AI spending remains the key semiconductor theme: strong outlooks for AVGO and AMBA, while SWKS improves long term but faces near-term pressure

Overview

This is a North American semiconductor weekly report previewing the upcoming quarterly results of AVGO and AMBA and summarizing information from SWKS’s Tech Day. The report considers AI capital spending the strongest ongoing theme and maintains constructive views on AVGO and AMBA. It is more optimistic about SWKS’s long-term business diversification, but near-term pressure leads it to maintain an Equal-weight rating.

Core views

The report first concludes that strong AI spending remains the most important industry theme during the off-cycle earnings period. AVGO and AMBA are both expected to report solid results and outlooks, although their drivers differ: AVGO is primarily driven by hyperscale customers’ custom ASIC, TPU, and AI networking buildouts, while AMBA is more exposed to automotive, low-power edge inference, and computer vision demand. The view on the North American semiconductor industry remains Attractive. For AVGO, Morgan Stanley expects the company to report after the market close on September 2. July-quarter AI revenue is expected to be $16.0 billion, up 48% sequentially, comprising $10.8 billion of custom ASIC revenue and $5.2 billion of AI networking revenue. October-quarter AI revenue is expected to rise another 32% sequentially to $21.2 billion, comprising $14.9 billion of custom ASIC revenue and $6.2 billion of AI networking revenue. Total July-quarter revenue is expected to be $29.4 billion, up 32.5% sequentially and 84.3% year over year, close to the $29.2 billion consensus estimate. Semiconductor Solutions revenue is expected to be $20.5 billion, up 36.6% sequentially and 123.6% year over year, while Infrastructure Software revenue is expected to be $8.9 billion, up 24.1% sequentially and 31.3% year over year. Gross margin is expected to be 74.0%, above the market’s 73.5%, and EPS is expected to be $3.24, slightly above the market’s $3.22. October-quarter total revenue is expected to be $34.8 billion, up 18.5% sequentially and 93.4% year over year, slightly above the market’s $34.7 billion. Semiconductor Solutions revenue is expected to be $25.9 billion, up 26.5% sequentially and 134.3% year over year, while Infrastructure Software revenue is expected to be $8.9 billion, flat sequentially and up 28.3% year over year. Gross margin is expected to be 72.9%, in line with the market, and EPS is expected to be $3.85, slightly above the market’s $3.84. The more important discussion for AVGO centers on 2027. Morgan Stanley maintains its approximately $120 billion AI revenue estimate and sees potential upside, while the company previously guided to 2027 AI revenue significantly above $100 billion and said growth could be sustained through 2028. Some investors’ expectations have already exceeded $150 billion. The report believes that level may be achievable but is not assured, so the key near-term risk is whether results can clear extremely high expectations rather than whether fundamentals are strong. TPU is expected to continue accounting for most custom ASIC revenue this year and next year, while new custom programs could gradually become important in 2027. Although MediaTek’s participation, AMD’s potential involvement in TPU v10, and Marvell’s warrant arrangement with Google all indicate that Google is broadening its supplier base, Morgan Stanley still expects AVGO to retain approximately 80% of the long-term TPU opportunity and views this as supplier diversification rather than AVGO being replaced. Recent downward revisions to some CoWoS forecasts are also viewed as noisy supply-chain information insufficient to change its strong growth outlook. AVGO remains one of Morgan Stanley’s preferred AI computing names, ranked behind only Top Pick NVIDIA. Supporting factors include its leadership in custom ASICs, strong networking business, gradual customer diversification, hyperscale capital-spending growth, potential cyclical recovery in non-AI semiconductor businesses, and opportunities from VMware integration, cost reductions, and stable cash flow. The $502 price target applies 28 times CY2027e ModelWare EPS of $17.92, equivalent to approximately 26 times non-GAAP EPS of $19.59. The report states that this valuation is broadly in line with or below AI peers. The risk-reward framework shows a current price of $368.79, a base case of $502, a bull case of $637, and a bear case of $308, implying 36% base-case upside. The bull case applies 31 times bull-case EPS of $20.55, while the bear case applies 20 times bear-case EPS of $15.40. The upside scenario depends on new xPU customers, continued networking strength, and VMware synergies. Key downside factors include networking share shifting to NVIDIA, ASIC competitiveness or customer production ramps falling short of expectations, and poor execution of the VMware integration. For AMBA, the report expects the company to report after the market close on September 3. July-quarter revenue is expected to be $108.3 million, up 7.9% sequentially and 13.3% year over year, above the $107.8 million consensus estimate. IP Security revenue is expected to rise 9.7% sequentially and 14.5% year over year, while automotive revenue is expected to rise 