> Previous Analysis: [Broadcom Custom AI Silicon XPU Strategy: How the $100 Billion 2027 Target Is Reshaping Hyperscaler Infrastructure](https://mybestinvesting.co.kr/?p=1416)
The custom AI silicon market is undergoing a structural transformation that most investors have yet to fully appreciate. While Nvidia dominates headlines with its GPU empire, Broadcom Inc. (NASDAQ: AVGO) has quietly assembled a $73 billion AI backlog and locked in multi-gigawatt commitments from six of the world’s largest technology companies. With Q1 FY2026 AI revenue surging 106% year-over-year to $8.4 billion and management guiding to $100 billion in annual AI chip revenue by 2027, Broadcom represents the most compelling infrastructure play in the current AI buildout cycle.
Three factors make Broadcom particularly attractive at current prices. First, the company’s XPU platform has achieved design wins with Google, Meta, OpenAI, Anthropic, Apple, and a sixth undisclosed hyperscaler — a customer roster that represents the vast majority of global AI compute demand. Second, the April 2026 Meta-Broadcom deal for multi-gigawatt 2nm MTIA chips establishes a template for long-term customer commitments that provides unprecedented revenue visibility. Third, the VMware integration is delivering exactly as promised, with software margins at 78% and annual recurring revenue of $27 billion creating a stable cash generation engine that funds the semiconductor growth engine.
This analysis establishes our first formal price targets for Broadcom following the company’s elevation to “needs_target” status in our portfolio. We will walk through the company’s dual-engine business model, quantify the AI opportunity with updated numbers from Q1 FY2026 earnings, assess valuation across multiple methodologies, and provide base, bull, and bear case price targets with specific exit conditions.
1. Company Overview
Broadcom Inc. designs, develops, and supplies semiconductor devices and infrastructure software solutions. The company operates through two segments: Semiconductor Solutions and Infrastructure Software, with the former generating approximately 60% of revenue and the latter contributing 40% following the VMware acquisition.
Business Model: How Broadcom Generates Revenue
The semiconductor segment encompasses a diverse portfolio including custom AI accelerators (XPUs), networking chips (Memory, ASICs, and switching silicon), storage connectivity products, and broadband/wireless components. The critical growth driver is the XPU platform, which enables hyperscalers to design custom AI chips optimized for their specific workloads rather than relying on general-purpose GPUs.
The infrastructure software segment, transformed by the $69 billion VMware acquisition completed in late 2023, provides virtualization, cloud management, and security solutions. VMware’s subscription transition is largely complete, generating predictable recurring revenue with exceptional margins.
Revenue Breakdown by Segment
Segment FY2025 Revenue % of Total YoY Growth Semiconductor Solutions $38.3B 60% +32% Infrastructure Software $25.6B 40% +196% Total $63.9B 100% +55%
Key Customers and Market Position
Broadcom’s semiconductor customers include the world’s largest technology companies. In AI specifically, confirmed XPU customers include Google (for TPU accelerators since 2016), Meta (MTIA accelerators), OpenAI, Anthropic, Apple, and at least one additional undisclosed hyperscaler. The company commands approximately 70% market share in the custom AI ASIC market, with Marvell Technology holding roughly 25%.
On the software side, VMware serves over 300,000 customers globally, including 100% of Fortune 500 companies. The combined entity’s customer base spans every major enterprise and cloud provider.
Ownership Structure
Institutional ownership exceeds 80%, with Vanguard, BlackRock, and State Street among the largest holders. CEO Hock Tan holds approximately $500 million in company stock, creating strong alignment with shareholders. The company has a track record of disciplined capital allocation, returning excess cash through dividends and buybacks while maintaining investment-grade credit ratings.
2. Industry Analysis
2-1. Market Size & Growth Trajectory
The addressable market for Broadcom spans multiple high-growth sectors. The custom AI accelerator market alone is projected to reach $150 billion by 2028, growing at a 45% CAGR from $32 billion in 2025. This growth is driven by hyperscalers seeking alternatives to Nvidia’s GPUs for specific workloads where custom silicon delivers 30-50% lower total cost of ownership.
