ENTITY: BROADCOM INC.

A Macro Intelligence Memo | June 2030 | Investor Edition

FROM: The 2030 Report DATE: June 28, 2030 RE: Broadcom's Valuation Normalization; AI Infrastructure Plateau and Shift to Infrastructure Utility Positioning; Strategic Pivot from Growth Play to Quality Infrastructure Provider

SUMMARY: THE BEAR CASE vs. THE BULL CASE

BEAR CASE: AI infrastructure growth slows to 2-3% CAGR. Custom silicon captures 50%+ market share. Margin compression continues (45-48%). Multiple compression to 14-15x. Stock declines to $90-110 (-26% to -7% downside).

BULL CASE: AI infrastructure growth reaccelerates to 12-15% annually. Custom silicon plateaus at 35-40%. HBM growth accelerates. Margins expand to 54-56%. Multiple expands to 20-22x. Stock rallies to $165-180 (+40% to +52% upside).

REALISTIC CASE: AI infrastructure 6-8% CAGR. Custom silicon stabilizes 35-40%. Margins stabilize 50-52%. Multiple holds 17-19x. Stock reaches $125-145 (-5% to +23% upside). Company trades as "quality infrastructure provider," not growth play.

EXECUTIVE SUMMARY

Broadcom Inc. (NASDAQ: AVGO) represents a significant case study in technology valuation correction and strategic repositioning within an artificial intelligence-driven market landscape that has proven less expansive than market consensus anticipated. The company achieved extraordinary valuation multiples during the AI infrastructure boom of 2024-2027 (peaking at 45x forward earnings in early 2027, implying market conviction in perpetual 30%+ growth rates), but has experienced material valuation correction (58% peak-to-trough decline) as empirical evidence of AI infrastructure growth constraints has accumulated between 2028-2030.

The correction, while painful for existing shareholders, has revealed Broadcom's actual strategic positioning: not as a high-growth artificial intelligence chip supplier, but as a high-quality, profitable infrastructure provider whose core networking, switching, and data center fabric businesses remain durable and valuable despite plateauing AI training workload growth. Current valuation (enterprise value approximately $185 billion, trading at 16-18x normalized earnings) represents fair-value pricing for a 6-8% organic growth infrastructure company with 55-60% operating margins and substantial capital return capacity.

Critical assessment: Broadcom is neither a value trap (unlikely to decline significantly further) nor a growth opportunity (near-term upside limited). The company is appropriately valued as a high-quality infrastructure provider with stable cash flows, but growth expectations should be reset from 20%+ (2024-2027 expectations) to 6-8% (2030+ realistic projection).

SECTION I: THE VANISHING EXPONENTIAL NARRATIVE (2024-2027)

The Bull Case at Peak Valuation (Early 2027)

The consensus narrative driving Broadcom's extraordinary valuation expansion during 2024-2027 was compelling and internally coherent:

Thesis Elements:

1. Exponential AI Scaling: Artificial intelligence model scale would expand according to scaling laws established by OpenAI and other research institutions. Larger models (100B → 500B → 1T+ parameters) would require proportionally more compute infrastructure.

2. Data Center Proliferation: Global data center deployment would expand from historical ~30 exabytes annual capacity addition to 100+ exabytes annually. Cloud providers would deploy 4-5x more data center capacity by 2030 relative to 2024 levels.

3. Custom Silicon Consolidation: While hyperscalers (Google, Amazon, Microsoft, Meta) would develop custom silicon for some applications, this would represent minority of workloads. Broadcom's universal networking and switching infrastructure would remain foundational to all data center architectures.

4. Margin Expansion: As AI infrastructure became standardized, Broadcom could achieve gross margins >70% and operating margins >60% on incremental capacity, driving exceptional earnings leverage.

5. Perpetual Demand: Unlike prior semiconductor cycles (which experienced periodic demand gluts), AI infrastructure demand would remain perpetually supply-constrained through 2035+, supporting above-cycle pricing power.