4.0% sequentially and 21.3% year over year. Gross margin is expected to be 59.8%, in line with the market, while EPS is expected to be $0.16, slightly below the market’s $0.17. October-quarter revenue is expected to be $121.2 million, up 12.0% sequentially and 11.8% year over year, above the $119.3 million consensus estimate. IP Security revenue is expected to rise 15.3% sequentially and 12.3% year over year, while automotive revenue is expected to rise 3.0% sequentially and 19.5% year over year. Gross margin is expected to be 59.7%, close to the market’s 59.6%, while EPS is expected to be $0.29, slightly below the market’s $0.30. The report expects the automotive business to grow approximately 20% for the full year and outpace IoT, driven by fleet telematics, AI video, safety applications, and rising AI content per vehicle. Enterprise security is relatively healthy, but consumer IoT could come under pressure in the second half due to rising storage and memory costs, representing the main near-term risk. AMBA’s medium- to long-term thesis is shifting from a sharp product-cycle upswing toward steadier execution. Drones are viewed as one of the most attractive near-term opportunities, spanning prosumer and commercial applications. The CV7 ramp is expected to generate incremental growth and support higher average selling prices. Automotive remains a long-term opportunity, but there was no major update last quarter on a broader autonomous-driving ramp for CV3, and the report explicitly notes slow progress in L2+ development. Morgan Stanley remains optimistic about AMBA’s technological position in low-power edge AI inference and computer vision, as well as its multiyear opportunities in automotive, drones, robotics, and physical AI infrastructure. Regarding unverified discussions of an NXP acquisition, the report does not confirm whether a transaction exists or will occur. It merely sees potential strategic complementarity between AMBA’s edge AI capabilities and NXP’s scale, automotive customer relationships, and OEM trust. Its Overweight thesis remains based primarily on the underlying technology rather than M&A expectations. AMBA’s $96 price target represents approximately 9 times CY2026e revenue of $410 million. The bull case of $136 represents 10 times revenue of $537 million. The company page shows a bear case of $36, corresponding to 3.5 times revenue of $352 million, while the industry risk-reward summary table shows a bear case of $34. Based on the summary table’s current price of $70.63, base-case upside is 36%. For SWKS, its Tech Day presentations reinforced the report’s positive view of its long-term opportunities. In mobile, the company won the mid-band position for the first time during Google’s roughly biennial rebid and now supplies nearly 100% of the RF front-end content in Pixel. The agreement was extended through 2030, with cumulative revenue expected to exceed $1 billion. The company also emphasized its close cooperation with MediaTek and early visibility into next-generation RF requirements. Beyond mobile, the company demonstrated how its RF, timing, isolation, sensing, and power capabilities could extend into higher-growth markets. Its precision-timing solution integrates network synchronization, jitter attenuation, clock generation, and software into a single chip. Data-center opportunities focus on high-voltage power architectures, isolation, gate driving, and sensing, with the company positioning itself in the 800V portions of the grid-to-rack chain. It also addressed automotive, WiFi/broadband, drones, and other edge-connectivity applications. Near-term Broad Markets conditions remain mixed, as strength in data centers and automotive is offset by IoT weakness. Smartphone demand is temporarily stable but still faces memory-cost pressure. Over the long term, a higher Broad Markets mix could reduce the company’s dependence on mobile, particularly amid continued Apple share losses. The Qorvo transaction and integration are key catalysts. The report maintains an Equal-weight rating and a $72 price target. The risk-reward table shows a current price of $65.79, a bear case of $43, a base case of $72, and a bull case of $105, implying 9% base-case upside. On valuation, the report estimates combined CY2027 EPS after transaction synergies at approximately $7, above SWKS’s standalone EPS of $5.31 and implying approximately 13 times earnings. However, given transaction and synergy uncertainty, it favors using standalone EPS and a 13.5-times multiple. Industry data show continued AI strength alongside traditional inventory pressure. Semiconductor company days of inventory stand at 114 days, up two days sequentially and 23 days above the historical median. Semiconductor customer days of inventory rose nine days sequentially to 60 days, compared with a seasonal increase of eight days, and remain above the historical median. Distributor days of inventory fell two days sequentially to 61 days, compared with a typical seasonal increase of four days, but remain seven days above the historical median. Short interest is also highly differentiated: as of August 28, 2026, AVGO, AMBA, and SWKS short interest as a percentage of shares outstanding was 1.2%, 9.3%, and 19.8%, respectively, compared with an average of 5.5% and a median of 5.0% among covered companies.