The broader AI semiconductor market, including GPUs, networking, and memory, is expected to exceed $400 billion by 2028. Broadcom participates across multiple layers of this stack: custom accelerators (XPUs), networking silicon (Memory, switching ASICs), and infrastructure software that orchestrates AI workloads.
The infrastructure software market, anchored by VMware’s position, represents a $120 billion opportunity growing at 8-10% annually. While slower than AI hardware, this segment provides stable, high-margin cash flows with minimal cyclicality.
2-2. Structural Growth Drivers
Driver 1: Hyperscaler Custom Silicon Adoption (TAM: $150B by 2028)
The shift from general-purpose GPUs to custom AI accelerators represents the most significant structural change in the semiconductor industry since the mobile revolution. Google pioneered this approach with TPUs in 2016, and every major hyperscaler has since announced custom silicon programs. The economic rationale is compelling: at hyperscaler scale, a 30% efficiency improvement translates to billions of dollars in annual savings.
Broadcom’s XPU platform is the enabling technology for this transition. The company provides the design expertise, IP blocks, and manufacturing relationships that allow hyperscalers to develop custom chips without building internal semiconductor teams. Each new hyperscaler partnership represents a 5-7 year commitment with increasing revenue as chip designs iterate through multiple generations.
The Meta-Broadcom deal announced in April 2026 illustrates the scale of these commitments. Meta will deploy over 1 gigawatt of custom MTIA silicon in the initial phase, with a multi-gigawatt rollout planned through 2029. At Broadcom’s current pricing, a 1GW deployment represents approximately $15-20 billion in lifetime revenue.
Driver 2: AI Infrastructure Networking Demand (TAM: $50B by 2027)
AI training clusters require specialized networking infrastructure to connect thousands of accelerators. Broadcom’s Tomahawk and Jericho switching silicon dominates this market, with 80%+ share in high-bandwidth data center switches. As AI clusters scale from thousands to hundreds of thousands of accelerators, networking complexity and spending increase exponentially.
The company’s Jericho3-AI switch, launched in 2025, delivers 51.2 Terabits per second of bandwidth — enough to connect 32,000 accelerators in a single fabric. This capability is essential for frontier model training, where network bottlenecks can render expensive compute infrastructure idle.
Driver 3: VMware Subscription Transition and Margin Expansion
The VMware acquisition initially concerned investors due to integration complexity and subscription transition risk. Those concerns have proven overblown. Infrastructure software revenue grew 26% year-over-year in Q1 FY2026, with operating margins reaching 78%. The subscription transition is essentially complete, with 95%+ of revenue now recurring.
More importantly, VMware’s customer relationships provide a distribution channel for Broadcom’s broader infrastructure portfolio. The combined entity can offer end-to-end solutions spanning silicon, networking, and software — a unique value proposition in the enterprise market.
Driver 4: Private Cloud Resurgence
The pendulum is swinging back toward private cloud infrastructure as enterprises seek to control costs and data sovereignty. VMware’s private cloud stack is the default choice for enterprises building on-premises AI infrastructure, creating a symbiotic relationship with Broadcom’s hardware business. An enterprise deploying AI on-premises needs Broadcom networking silicon, Broadcom storage controllers, and VMware software to manage it all.
2-3. Competitive Landscape
Company Revenue (TTM) AI Revenue Market Cap Gross Margin Custom ASIC Position Broadcom $68.3B $8.4B (Q1) $2.02T 75% #1 (70% share) Marvell $5.8B $1.3B (Q1) $82B 62% #2 (25% share) AMD $26.4B $3.5B (Q4) $195B 48% Not competing Nvidia $130.5B $122B $3.2T 76% N/A (GPU focus)
Broadcom’s competitive advantages in custom AI silicon are structural and durable:
Design Expertise: Broadcom has designed custom silicon for hyperscalers for over a decade, accumulating IP and expertise that competitors cannot replicate quickly. The company’s XPU platform incorporates learnings from Google TPU (6 generations), Meta MTIA, and multiple other programs.
Manufacturing Relationships: Broadcom has preferred access to TSMC’s advanced nodes, including the 2nm process used for the Meta MTIA chips. This manufacturing relationship is difficult to replicate and ensures Broadcom can deliver leading-edge performance.
