Executive Summary

Investment Thesis (3 Sentences)

AMD is a semiconductor company executing exceptionally well on its multi-year transformation from a struggling CPU maker to a leading provider of high-performance computing solutions for data centers, PCs, and gaming. The company's Instinct AI accelerators (MI300/350 series) position it as the only credible competitor to Nvidia's dominance in the $500B+ AI accelerator market by 2028. However, current valuation at 122x trailing P/E and 37x forward P/E prices in near-flawless execution on AI growth, leaving minimal margin of safety.

Key Metrics Dashboard

Decision: WAIT - Buy at $160 or below

Position Size: 2-3% at target price Primary Risk: Nvidia's ecosystem lock-in; AI capex slowdown Catalyst: MI350 launch mid-2025; Data center GPU revenue scaling

---

Phase 0: Opportunity Identification (Klarman)

Why Does This Opportunity Exist?

Answer: It doesn't exist at current prices.

AMD trades at a significant premium reflecting AI growth expectations. The opportunity will only emerge if:

1. Market overreaction to near-term earnings miss - One weak quarter could cause a 30%+ selloff given high expectations
2. Nvidia competitive announcement - New Blackwell Ultra or pricing pressure could compress AMD's perceived opportunity
3. AI infrastructure spending pause - Hyperscaler capex slowdown would hit all AI hardware names
4. Macro correction - High-beta stocks like AMD (β=1.95) fall more than market in downturns

Current State: No mispricing evident. Stock is up 91% in past year, trading near 52-week highs. Institutional ownership at 70%. Analyst consensus is overwhelmingly bullish (40 Buy, 11 Hold, 0 Sell).

---

Phase 1: Risk Analysis (Inversion Thinking)

"How Could This Investment Lose 50%+ Permanently?"

Top 10 Risk Events

Total Expected Downside: -67.25% (risk-adjusted weighted)

Deep Dive: Top 3 Risks

1. Nvidia Ecosystem Lock-in (CUDA Moat)

• Nvidia's CUDA has 15+ years of developer adoption
• 95% of AI frameworks optimized for CUDA first
• ROCm is improving but still requires porting effort
• Hyperscalers with internal software teams (Microsoft, Meta) can adapt; enterprises may struggle
• Mitigation: AMD's open-source strategy, growing ROCm support (1M+ Hugging Face models)

2. AI Capital Expenditure Cycle

• 2024-2025 hyperscaler capex is at record levels ($200B+ combined)
• DeepSeek-style efficiency gains could reduce compute requirements per model
• "Stargate" and similar announcements suggest multi-year runway, but cycles are inevitable
• A 6-12 month pause would devastate AMD's growth narrative at 122x P/E
• Mitigation: Diversification across PC, gaming, embedded (50% of revenue)

3. TSMC Dependency

• 100% of leading-edge chips manufactured by TSMC
• Any Taiwan geopolitical event = catastrophic
• TSMC prioritizes Apple > Nvidia > AMD in supply allocation
• CoWoS capacity constraints already impacting MI300 ramp
• Mitigation: Long-term agreements; ZT Systems acquisition for system-level integration

The Bear Case (3 Sentences)

AMD's AI GPU revenue ($5B in 2024) is 10% of Nvidia's ($50B+), with inferior margins and weaker ecosystem lock-in. The company's trailing ROE of 2.9% and 122x P/E suggest investors are paying for a future that requires AMD to become the next Nvidia—a low-probability outcome. If AI infrastructure spending normalizes or Nvidia maintains >80% market share, AMD's valuation compresses to 30x earnings ($60-80/share).

Pre-Defined Sell Triggers (Non-Price)

1. Thesis Break: MI350 market share below 10% of addressable market by Q4 2025
2. Moat Erosion: Microsoft, Meta, or Google announce exclusive long-term Nvidia deals
3. Management Failure: Lisa Su departure without credible successor; R&D investment cuts
4. Competitive: Intel Gaudi 3 or custom hyperscaler chips gain 20%+ AI training share

---

Phase 2: Financial Analysis

ROE Decomposition (DuPont, 5 Years)

Analysis: 2022-2024 ROE collapsed due to Xilinx acquisition ($49B) massively increasing equity base and intangible amortization depressing net income. Pre-acquisition, AMD was a high-ROE compounder. Current ROE is distorted by purchase accounting; normalized ROE (ex-intangibles) would be ~8-10%.

Owner Earnings Calculation

Owner Earnings (2024):
  Net Income:          $1.64B
  + D&A:               $3.18B
  - Maintenance CapEx: $0.32B (50% of total CapEx)
  - ΔWorking Capital:  $0.80B
  ─────────────────────────────
  Owner Earnings:      $3.70B

Per Share (1.63B shares): $2.27

Valuation Trinity

1. Liquidation Value (Floor)

Net Current Assets: $19.0B - $7.3B = $11.7B
Tangible Book: $57.6B - $24.8B (GW) - $18.9B (Intang.) = $13.9B

Per Share: $13.9B / 1.63B = $8.53

Not relevant for going concern; AMD is not a cigar butt.

