and AMAT’s fiscal years end in October, their earnings reports come out one month later than peers, and given their relatively long-side congestion and cheaper volatility, they make ideal paired trading targets. In terms of congestion, after hitting a record high in April, sentiment in the semiconductor sector became even more positive. LRCX, AVGO, MU, and AMD had the highest long-side congestion; among 60 stocks, 10 reached or exceeded +24 (out of a maximum score of +30); the highest short-side congestion included PI (-14.1) and INDI (-8.7). NVDA’s congestion remained at a high of +25.1. Ultra-Large-Scale Manufacturers’ AI Capital Expenditures: Earnings reports from AMZN, GOOG, MSFT, and META reinforced the strong outlook for AI infrastructure spending. Driven by limitations in computing power and power capacity, growing customer demand for AI, and multi-year construction plans, capital expenditures continue to rise. UBS raised its 2027 capital expenditure forecast for ultra-large-scale manufacturers by 16% to $968 billion and by 11% to $1039 billion for 2028; these four giants account for over 80% of the updated industry forecast. Specifically, AMZN raised its 2027 capital expenditure forecast from $149 billion to $170 billion; META raised its 2026 capital expenditure guidance and mentioned collaborating with AVGO to develop self-designed chips exceeding 1 GW; MSFT noted that AI demand remains constrained by GPU, network, energy, and data center capacity limits; GOOG expanded capital expenditures due to Gemini and Cloud demand and supplied TPU hardware exceeding several GW.

Analysis framework

UBS’s analysis unfolds along several main lines—from macro to micro: Macroeconomic conditions and performance validation: First, use ISM/PMI manufacturing data to determine the direction of macroeconomic conditions, then combine it with actual performance and guidance changes from disclosed analog chip companies to validate industry demand turning points from bottom up, further breaking down revenue by end markets such as automotive, industrial, and data centers to assess the strength of recovery in each segment. Quantitative congestion analysis: Using UBS Quant Answers congestion factors (integrating prime broker positions, 13F regulatory filings, stock lending data, etc.), we quantify the long-short congestion levels of semiconductor subsectors and individual stocks (ranging from -30 to +30), helping to identify whether investor positions are overly concentrated and potential reversal risks. Derivatives volatility trading: Given that early-released companies provide valuable information for later-released peers, we analyze the predictable relationship between volatilities among peers to identify Gamma leasing trading opportunities (e.g., taking advantage of the time difference when ADI/AMAT release earnings later than peers). Industry chain capital expenditure transmission: By dissecting the capital expenditure guidance, capacity bottlenecks, and changing customer demands of ultra-large-scale cloud providers, we trace from top to bottom how AI infrastructure investment trends impact incremental demand for downstream chip companies (e.g., the change in NVDA’s data center revenue share of ultra-large-scale capital expenditures).

Methodology notes

• Quantitative/Factor/Portfolio TheoryStyle factor analysis

  Congestion Factor

  By integrating prime broker position data, 13F institutional holdings reports, stock lending data, and other sources, we quantitatively assess the congestion level of a stock’s long or short positions (scores ranging from -30 to +30). Higher positive scores indicate greater long-side congestion, while lower negative scores indicate greater short-side congestion. Extremely high congestion scores suggest investors may start shifting positions in the opposite direction, helping to identify potential reversal risks for the target stock.

• Event Gaming and Behavioral FinanceFund Flow/Chip Analysis

  Gamma Leasing Strategy

  Given that early-released companies during earnings season provide valuable information for later-released peers, we analyze the predictable relationship between volatilities among peers to execute paired trades or straddle options strategies. For example, since ADI/AMAT releases earnings one month later than peers, we can take advantage of the volatility changes brought by earlier companies like TXN to trade.

• Industry/Supply Chain Analysis FrameworkVolume-price decomposition

  End-Market Volume-Price Decomposition

  We break down semiconductor company revenues by downstream end markets such as automotive, industrial, data centers, and consumer electronics, further distinguishing between unit growth in sales volume and value growth from semiconductor content. For example, NXP pointed out that the core of its automotive semiconductor business lies in the growth of semiconductor content per vehicle overcoming fluctuations in vehicle production volumes.