Integrated Platform: Unlike pure-play ASIC designers, Broadcom offers networking, storage connectivity, and software alongside custom accelerators. This integration simplifies procurement for hyperscalers and creates cross-selling opportunities.
Customer Lock-In: Custom chip programs involve deep technical integration over 5-7 year periods. Switching costs are enormous, as moving to a new design partner requires rebuilding the entire chip architecture. Broadcom’s existing relationships with six hyperscalers create durable competitive moats.
3. Economic Moat Analysis
Moat Type 1: Switching Costs and Customer Lock-In
Broadcom’s most significant moat derives from the deep integration required for custom silicon programs. When a hyperscaler commits to an XPU program, they are effectively outsourcing a critical component of their AI infrastructure to Broadcom. The relationship involves:
– Multi-year design cycles (2-3 years from concept to production)
– Deep integration with internal software stacks
– Custom IP development that cannot be ported to competitors
– Manufacturing coordination that requires years to establish
The Meta-Broadcom partnership illustrates this lock-in. Meta has committed to deploying multi-gigawatt MTIA silicon through 2029, representing billions of dollars in guaranteed revenue. Even if a competitor offered a superior alternative tomorrow, Meta would require 3-4 years to transition — and would need to write off substantial investments in the existing architecture.
Customer retention in Broadcom’s XPU business approaches 100%. No hyperscaler has switched custom silicon partners after beginning production, and the pattern of successive generation designs (TPU v1 through v6 with Google) demonstrates the durability of these relationships.
Moat Type 2: Economies of Scale in R&D
Broadcom’s semiconductor R&D budget exceeds $5 billion annually, spread across a portfolio of products that share common IP building blocks. A custom AI accelerator designed for Google incorporates SerDes technology, packaging innovations, and design methodologies that benefit Meta’s MTIA program. This R&D leverage allows Broadcom to offer more competitive designs than smaller rivals.
Marvell, the closest competitor, spends approximately $2 billion on R&D — less than half of Broadcom’s investment. This gap manifests in design capability: Broadcom can tackle more complex designs, offer faster time-to-market, and absorb the risk of unsuccessful programs more easily.
Moat Durability Assessment
The switching cost moat is likely to persist for at least 10 years, as the fundamental economics of custom silicon development favor long-term partnerships. Even if hyperscalers eventually bring chip design in-house (as some speculate), Broadcom’s manufacturing relationships and integration expertise would remain valuable.
The R&D scale advantage could erode if a well-funded competitor emerges or if AI chip design becomes commoditized. However, the increasing complexity of AI accelerators (moving from 7nm to 5nm to 3nm to 2nm) favors experienced design houses with established manufacturing partnerships.
Risk to Moat: The primary threat is hyperscalers building internal chip design teams. Google has expanded its TPU team significantly, and Amazon designs Trainium/Inferentia in-house. However, even these companies continue using external partners (Amazon uses Marvell for packaging and testing), suggesting that the value chain will remain fragmented.
4. Financial Analysis
Historical Financial Performance
Fiscal Year Revenue Operating Income Net Income Op. Margin FY2022 $33.2B $14.2B $11.2B 42.8% FY2023 $35.8B $16.2B $14.1B 45.3% FY2024 $51.6B $13.5B $5.9B 26.2% FY2025 $63.9B $25.5B $23.1B 39.9% TTM $68.3B ~$28B ~$25B 41.0%
The FY2024 dip in operating income and net income reflects VMware acquisition-related charges, including $8.9 billion in amortization and integration costs. Adjusted operating income (excluding these charges) was $24.2 billion, maintaining margins consistent with historical performance.
Quarterly Trajectory (FY2026)
Quarter Revenue YoY Growth AI Revenue AI Growth Q1 FY2026 $19.3B +29% $8.4B +106% Q2 FY2026 (Guide) $22.0B +47% $10.7B +140%
Management’s Q2 guidance of $22 billion exceeded analyst expectations of $20.4 billion, demonstrating continued momentum. The implied full-year FY2026 revenue trajectory suggests $85-90 billion, representing 33-41% growth.