Market Validation: - Broadcom's own guidance (August 2026): "AI infrastructure revenue will compound at 40% CAGR through 2032" - Wall Street consensus: 12-month price targets implied 35-50x earnings multiples - Analyst commentary: "Only question is how fast AI infrastructure grows, not whether growth occurs" - Stock performance: 380% appreciation from January 2024 to peak in June 2027 (stock price $35 → $170)

The Inflection Point (August 2028 - March 2029)

Between August 2028 and March 2029, three separate developments undermined the consensus bull case:

Problem 1: Energy Economics of AI Training Plateau

By late 2028, academic research (particularly from MIT, Stanford, UC Berkeley) demonstrated that frontier AI model training had reached energy consumption levels that made further scaling economically irrational:

Computational Requirements: - GPT-4 equivalent training: ~13.5 gigawatt-hours electricity consumption - Electricity cost (at $60/MWh wholesale): ~$810,000 - Additional infrastructure amortization: $45-50 million - Total training cost: $47-52 million per frontier model

Economic Constraint: At $47-52 million per frontier model, only 20-30 frontier models could be economically trained annually globally. This represented a fundamental ceiling on model development—models would become more efficient (fewer parameters, better architectures) rather than larger.

Implication for Broadcom: Demand for infrastructure supporting 1T+ parameter models (which would require proportionally more networking and switching infrastructure) disappeared. Instead, demand shifted to supporting 7B-70B parameter models with different infrastructure requirements—models that could operate on standard cloud infrastructure without premium networking layers.

Revenue Impact: - Broadcom's infrastructure/networking revenue: 67% YoY growth in Q4 2027 - Decelerated to 12% YoY growth by Q2 2029 - Gross margin compression: 73% (2027) → 61% (2029)

Problem 2: Hyperscaler Custom Silicon Adoption

Concurrent with AI training plateau recognition, hyperscalers (Amazon, Google, Microsoft, Meta) accelerated custom silicon strategies specifically designed to bypass premium Broadcom networking and switching infrastructure:

Custom Chip Examples: - Amazon AWS: Trainium (training) and Inferentia (inference) chips with integrated switching fabric - Google: TPU v5+ with custom interconnect architecture - Microsoft: Maia chips with proprietary networking stack - Meta: Custom silicon initiatives targeting specific AI workloads

Market Share Impact: - Custom silicon share of AI training workloads: 8% (2026) → 34% (2029) - Broadcom addressable market reduction: Estimated $8-12 billion annually by 2029 - Margin pressure: Custom silicon initiatives forced hyperscalers to seek lower-cost networking solutions from alternative suppliers

Problem 3: AI Model Efficiency Shift

As AI model sizes plateaued, competition shifted from scale (larger models) to efficiency (better models with fewer parameters). Efficiency-focused architectures fundamentally changed infrastructure requirements:

Efficiency Trends: - Sparse architectures (not all parameters activated for every inference) - Distillation (compression of larger models into smaller, faster variants) - Retrieval-augmented generation (using external databases rather than model parameters) - Mixture-of-experts (selective activation of model subsets)

Infrastructure Impact: These efficiency approaches require less networking bandwidth, lower compute intensity, and lower memory access patterns than dense, large-scale models. They can operate effectively on standard cloud infrastructure without Broadcom's premium networking fabric.

Specific Margin Impact: - 7-billion parameter models: Infrastructure margin ~35-40% - 175-billion parameter models: Infrastructure margin ~68-72% - Shift from large to small models compressed margins by 30-40%

SECTION II: THE VALUATION CORRECTION AND MARKET OVERREACTION

The Correction Process (June 2027 - June 2029)

Broadcom's stock price trajectory during this period reflected market repricing:

Phases of Market Repricing

Phase 1: Denial (June 2027 - Feb 2028) - Market interpreted slower-than-expected growth as temporary - Analyst consensus maintained 30%+ growth forecasts - Stock decline limited to -8% from peak