Analysis framework

The report first uses AI capital spending and industry inventories to identify the primary cyclical theme. It then builds quarterly revenue, segment, gross-margin, and EPS models for AVGO and AMBA and compares them with consensus expectations to identify potential earnings or guidance surprises. It subsequently uses bull, base, and bear scenarios and valuation multiples to establish company-level risk-reward ranges, mapping supplier changes, customer demand, product ramps, and M&A integration to scenario drivers. For SWKS, it combines Tech Day presentations, customer wins, business mix, and the Qorvo transaction to assess the long-term diversification path, and finally supplements its industry and market-positioning analysis with days of inventory and short-interest data.

Methodology notes

• Event-Driven Trading and Behavioral FinanceExpectation Gap/Expectation Management

  Comparison of Morgan Stanley forecasts with consensus expectations

  The report compares revenue, gross margin, EPS, and next-quarter guidance with market expectations item by item to determine whether, despite strong fundamentals, the figures are sufficient to exceed the high bar already set by investors.

• Valuation methodsPE/PEG valuation

  P/E valuation based on forward EPS

  AVGO’s price target is calculated using 28 times CY2027e ModelWare EPS of $17.92; due to transaction risk, SWKS is valued using standalone EPS of $5.31 and a 13.5-times multiple.

• Valuation methods

  EV/Sales valuation

  AMBA’s base-case price target is valued at approximately 9 times CY2026e revenue of $410 million, while the bull and bear cases alter both revenue assumptions and enterprise-value-to-revenue multiples.

• Event-Driven Trading and Behavioral Finance

  Bull, base, and bear scenarios with option-implied probabilities

  The report establishes different operating and valuation assumptions for each company and uses option-implied volatility as of August 28, 2026 to estimate approximate risk-neutral probabilities that share prices exceed scenario prices within three months or one year.

• Cycle and Business Conditions Framework

  Comparison of days of inventory with seasonality and historical medians

  The report separately tracks days of inventory and sequential changes for semiconductor companies, customers, and distributors to assess whether supply-chain inventory pressure is improving or accumulating.

Asset mapping & comparison

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

• Broadcom (AVGO.US)

  AI capital spending drives company revenue through custom ASICs, TPUs, and AI networking buildouts; the report ranks it as the preferred AI computing name behind only NVIDIA.

  Strengths

  Leadership in custom ASICs, data-center networking advantages, customer diversification, management execution, and VMware cash-flow synergies.

  Weaknesses

  Market expectations are extremely high, while TPU supplier diversification continues to create valuation and sentiment pressure.

  Comparison

  The price target represents approximately 26 times CY2027e non-GAAP EPS, which the report says is broadly in line with or below AI peers; it ranks behind only NVIDIA in preference.

  Risks

  Networking share shifting to NVIDIA, declining ASIC competitiveness or customer programs failing to reach production, and VMware synergies falling short of expectations.

• Ambarella (AMBA.US)

  Automotive, IP security, drones, and other edge AI applications constitute its sources of growth.

  Strengths

  Leading low-power AI inference and computer vision technology, with automotive design wins and CV7 expected to support growth and higher average selling prices.

  Weaknesses

  L2+ autonomous-driving development is relatively slow, while consumer IoT demand may be affected by rising memory and storage costs.

  Comparison

  Approximately 9 times CY2026e EV/Sales, representing a modest premium to small-cap semiconductor companies.

  Risks

  Automotive revenue ramping more slowly than expected, stalled CV replacement in security cameras, and CV products failing to expand beyond initial surveillance applications.

• Skyworks Solutions (SWKS.US)

  The Pixel RF front-end win and expansion in data centers, automotive, and precision timing support long-term business diversification.

  Strengths

  It has secured nearly all Pixel RF front-end content and possesses a combined portfolio of RF, timing, isolation, sensing, and power capabilities.

  Weaknesses

  Broad Markets performance is mixed, IoT is soft, and Apple share continues to decline.

  Comparison

  Equal-weight maintained; the $72 price target represents 9% base-case upside from the current price of $65.79.

  Risks

  Cancellation or poor integration of the Qorvo transaction, competition from RF localization in China, concentration and pricing pressure from Apple accounting for approximately 60% of total sales, and failure of gross margins to recover.

• NVIDIA (NVDA.US)

  As the Top Pick within Morgan Stanley’s semiconductor coverage, it is the principal benchmark for AVGO’s AI computing position and networking competitive risk.

  Strengths

  The report ranks it as its first choice among preferred AI computing names.

  Comparison

  AVGO is ranked as the second-preferred AI computing name, immediately behind NVIDIA.

  Risks

  Risks listed in the report include weaker-than-expected AI end markets, customers reducing GPU purchases, AMD reemerging as an effective GPU competitor, and cloud customers developing proprietary hardware.