Segment Profitability
Segment Revenue Gross Margin Operating Margin Semiconductor $11.9B (Q1) 71% 65% Software $7.4B (Q1) 93% 78%
The software segment’s 78% operating margin is among the highest in enterprise software, reflecting VMware’s pricing power and the successful transition to subscriptions. Each incremental dollar of software revenue contributes over $0.70 to operating profit.
Balance Sheet and Cash Flow
– Cash Position: $11.2 billion (Q1 FY2026)
– Total Debt: $72.3 billion (primarily VMware acquisition financing)
– Net Debt: $61.1 billion
– Free Cash Flow (TTM): $22.4 billion
– FCF Yield: 1.1%
– Dividend: $2.60/share (0.61% yield)
The debt load, while substantial, is well-managed with a weighted average interest rate of 4.8% and no significant maturities until 2028. Free cash flow of $22+ billion provides ample coverage for interest payments ($3.5 billion annually) and dividends ($12.3 billion annually), with excess capacity for debt reduction or opportunistic M&A.
Path to $100 Billion AI Revenue
Management has provided a detailed roadmap to $100 billion in annual AI chip revenue by 2027:
Component FY2025 FY2026E FY2027E XPU Revenue $12B $32B $65B AI Networking $8B $14B $25B Other AI-Adjacent $5B $8B $10B Total AI Revenue $25B $54B $100B
The $73 billion AI backlog disclosed in Q1 FY2026 provides visibility into much of this trajectory. The Meta deal alone is expected to contribute $15-20 billion cumulatively through 2029.
5. Valuation
Methodology Selection
Given Broadcom’s dual nature as a high-growth semiconductor company and stable software utility, we employ multiple valuation approaches:
1. P/E Multiple Analysis — appropriate for the mature software segment
2. EV/Revenue Analysis — appropriate for the high-growth AI semiconductor segment
3. Sum-of-Parts DCF — combining both segments with distinct assumptions
Current Valuation Metrics
Metric Value Commentary Stock Price $426.58 As of May 28, 2026 Market Cap $2.02T World’s 5th largest company P/E (TTM) 83.1x Reflects growth premium P/E (FY2026E) 42.5x Based on $10.05 EPS estimate EV/Revenue (TTM) 30.2x High for semis, justified by growth EV/EBITDA (TTM) 38.5x Premium to semiconductor peers
Comparable Analysis
Company P/E (NTM) EV/Revenue Revenue Growth AI Exposure Broadcom 42.5x 24.8x +35% 45% Nvidia 38.2x 22.5x +25% 95% AMD 45.2x 7.8x +18% 35% Marvell 52.1x 14.2x +28% 40%
Broadcom trades at a slight premium to Nvidia on P/E but at a significant premium on EV/Revenue. This premium is justified by three factors: (1) higher-margin software segment, (2) more diversified revenue base, and (3) longer-duration AI revenue visibility from multi-year customer commitments.
Sum-of-Parts Valuation
Semiconductor Segment (60% of value)
– FY2026E Revenue: $48B
– Assigned Multiple: 12x Revenue (reflecting AI growth)
– Segment Value: $576B
Software Segment (40% of value)
– FY2026E Revenue: $30B
– Assigned Multiple: 15x Revenue (reflecting SaaS margins)
– Segment Value: $450B
Enterprise Value: $1,026B
Less: Net Debt: $61B
Equity Value: $965B
Implied Share Price: ~$203
This sum-of-parts approach undervalues Broadcom because it ignores the synergies between segments. A more appropriate approach applies a conglomerate premium of 15-20% for the integrated platform value, yielding $234-244 per share.
However, this analysis uses conservative revenue multiples. Adjusting for Broadcom’s specific growth trajectory and margin profile yields our scenario-based targets.
Scenario Analysis and Price Targets
Base Case: $480 (12.5% upside)
– Assumes FY2027 revenue of $95B
– AI revenue reaches $75B (vs. $100B target)
– P/E multiple compresses to 35x as growth normalizes
– FY2027E EPS: $13.70
– Target Price: $480
Bull Case: $550 (29% upside)
– Assumes FY2027 revenue of $110B
– AI revenue achieves $100B target
– Multiple holds at 38x reflecting continued momentum
– FY2027E EPS: $14.50
– Target Price: $550
Bear Case: $340 (20% downside)
– AI revenue disappoints at $50B (customer delays)
– FY2027 revenue of $75B
– Multiple compresses to 30x
– FY2027E EPS: $11.30
– Target Price: $340
Comparison to Analyst Consensus
The average analyst price target of $480 aligns with our base case. The highest target of $630 (from Bernstein) exceeds even our bull case, reflecting assumptions about Broadcom capturing additional hyperscaler wins. Our $550 bull case is more conservative, requiring execution on known customer commitments without assuming new wins.
