Morgan Stanley Bullish on Broadcom AI: Target Price Raised to $485

Overview

This is a pre-earnings preview commentary by Morgan Stanley on Broadcom (AVGO) before its earnings release after market close on June 3, 2026. The firm maintains an Overweight rating and raises the target price from $470 to $485. The core thesis is: against the backdrop of strong demand for both AI networking and custom ASICs, Broadcom can meet high market expectations, but the current quarter may not be one of significant outperformance and guidance raise; the truly bigger growth story lies in 2027. The firm also modestly revised down its AI growth assumptions for H2 2026, due to timing delays in some rack deployments rather than demand loss.

Core views

Overall Assessment: The firm believes market expectations for this quarter are already very high, with channel checks on AI networking showing very strong results and sustained TPU demand. The logic is that frontier large model developers remain compute-constrained and will seize any available capacity, which benefits the GPU-alternative ecosystem (including Broadcom's ASIC/TPU) as a whole. The firm expects Q4 AI revenue of $10.8 billion and Q7 AI revenue of $16.9 billion; while not viewing this as a quarter of significant outperformance and substantial guidance raise, as long as management reiterates the long-term demand outlook, the stock can still move upward. 2027 Share Gain Story: The firm emphasizes the more important focus is XPU (custom chip) share gains in 2027. Broadcom's current AI market share is only ~10% and is expected to continue rising. The company currently sees customer compute scale approaching 10GW by FY2027, and using a rough framework of ~$20 billion per GW, the corresponding revenue opportunity is massive (though value per GW varies significantly by customer). The firm views this methodology as optimistic-biased but uses it to illustrate the magnitude of opportunity, currently modeling ~$120 billion in FY2027 AI revenue and believes there is still upside revision potential as ASIC projects ramp and AI networking outperforms. H2 Outlook Adjustment: The firm's view on H2 2026 has changed due to recent rack-related developments, believing original rack deployments have been postponed to 2027, likely including the previously discussed ~$21 billion opportunity. Crucially, the firm does not view this as demand loss but rather capacity being reallocated to other TPU customers, thus the originally expected 'hockey stick' steep inflection in H2 2026 will become more moderate—growth remains strong but the previously expected upside is more delayed than canceled. CoT and Competitive Risk: On CoT (packaging-related) issues, the firm continues to view MediaTek as a risk but not a thesis-disrupting factor, estimating ~10-15% of TPU value content is realistically at risk, but still believes Broadcom will retain most TPU share; replacing an incumbent supplier at this scale is very difficult, and Broadcom's TPU position is more solid than many investors realize. Non-AI Business: The firm expects stable but unremarkable performance, having not yet seen meaningful cyclical recovery signs in the broader semiconductor portfolio, but trends have stabilized; infrastructure software continues steady execution and remains a stable growth and cash flow source. Model and Valuation Adjustments: The firm lowered H2 2026 AI growth assumptions to reflect the above rack dynamics, keeping Q4 AI revenue unchanged at $10.8 billion, revising Q7 down from $19.5 billion (+80% QoQ) to $16.9 billion (+56% QoQ); expects Q4 total revenue of $22.1 billion and Q7 total revenue of $28.8 billion. Despite the Q7 outlook cut, full-year forecasts remain largely unchanged as the firm views the ~$3 billion AI revenue reduction as primarily timing misalignment related to rack availability rather than underlying demand changes, with most impact expected to be recouped in subsequent quarters. The firm currently forecasts 2026 calendar year AI revenue of $74 billion and 2027 of $133 billion. On valuation, raising the 2027 EPS multiple (including SBC) from 27x to 28x yields the higher $485 target price, and believes Broadcom has significant tailwinds on both custom silicon and networking fronts, is its preferred ASIC play, and sits at the core of AI infrastructure's most important trends.

Analysis framework

The firm's analytical framework follows a dual-layer structure of 'near-term earnings delivery + long-term share story'. In the near term, it uses channel checks (AI networking checks) and quarterly revenue/EPS modeling to judge whether Broadcom can meet high expectations, comparing against Street consensus and clearly stating that the key variable this quarter is management's commentary on long-term demand rather than the magnitude of single-quarter outperformance. In the long term, the firm quantifies opportunity using a 'compute scale × unit value' approach: first presenting ~10GW customer compute scale by FY2027, then deriving revenue opportunity using an empirical framework of ~$20 billion per GW, while acknowledging this is optimistic-biased and varies by customer content, thus using it as a tool to illustrate magnitude rather than precise forecasting. When handling the H2 growth revision, the firm specifically distinguishes between 'timing shift' and 'demand loss': interpreting the ~$21 billion rack opportunity delay as capacity reallocation among TPU customers, thereby maintaining full-year forecasts largely unchanged. This approach of 'splitting one-off timing disturbances from underlying trends' is key to its unwavering bullish thesis. On valuation, it uses P/E methodology, deriving target price from 2027 EPS multiplied by valuation multiple, and expresses enhanced confidence in growth prospects by raising the multiple.

Methodology notes

• Valuation MethodPE/PEG valuation

  Multiply future year EPS by valuation multiple to derive target price

  This report prices using 2027 EPS multiplied by valuation multiple: raising the multiple from 27x to 28x lifts the target price from $470 to $485. A higher multiple represents the market's willingness to pay more for the same earnings, and raising the multiple expresses enhanced confidence in growth prospects.