Phase 2: Skepticism (March 2028 - July 2028) - Analyst research notes questioning sustainability of growth - First 2029 guidance cuts (-15% revenue expectations) - Stock decline accelerated to -25% cumulative

Phase 3: Panic (August 2028 - March 2029) - Academic papers on energy constraints published - Custom silicon adoption becoming obvious - Management guidance cut another -35% - Stock bottomed at $95 (-44% from peak)

Phase 4: Stabilization (April 2029 - June 2030) - Market re-evaluated Broadcom as "infrastructure utility" rather than growth play - Investors recognized residual value in core networking/switching business - Stock stabilized at $115-125 range

Market Overreaction Assessment

The question for June 2030 investors: Was the correction appropriate, or did it overestimate the severity of Broadcom's challenges?

Arguments for Overreaction: - Broadcom's core networking business (42% of revenue) remained durable - Operating margins, while compressed, remained 50-55% (above most technology companies) - Customer switching costs were high; architectural change takes 2-3 years - Capital return ($9-10B annually in dividends and buybacks) supported stock valuation

Arguments for Appropriate Correction: - Growth rate deceleration from 30%+ to 6-8% justified 3x+ valuation compression - Custom silicon threat was existential to historical margin profile - Margin compression from 73% to 61% represented meaningful operating leverage loss

Assessment: Market correction was appropriate; possibly slight overreaction on downside (-44% decline) but not dramatically so.

SECTION III: THE RESILIENT POSITIONING AND HIDDEN VALUE

The Enduring Data Center Fabric Challenge

Despite AI training workload plateau, the operational complexity of global data center infrastructure continues to drive demand for Broadcom's core networking and switching products:

Demand Drivers: 1. Inference Distribution: AI inference deploying across millions of edge devices, autonomous vehicles, edge servers globally; coordination requires sophisticated fabric 2. Heterogeneous Compute: Mixture of custom silicon (Google TPU), standard processors (Intel, AMD), and specialized hardware (FPGAs) requires orchestration 3. Multi-Cloud Operations: Enterprises running workloads across AWS, Azure, Google Cloud, and on-premises requiring unified networking layer 4. Real-Time Streaming: Distributed systems require low-latency, high-bandwidth networking Broadcom specializes in

Financial Resilience: - Broadcom networking/switching segment: 12.4% CAGR even during 2028-2029 downturn - Operating margin: 58% (among highest in semiconductor industry) - Customer concentration: Top 5 customers represent ~40% of revenue (distributed across industry) - Switching costs: High; replacing Broadcom infrastructure requires multi-year architecture redesign

VMware Strategic Positioning

The $61 billion VMware acquisition (2023) was widely criticized as bubble-era excess. By 2030, the acquisition proved strategically prescient:

Original Thesis: Broadcom sought to acquire end-to-end data center control (hardware networking + software orchestration). Skeptics argued this was defensive acquisition before growth decelerated.

2030 Reality: As enterprises fragmentized their AI infrastructure strategies (utilizing multiple cloud providers, custom chips, and on-premises systems), VMware's orchestration software became critical glue:

VMware 2030 Business: - Software revenue: $3.2-3.5 billion (grew 23% YoY despite overall sector challenges) - Gross margin: 74% (higher than legacy Broadcom) - Growth vector: AI/ML workload orchestration across multi-cloud environments - Customer penetration: 65%+ of Fortune 500 enterprises

Strategic Value: VMware transformed from "legacy virtualization company" to "critical AI infrastructure orchestration platform." This positioned Broadcom as end-to-end provider (hardware + software) for complex AI deployments.

High-Bandwidth Memory (HBM) Interconnect Opportunity

An overlooked opportunity for Broadcom: As AI model architecture shifted from compute-intensive to memory-intensive (particularly with attention mechanisms requiring massive memory bandwidth), high-bandwidth memory systems became critical bottleneck.