• Industry/Supply Chain Analysis FrameworkIndustry Chain Upstream-Midstream-Downstream Transmission

  Capital Expenditure Transmission Analysis

  By tracking the capital expenditure changes and capacity bottlenecks of ultra-large-scale cloud providers (upstream buyers), we derive the incremental demand impact on downstream chip suppliers (such as GPU and network chip makers). For example, by calculating the proportion of NVDA’s data center revenue in ultra-large-scale capital expenditures, we assess how much AI investment benefits chip manufacturers.

Asset mapping & comparison

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

• NVDA (NVIDIA)

  Benefit: Ultra-large-scale manufacturers’ AI capital expenditures continue to rise, and AI demand exceeds available computing capacity

  Strengths

  Data center revenue share of ultra-large-scale capital expenditures continues to rise, and AI chip demand is strong

  Weaknesses

  Congestion remains at a high of +25.1, with obvious long-side congestion

• AMD

  Benefit: Expanding AI computing demand and growing custom chip demand from ultra-large-scale manufacturers

  Strengths

  The Agentic CPU theme has gained market attention, and congestion is rising

  Weaknesses

  High long-side congestion

  Comparison

  Compared with ARM and INTC, both belong to the computing sector and are favored by the market

• ADI

  Benefit: Recovery of the analog chip industry and Gamma leasing trading opportunities

  Strengths

  Is the best paired target for this quarter’s Gamma leasing strategy; its earnings report coming out one month later than peers provides a trading window

  Weaknesses

  Congestion declined over the past month, making it the biggest loser

  Comparison

  Its performance in the analog chip sector diverged from that of TXN and others

• AVGO (Broadcom)

  Benefit: Collaboration with META on custom chips and network chip demand

  Strengths

  Collaborated with META to develop self-designed chips exceeding 1 GW; high congestion makes it popular in the market

Key data

• S&P PMI54.5Higher than the previous reading of 54.0 and the expected 1.5 percentage points; new orders rose month-on-month
• First-Quarter Analog Chip Revenue Overperformance+2.5%Operating profit exceeded expectations by about 8%
• Second-Quarter Analog Chip Revenue Guidance Raise+8.5%Operating profit guidance raised by about 15%
• Full-Year 2026 Analog Chip Revenue Forecast Raise+8%Operating profit forecast raised by about 12%
• 2027 Ultra-Large-Scale Manufacturer Capital Expenditure Forecast$968 billionRaised by 16% from the previous forecast
• 2028 Ultra-Large-Scale Manufacturer Capital Expenditure Forecast$1039 billionRaised by 11% from the previous forecast
• NVDA Congestion Score+25.1Remained high, with obvious long-side congestion
• AMZN 2027 Capital Expenditure Forecast$170 billionRaised from the previous forecast of $149 billion
• Most Short-Side Congested Stock PI-14.1Highest short-side congestion

Impact & implications

UBS believes that the comprehensive overperformance of analog chip earnings and raised guidance indicate that the industry is in a broad recovery phase, especially driven by the widespread rebound in the industrial sector and the long-term growth potential of automotive ADAS, providing solid support for the industry. The continued upward revision of AI capital expenditure forecasts by ultra-large-scale manufacturers suggests that AI infrastructure investment remains in an acceleration phase; constraints in computing power and power capacity will be difficult to ease in the short term, providing marginal benefits for chip stocks such as NVDA, AMD, AVGO, MRVL, and ALAB. The momentum of capital expenditure growth also supports broad opportunities for general-purpose computing chips and custom accelerators/network chips.

Risks

• Investments in the technology sector carry above-average risks, including low sales visibility, rapid innovation and technological change, intense competition, frequent mergers and acquisitions, and many low barriers to market entry.
• Extremely high congestion scores (whether long or short side) suggest that investor positions may start shifting in the opposite direction.
• Some securities show above-average volatility.

What to watch

• Whether subsequent semiconductor companies’ earnings meet the volatility expectations of the Gamma leasing strategy
• The continued upward trend in ultra-large-scale manufacturers’ capital expenditure forecasts and whether AI demand can exceed the growth in computing power and power capacity
• Whether semiconductor distribution channels shift from inventory normalization to substantial restocking
• The sustained pull effect of increasing automotive ADAS penetration on semiconductor content