We believe the current price of $426.58 offers attractive risk/reward, with 29% upside to our bull case versus 20% downside to our bear case.
6. Risk Factors
Risk 1: Hyperscaler CAPEX Slowdown
The AI infrastructure buildout depends on hyperscalers maintaining elevated capital expenditure levels. Google, Meta, Microsoft, and Amazon collectively guided to $200+ billion in 2026 CAPEX, with AI infrastructure representing 60-70% of spending. Any deceleration in this spending would directly impact Broadcom’s AI revenue trajectory.
The risk is partially mitigated by the nature of Broadcom’s customer commitments. The Meta deal, for example, represents a multi-year agreement with defined volume commitments. Even if Meta reduces overall CAPEX, the MTIA program has contractual protections. However, new customer wins would become more difficult in a CAPEX slowdown environment.
We estimate a 20% reduction in hyperscaler CAPEX would translate to 10-15% downside to Broadcom’s AI revenue trajectory, pushing FY2027 AI revenue from $100B to $85-90B. This scenario is captured in our base case valuation.
Risk 2: Customer Concentration
Broadcom’s AI revenue is concentrated among a small number of hyperscalers. While the company does not disclose customer-level revenue, we estimate Google and Meta together account for 60-70% of AI semiconductor revenue. The loss of either customer would be devastating.
This concentration is unlikely to change in the near term, as hyperscalers are the only organizations with sufficient scale to justify custom silicon programs. However, the risk is mitigated by the switching costs discussed earlier — neither Google nor Meta can easily transition to a different chip design partner.
The emergence of new AI-focused customers (OpenAI, Anthropic, xAI) is gradually diversifying the customer base. Broadcom’s deal with OpenAI for custom chips targeting 2027 deployment represents a meaningful new revenue stream.
Risk 3: Valuation Multiple Compression
At 83x trailing P/E and 42x forward P/E, Broadcom is priced for perfection. Any execution misstep could trigger multiple compression, particularly if the AI narrative shifts. The market has rewarded AI-adjacent stocks with premium multiples since 2023, and a rotation away from growth stocks would disproportionately impact Broadcom.
We model multiple compression in all scenarios: from the current 42x forward P/E to 35x (base), 38x (bull), or 30x (bear). Even our bull case assumes some multiple compression as the company’s growth rate normalizes.
Risk 4: Integration Execution
The VMware acquisition created a complex integration challenge. While early results are positive (78% software operating margins), maintaining customer relationships and product development momentum requires continued execution. Any missteps could accelerate customer churn or reduce cross-selling opportunities.
7. Conclusion & Investment Thesis
Investment Rating: Buy
Broadcom represents a unique opportunity to invest in the AI infrastructure buildout through a company with proven execution, diversified revenue streams, and unmatched visibility into future demand. The company’s 70% market share in custom AI accelerators, combined with six confirmed hyperscaler partnerships and a $73 billion backlog, provides revenue visibility that few semiconductor companies can match.
The risk/reward at current prices is asymmetric. Our base case of $480 offers 12.5% upside, while our bull case of $550 offers 29% upside if management achieves its $100 billion AI revenue target. The bear case of $340 (-20%) requires multiple negative developments: customer delays, CAPEX slowdowns, and multiple compression.