HBM Market Dynamics: - HBM demand: 156% growth between 2027-2029 - HBM pricing: $3,000-5,000 per stack (premium pricing) - Broadcom's role: Supplies interconnect technology (HBM stacks to GPUs/processors) - Interconnect revenue: 32% YoY growth 2028-2029

Future Opportunity: HBM represents growth vector for Broadcom within constraint environment. While AI compute growth plateaus, memory architecture complexity and HBM demand continue expanding.

Geopolitical Protection

U.S.-China semiconductor relationship deteriorated significantly in 2028-2029, creating regulatory environment favorable to Broadcom:

Protective Dynamics: 1. Export Restrictions: U.S. restricted export of advanced semiconductors to China; Broadcom benefited from U.S. government classification as "national security critical" 2. Reshoring Incentives: U.S. government provided subsidies for semiconductor domestic manufacturing; Broadcom benefited from subsidies for U.S. production 3. Allied Demand: NATO allies and allies (Japan, Korea, Taiwan, Australia) required U.S.-origin semiconductors; Broadcom captured this demand 4. Competitor Disadvantage: Taiwan Semiconductor and Samsung faced export restrictions or customer concentration risk

Estimated Impact: Geopolitical environment provided 5-10% revenue upside for Broadcom relative to prior baseline assumptions.

SECTION IV: FINANCIAL PROFILE AND VALUATION ANALYSIS

Current Financial Metrics (June 2030)

Valuation Multiples and Comparables

Current Valuation (June 2030): - Stock price: $118 - Diluted shares outstanding: ~1.6 billion - Market capitalization: $188 billion - Enterprise value: ~$185 billion

Valuation Multiples: - Price-to-Earnings (forward): 16.8x (on $7.05 EPS 2030 estimate) - Price-to-Sales: 4.8x - EV-to-EBITDA: 10.2x - Dividend yield: 2.4% - FCF yield: 6.2%

Comparable Analysis: | Company | Industry | P/E | Growth | |---------|----------|-----|--------| | Broadcom | Semiconductors | 16.8x | 5-6% | | Intel | Semiconductors | 22.1x | 3-4% | | Nvidia | Semiconductors | 38.2x | 18-22% | | Cisco | Networking | 15.3x | 4-5% | | Arista Networks | Networking | 28.4x | 15-18% |

Assessment: Broadcom's 16.8x multiple is appropriate for low-single-digit growth infrastructure company; positioned between low-growth traditional semiconductor companies and high-growth companies like Nvidia/Arista.

Scenario-Based Valuation Analysis

Bull Case (Probability: 20%)

Assumptions: - AI infrastructure growth reaccelerates to 12-15% annually due to new applications (robotics, autonomous vehicles, scientific computing) - Custom silicon adoption plateaus at 35-40% market share - Broadcom captures additional HBM-related revenue and margin expansion - Operating margin expands to 54-56% - Revenue reaches $45-48B by 2032

Valuation: - 2032 Operating Income: $24-27B - Valuation multiple: 20-22x (growth multiple) - Implied enterprise value: $480-594B - Implied 2030 stock price (present value discounted 3 years): $165-180 - Upside from current: 40-52%

Base Case (Probability: 55%)

Assumptions: - AI infrastructure growth 6-8% CAGR through 2032 - Custom silicon stabilizes at 35-40% market share - Margins stabilize at 50-52% - Revenue reaches $42-44B by 2032 - Modest multiple expansion to 17-19x as growth stabilizes

Valuation: - 2032 Operating Income: $21-23B - Valuation multiple: 17-19x - Implied enterprise value: $357-437B - Implied 2030 stock price: $125-145 - Downside/upside from current: -5% to +23%

Bear Case (Probability: 25%)

Assumptions: - AI infrastructure growth slows to 2-3% CAGR - Custom silicon captures 50%+ market share - Margin compression continues to 45-48% - Revenue grows to only $40-41B by 2032 - Multiple compression to 14-15x due to growth disappointment

Valuation: - 2032 Operating Income: $18-20B - Valuation multiple: 14-15x - Implied enterprise value: $252-300B - Implied 2030 stock price: $90-110 - Downside from current: -26% to -7%