• Industry/Industrial Analysis FrameworkVolume-Price Breakdown

  Use 'compute scale (volume) × value per unit (price)' to estimate revenue opportunity, i.e., ~$20 billion per GW framework

  The firm first estimates customer compute approaching 10GW by FY2027, then multiplies by ~$20 billion per GW to derive potential revenue magnitude. This method of breaking total volume into 'quantity × unit price' facilitates quick market size estimation, but unit price (value per GW) varies by customer, so the result is only a magnitude reference rather than precise value.

• Event Game Theory and Behavioral FinanceExpectation Gap/Expectation Management

  When market expectations are already high, focus on the gap between actual results and expectations rather than absolute growth rate

  The report notes this quarter's 'expectations are already high', so even strong data may not constitute a significant outperformance quarter, and whether the stock can rise depends more on whether management reiterates long-term demand. This illustrates that for high-expectation stocks, the key is 'beating expectations or not' rather than looking at growth alone.

• Cycle and Prosperity FrameworkInventory cycle (Kitchin)

  Use inventory days (DOI) relative to historical median and seasonal changes to judge industry prosperity position

  The report tracks inventory days (DOI) for semiconductor companies and customers (e.g., industry inventory at 111 days, 22 days above historical median) to judge whether non-AI semiconductors remain in destocking phase and whether cyclical recovery is near. Inventory above normal levels typically means demand recovery will take more time.

Asset mapping & comparison

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

• Broadcom (Broadcom Inc., AVGO.US)

  Core subject of this report: As a key beneficiary of the GPU-alternative ecosystem, with strong AI networking and custom ASIC/TPU demand, significant share gain potential in 2027

  Strengths

  Leading position in merchant silicon and ASICs in data centers; strong AI networking channel checks, sustained TPU demand; AI market share only ~10% with room for improvement; infrastructure software provides stable cash flow; firm believes its TPU position is more solid than market perception

  Weaknesses

  Non-AI semiconductor portfolio has not shown meaningful cyclical recovery; Q7 gross margin (74.1%) below market expectations (76.0%); H2 AI growth revised down, ~$21 billion rack opportunity deferred to 2027

  Comparison

  Has the second-largest AI absolute dollar exposure in Morgan Stanley's semiconductor coverage; firm assigns Overweight rating and maintains its premium valuation relative to AI peers

  Risks

  Networking share loss to NVIDIA (Mellanox); insufficient ASIC chip competitiveness, customer churn to competitors; VMware acquisition integration execution risk; CoT/MediaTek could erode ~10-15% of TPU value content; new customer partnerships failing to ramp, revenue growth and synergies falling short (bear case)

Key data

• Target Price$485Raised from $470, corresponding to 2027 EPS valuation multiple rising from 27x to 28x
• Q4 AI Revenue$10.8 billionMaintain original forecast unchanged
• Q7 AI Revenue$16.9 billionQoQ +56%, revised down from $19.5 billion (QoQ +80%)
• Q4 Total Revenue$22.1 billionQoQ +14.6%, YoY +27.5%, largely in line with Street consensus of $22.0 billion
• Q7 Total Revenue$28.8 billionSlightly above Street consensus of $28.1 billion
• 2026/2027 Calendar Year AI Revenue$74 billion / $133 billionFull-year forecast largely unchanged, ~$3 billion reduction viewed as timing shift
• FY2027 AI Revenue Model~$120 billionStill has upside revision potential as ASIC ramps and AI networking outperforms
• Broadcom Current AI Market Share~10%Firm expects continued improvement
• FY2027 Customer ComputeApproaching 10GWRevenue opportunity estimated using ~$20 billion per GW framework
• Rack-Related Opportunity Deferred~$21 billionPostponed to 2027, viewed as timing shift rather than demand loss
• TPU Value Content Risk Exposure~10-15%Stemming from CoT/MediaTek competition, but firm believes Broadcom retains most share
• Q4 EPS / Gross Margin$2.40 / 76.9%EPS slightly above Street $2.39, gross margin above Street 76.1%
• Q7 EPS / Gross Margin$3.19 / 74.1%EPS slightly above Street $3.16, gross margin below Street 76.0%
• Scenario Range (Bull/Base/Bear)$619 / $485 / $298Corresponding to 2027 EPS multiples of 31x, 28x, 20x respectively

Impact & implications

The firm believes Broadcom sits at the core of AI infrastructure's most important trends and is its preferred ASIC play; 2026 and 2027 are poised to be strong years for XPU (custom chip) share gains, with the TPU platform continuing to gain momentum and driving incremental custom silicon demand. Near-term, H2 AI growth was modestly revised down, but as it's viewed as rack timing shift with unchanged demand, full-year forecasts remain largely unchanged; as long as management reiterates the long-term demand outlook in earnings, the stock can still move upward. Non-AI business and infrastructure software provide stable but unremarkable support.

Risks

• Networking share loss to NVIDIA (Mellanox)
• Insufficient ASIC chip competitiveness, customer churn to competitors
• VMware acquisition integration execution risk
• CoT-related competition from MediaTek, could impact ~10-15% of TPU value content
• New customer partnerships failing to ramp, revenue growth and synergies falling short (bear case)

What to watch

• Management's commentary on long-term demand outlook in the June 3 after-market-close earnings
• CoT/MediaTek-related developments (firm explicitly lists as area to monitor)
• Progress on rack deployments deferred to 2027 and capacity reallocation situation
• Cyclical recovery signs in core (non-AI) semiconductor business