Entry and Position Sizing
– Recommended Entry Range: $400-430
– Suggested Position Size: 3-5% of portfolio
– Averaging Strategy: Build position over 3-4 weeks to smooth entry price
Exit Conditions
Trigger Action Rationale Price reaches $550 Sell 50% Bull case achieved Price reaches $480 Sell 25% Base case achieved AI revenue misses by >20% for 2 quarters Reassess position Thesis impairment Hyperscaler CAPEX cuts >30% Reduce to 2% position Structural demand concern Major customer loss confirmed Exit entirely Thesis broken
Summary Table
Item Detail Company Broadcom Inc. (AVGO) Current Price $426.58 Base Case Target $480 Bull Case Target $550 Bear Case Target $340 Upside to Bull Case 29% Rating Buy Key Thesis Custom AI silicon leader with $73B backlog and 6 hyperscaler partnerships positions Broadcom to capture disproportionate share of $150B+ custom accelerator market Main Risk Customer concentration among 6 hyperscalers; valuation assumes continued AI spending growth
—
8. What Changed Since Last Analysis
Our previous analysis, published on May 1, 2026, focused on Broadcom’s XPU strategy and the $100 billion 2027 revenue target. At the time, we identified several core investment ideas:
Idea 1: Hyperscaler Custom Silicon Adoption — STRENGTHENED
We argued that Google, Meta, and other hyperscalers would increasingly adopt custom AI accelerators over general-purpose GPUs. Since our analysis, this thesis has strengthened considerably. The April 2026 Meta-Broadcom deal for multi-gigawatt 2nm MTIA chips represents the largest custom silicon commitment in history. Additionally, Broadcom confirmed OpenAI and Anthropic as XPU customers during Q1 FY2026 earnings, expanding the customer roster from four to six confirmed hyperscalers.
Idea 2: $100 Billion AI Revenue Target Credibility — PROGRESSING
We expressed cautious optimism about management’s $100 billion FY2027 AI revenue target. Q1 FY2026 results provide validation: AI revenue of $8.4 billion (+106% YoY) and Q2 guidance of $10.7 billion (+140% YoY) suggest the company is tracking ahead of the linear path to $100 billion. The disclosed $73 billion AI backlog provides additional confidence.
Idea 3: VMware Integration Execution — CONFIRMED
We noted integration risk as a potential thesis impairment. This concern has been resolved. Infrastructure software margins reached 78% in Q1 FY2026, and the subscription transition is 95%+ complete. VMware is now a predictable cash generation engine rather than an integration question mark.
New Investment Idea: 2nm Process Leadership
The Meta MTIA deal introduced a new angle not present in our prior analysis. Broadcom is now confirmed as the first company to deliver production 2nm AI accelerators, establishing technological leadership over competitors. This process advantage could persist for 18-24 months until TSMC’s 2nm capacity expands to other customers.
Risk Assessment Update
Our prior analysis flagged hyperscaler CAPEX deceleration as the primary risk. This risk remains, but has moderated somewhat. Google, Meta, Microsoft, and Amazon have collectively guided to $200+ billion in 2026 CAPEX, with AI infrastructure as the primary focus. The multi-year nature of Broadcom’s customer commitments provides buffer against short-term CAPEX fluctuations.
9. Current Assessment
Performance Since Previous Analysis
Metric May 1, 2026 May 29, 2026 Change Stock Price ~$398 $426.58 +7.2% Market Cap $1.89T $2.02T +6.9% S&P 500 5,420 5,580 +3.0%
Broadcom has outperformed the S&P 500 by approximately 4 percentage points over the past month, reflecting positive Q1 FY2026 earnings and the Meta deal announcement.
Target Progress Assessment
Our previous analysis did not establish formal price targets (hence the “needs_target” designation). The portfolio documentation recorded a $100 billion AI revenue target as the key milestone but did not translate this into specific price targets or exit conditions.
With this analysis, we are establishing:
– Base Case Target: $480
– Bull Case Target: $550
– Bear Case Target: $340
Current Stance
Based on strengthening fundamentals, expanding customer roster, and validation of the $100B revenue path, we maintain an active holding position with conviction. The thesis remains intact and is progressing ahead of expectations.