Probability-Weighted Fair Value

Weighted fair value = (20% × $170) + (55% × $135) + (25% × $100) = $34 + $74 + $25 = $133

Current price: $118 Implied upside: +12.7%

SECTION V: INVESTMENT RECOMMENDATION AND STRATEGY

Portfolio Allocation Guidance

Recommended allocation: 3-5% of diversified portfolio

Rationale: - Quality of business: Broadcom operates profitable, cash-generative infrastructure company - Valuation: At $118, stock trades near fair value; limited risk/reward asymmetry - Growth: Expect 5-6% annual revenue growth (below market average but predictable) - Income: 2.4% dividend yield plus capital return via buybacks; total shareholder yield ~6% - Risk: Potential headwinds from continued custom silicon adoption or macroeconomic slowdown

Investment Thesis

Broadcom is appropriate for investors seeking: - Stable cash flow generation (high FCF yield) - Recession-resistant revenues (infrastructure remains essential in downturns) - Capital return (dividend + buyback program) - Modest growth (5-6% organic CAGR realistic)

Broadcom is NOT appropriate for investors seeking: - High growth (not a 15%+ CAGR company anymore) - Significant upside appreciation (limited from current valuation) - Technology leadership (infrastructure company, not innovation leader) - Defensive beta (higher than bond-like yields suggest)

THE DIVERGENCE: BEAR vs. BULL INVESTMENT OUTCOMES

FINAL INVESTMENT ASSESSMENT

BEAR CASE PATH: AI infrastructure growth disappoints (2-3% CAGR). Custom silicon threat materializes. Margins compress. Stock declines to $90-110 (-26% to -7%).

BULL CASE PATH: AI infrastructure reaccelerates (12-15% CAGR). HBM becomes growth vector. Margins expand. Stock rallies to $165-180 (+40% to +52%).

BASE CASE PATH: (Most likely, 55% probability) Infrastructure utility positioning. 5-6% organic growth. Margins stabilize 50-52%. Stock reaches $125-145 by 2032 (-5% to +23% upside). 3-year CAGR: 2-7%.

INVESTMENT RECOMMENDATION: RATING: HOLD (Fair Value)

At $118/share (16-18x earnings), Broadcom is appropriately priced as high-quality infrastructure provider. Neither value trap nor growth opportunity. Company generates substantial free cash flow (6%+ FCF yield) but faces limited multiple expansion potential.

Suitable for investors seeking: - Stable cash flow generation - Recession-resistant revenues - Dividend growth (modest) - Limited volatility

Not suitable for investors seeking: - High growth (5-6% CAGR realistic) - Significant appreciation upside - Technology leadership positioning

Probability-weighted fair value: $133/share (12% upside from current). Hold positions; wait for weakness below $105 before accumulating.

This assessment is prepared by The 2030 Report Investment Analysis division. Information is current as of June 28, 2030.

REFERENCES & DATA SOURCES

1. Broadcom 10-K Annual Report, FY2029 (SEC Filing)
2. Bloomberg Intelligence, "Semiconductor Design and Fabless Models: AI Chip Concentration," Q2 2030
3. McKinsey Global Institute, "AI Infrastructure Hardware: Design, Manufacturing, and Supply Chain," 2029
4. Gartner, "Magic Quadrant for Semiconductor Intellectual Property Core Processors," 2030
5. IDC, "Worldwide AI and Data Center Processor Market Share, 2025-2030," 2029
6. Goldman Sachs Equity Research, "Broadcom: Networking and AI Chip Exposure Analysis," April 2030
7. Morgan Stanley, "Fabless Design and Chip Supply Concentration: Geopolitical Risk," May 2030
8. Bank of America, "Broadcom's Diversification: Infrastructure Software and Networking," March 2030
9. Jefferies Equity Research, "Semiconductor Cycles: When Does Demand Normalize?," June 2030
10. Wedbush Securities, "Broadcom: VMware Integration and Margin Expansion," April 2030