Key data

• AVGO July-quarter AI revenue$16.0 billionUp 48% sequentially; custom ASICs account for $10.8 billion and AI networking for $5.2 billion
• AVGO October-quarter AI revenue$21.2 billionUp 32% sequentially; the report breaks this down into $14.9 billion of custom ASIC revenue and $6.2 billion of AI networking revenue
• AVGO 2027 AI revenue estimateApproximately $120 billionThe company previously guided to significantly above $100 billion; some market expectations exceed $150 billion
• AVGO July-quarter total revenue and EPS$29.4 billion; $3.24Market expectations are $29.2 billion and $3.22, respectively
• AVGO October-quarter total revenue and EPS$34.8 billion; $3.85Market expectations are $34.7 billion and $3.84, respectively
• AVGO price target and risk-rewardBase case $502; bull case $637; bear case $308Current price $368.79, with 36% base-case upside
• AMBA July-quarter revenue$108.3 millionUp 7.9% sequentially and 13.3% year over year, above the $107.8 million consensus estimate
• AMBA October-quarter revenue$121.2 millionUp 12.0% sequentially and 11.8% year over year, above the $119.3 million consensus estimate
• AMBA full-year automotive business growthApproximately 20%Expected to outpace IoT, driven by fleet telematics, AI video, safety applications, and rising AI content per vehicle
• AMBA price target$96Current price $70.63, with 36% base-case upside shown in the report; bull case $136
• SWKS Pixel agreementExtended through 2030, with cumulative revenue exceeding $1 billionSWKS has secured nearly 100% of Pixel’s RF front-end content
• SWKS risk-rewardBear case $43; base case $72; bull case $105Current price $65.79, with 9% base-case upside
• Supply-chain days of inventoryCompanies 114 days; customers 60 days; distributors 61 daysUp two days, up nine days, and down two days sequentially, respectively; all three remain above historical medians
• Short interest in key companiesAVGO 1.2%; AMBA 9.3%; SWKS 19.8%As of August 28, 2026; covered-company average 5.5% and median 5.0%

Impact & implications

The report believes AI infrastructure investment remains sufficient to support high growth in AVGO’s custom ASIC and networking businesses, with 2027 and beyond representing the principal source of fundamental value. However, the near-term share-price reaction will depend more on whether the company can exceed extremely high expectations. AMBA is transitioning from a rapid product cycle toward steady execution. Automotive, drones, and CV7 could expand its edge AI opportunity, while consumer IoT cost pressure and the pace of automotive ramps will determine the near-term trajectory. SWKS’s technology portfolio and Google Pixel win demonstrate a path to reducing its dependence on mobile, but Broad Markets have not yet fully recovered, while Apple share losses and execution of the Qorvo transaction keep the near-term view neutral. Overall industry inventories remain above historical levels, although sequential improvement at distributors indicates that recovery is not synchronized across the supply chain.

Risks

• Even if AVGO’s results are very strong, they could trigger a negative reaction if they fail to exceed investors’ extremely high expectations.
• AVGO’s TPU supplier diversification, changes in CoWoS forecasts, and the progress of new custom programs could affect the market’s assessment of long-term AI revenue.
• AVGO could lose networking share to NVIDIA or lose customers due to insufficient ASIC competitiveness.
• VMware integration, cost reductions, and synergy realization may fall short of expectations.
• AMBA’s consumer markets could face additional demand pressure from rising memory and storage costs in the second half.
• AMBA’s automotive revenue and CV3 autonomous-driving business could ramp later than expected, while CV penetration in security cameras may also slow.
• SWKS faces Apple share losses, approximately 60% customer concentration, and potential pricing pressure.
• SWKS’s Qorvo transaction could be canceled or poorly integrated, while RF localization competition in China and gross-margin recovery also present risks.
• Days of inventory at semiconductor companies, customers, and distributors remain above their respective historical medians.

What to watch

• Monitor AVGO’s July-quarter results and October-quarter AI revenue guidance, scheduled after the market close on September 2, 2026.
• Monitor whether AVGO updates its 2027 AI revenue framework and the 2027 ramp timing of new custom ASIC programs.
• Monitor changes in TPU v10 suppliers, AVGO’s long-term TPU share, and AI networking demand.
• Monitor AMBA’s results after the market close on September 3, 2026 and whether October-quarter revenue reaches the $121.2 million model.
• Monitor approximately 20% full-year growth in AMBA’s automotive business, the CV7 ramp, drone demand, and consumer IoT cost pressure.
• Monitor whether verifiable new information emerges regarding the unconfirmed AMBA and NXP discussions.
• Monitor completion and integration progress for the SWKS-Qorvo transaction and whether data centers and automotive can drive a recovery in Broad Markets.
• Monitor subsequent changes in days of inventory at semiconductor companies, customers, and distributors.