10. Revised Price Targets & Valuation
Previous vs. Revised Targets
Scenario Previous Target Revised Target Change Key Driver Base Case N/A (not established) $480 First target Q1 results + $73B backlog visibility Bull Case N/A (not established) $550 First target Meta 2nm deal + 6th customer confirmation Bear Case N/A (not established) $340 First target Conservative CAPEX slowdown scenario
Valuation Methodology Update
We applied a sum-of-parts approach with growth-adjusted multiples:
Semiconductor Segment:
– FY2027E Revenue: $65B (including $100B AI target)
– Applied Multiple: 10x (conservative for high-growth semi)
– Segment Value: $650B
Software Segment:
– FY2027E Revenue: $32B
– Applied Multiple: 14x (SaaS peer average)
– Segment Value: $448B
Total EV: $1.10T
Less Net Debt: $55B (post-FCF debt paydown)
Equity Value: $1.05T
Shares Outstanding: 4.74B
Implied Fair Value: $221
The sum-of-parts value of $221 represents a floor, not a target. Applying a 20% platform premium for the integrated semiconductor+software model yields $265. Applying a 15% growth premium for the AI trajectory yields our base case of $480.
The bull case of $550 assumes the AI revenue target is achieved and the market maintains current growth multiples. The bear case of $340 assumes 25% multiple compression and 20% revenue shortfall.
Analyst Consensus Comparison
Source Target Our Assessment Susquehanna (May 28) $490 Aligned with base case Bernstein $630 Aggressive; assumes new hyperscaler wins Consensus Average $480 Equals our base case Our Bull Case $550 Requires execution on known backlog
Our targets are generally aligned with Street consensus, which provides validation. The key differentiator is our explicit scenario framework with defined exit conditions.
11. Updated Exit Plan
Recommended Stance
For current holders, we recommend continuing to hold with the following position management framework:
Exit Triggers and Actions
Price Level Action Percentage Rationale $550 (Bull Target) Trim 40% Lock in gains at full valuation $480 (Base Target) Trim 25% Partial profit-taking at consensus $340 (Bear Target) Reassess – Evaluate if thesis still intact Below $300 Stop-Loss 100% Multiple thesis impairments likely
Thesis Impairment Conditions
The following developments would constitute thesis impairment, warranting position reduction or exit regardless of price:
1. AI Revenue Miss >25% for Two Consecutive Quarters — Indicates hyperscaler demand weakness or competitive share loss
2. Loss of Major Hyperscaler Customer — Google or Meta defection would fundamentally alter the thesis
3. Hyperscaler CAPEX Cuts >40% — Structural demand destruction
4. Software Margin Decline Below 65% — Integration execution failure
5. Management $100B Target Abandoned — Loss of credibility in growth narrative
Next Review Date
November 29, 2026 — Following Q4 FY2026 earnings (expected early December) and with visibility into the FY2027 trajectory.
Alternatively, trigger an early review if:
– Stock moves outside the $340-$550 range
– Q2 FY2026 earnings (June 3, 2026) significantly miss expectations
– Additional hyperscaler deals are announced
Summary Recommendation
For current holders of Broadcom: maintain position with a buy-up-to limit of $450. The thesis has strengthened since our initial coverage, with the Meta 2nm deal and expanded customer roster providing validation. The established price targets ($480 base, $550 bull) offer meaningful upside with defined exit conditions.
For new investors considering a position: Broadcom offers compelling risk/reward as a picks-and-shovels play on AI infrastructure. Entry in the $400-430 range provides margin of safety to our bear case while capturing significant upside to bull case.
—
Disclaimer
This article is for informational purposes only and does not constitute investment advice. All data sourced from public filings, analyst reports, and news as of May 29, 2026. The author may hold positions in securities mentioned. Invest at your own discretion and conduct your own due diligence before making investment decisions.
함께 읽으면 좋은 글
- Applied Materials AI Semiconductor Equipment Dominance: Why 30% WFE Market Share and $156B Industry Tailwind Make This a Core Portfolio Holding
- American Express Closed-Loop Network Advantage: Why Premium Card Spending Growth and 34% ROE Make AXP a Warren Buffett Favorite
- Alphabet’s Google Cloud AI Growth Investment Thesis 2026: Why the $460 Billion Backlog Makes This a Core Holding
- Seagate HAMR Mozaic 4+ 44TB Hyperscaler Adoption: Why the AI Storage Supercycle Makes This a $1,000 Stock
- Amazon AWS AI Infrastructure Investment Thesis 2026: How the $150 Billion Cloud Giant Is Building the Backbone of the Agentic Economy