2. DCF Valuation (Owner Earnings)

Conservative Assumptions:

• Owner Earnings 2024: $3.7B
• Growth Years 1-5: 25% CAGR (AI ramp)
• Growth Years 6-10: 10% CAGR (maturation)
• Terminal Growth: 3%
• Discount Rate: 11% (WACC)

Terminal Value: $17.95B × (1.03) / (0.11 - 0.03) = $231B

DCF Value: $178B → $109/share

Bull Case (35% CAGR years 1-5): $156/share Bear Case (15% CAGR years 1-5): $78/share

3. Private Market Value

Recent semiconductor M&A multiples:

• Xilinx: 11.4x EV/Revenue (2022)
• Nvidia market: 25x+ EV/Revenue

AMD at 5x Revenue (private market discount): $130B → $80/share AMD at 8x Revenue (strategic premium): $208B → $128/share

Valuation Summary

Conclusion: Current price ($232) is significantly above all valuation methods. No margin of safety exists.

Margin of Safety Calculation

Intrinsic Value Estimate (weighted average):

• 40% DCF Conservative: $109
• 30% DCF Bull: $156
• 20% Private Market (8x): $128
• 10% Owner Earnings (25x): $57

Weighted IV: $119/share

Current MOS: ($119 - $232) / $119 = -95% (OVERVALUED)

---

Phase 3: Moat Analysis

Moat Sources

Moat Width Assessment: NARROW

AMD has a defensible position but lacks the wide moat of Nvidia (CUDA ecosystem) or Intel's historical dominance (now eroding). The chiplet architecture is replicable; TSMC manufactures for everyone.

Forces of Erosion

Key Question: "Wider or Narrower in 10 Years?"

Assessment: Narrower → Same

• Data center GPU moat depends on ROCm achieving CUDA parity (uncertain)
• x86 CPU moat stable if Intel execution remains poor
• Gaming and embedded moats stable but not growing
• AI accelerator landscape highly dynamic; winner not determined

---

Phase 4: Management & Capital Allocation

Lisa Su - CEO Since 2014

Track Record:

• Transformed AMD from near-bankruptcy to $377B market cap
• Executed Zen architecture pivot, regained CPU competitiveness
• Acquired Xilinx ($49B) for adaptive computing
• Accelerated AI roadmap to annual product cadence

Compensation (2024):

• Base Salary: ~$1.3M
• Total Compensation: ~$30M (90% equity-linked)
• Ownership: 0.44% ($1.7B+)

Assessment: Exceptional operator with strong alignment. One of the best CEOs in tech.

Capital Allocation (Last 5 Years)

Verdict: Excellent capital allocation focused on long-term competitiveness.

Insider Activity (24 Months)

• Lisa Su: Regular planned sales (10b5-1) for diversification
• No unusual insider buying at current prices
• Low insider ownership (0.44%) typical for mature tech

---

Phase 5: Catalyst Analysis

Potential Catalysts

Negative Catalysts

---

Phase 6: Decision Synthesis

Position Sizing Formula

At current prices ($232), ZERO position recommended.

Target entry prices:

• Strong Buy: $95 (20% below DCF Bear) - 59% decline needed
• Accumulate: $120 (intrinsic value estimate) - 48% decline needed
• Fair Value: $150 (DCF Bull scenario)
• Take Profits: N/A (not buying)

Expected Return Probability Tree

Monitoring Metrics

---

Final Recommendation

┌─────────────────────────────────────────────────────────────────┐
│                     INVESTMENT RECOMMENDATION                    │
├─────────────────────────────────────────────────────────────────┤
│ Company: Advanced Micro Devices      Ticker: AMD                │
│ Current Price: $231.83               Date: 2026-01-17           │
├─────────────────────────────────────────────────────────────────┤
│ VALUATION SUMMARY                                                │
│ ┌─────────────────────────┬─────────────┬─────────────────────┐ │
│ │ Method                  │ Value/Share │ vs Current Price    │ │
│ ├─────────────────────────┼─────────────┼─────────────────────┤ │
│ │ DCF (Conservative)      │ $109        │ -113% (overvalued)  │ │
│ │ DCF (Bull)              │ $156        │ -49% (overvalued)   │ │
│ │ DCF (Bear)              │ $78         │ -197% (overvalued)  │ │
│ │ Owner Earnings (15x)    │ $34         │ -582% (overvalued)  │ │
│ │ Owner Earnings (25x)    │ $57         │ -307% (overvalued)  │ │
│ │ Private Market Value    │ $80-128     │ -81% to -113%       │ │
│ └─────────────────────────┴─────────────┴─────────────────────┘ │
│                                                                  │
│ INTRINSIC VALUE ESTIMATE: $119 (weighted average)               │
│ MARGIN OF SAFETY: -95% (OVERVALUED)                             │
├─────────────────────────────────────────────────────────────────┤
│ RECOMMENDATION:  [ ] BUY  [ ] HOLD  [ ] SELL  [X] WAIT          │
├─────────────────────────────────────────────────────────────────┤
│ STRONG BUY PRICE:        $95  (59% below current)               │
│ ACCUMULATE PRICE:        $120 (48% below current)               │
│ FAIR VALUE:              $150                                   │
│ TAKE PROFITS:            N/A                                    │
│ SELL PRICE:              N/A                                    │
├─────────────────────────────────────────────────────────────────┤
│ POSITION SIZE: 0% (wait for entry)                              │
│ CATALYST: MI350 launch mid-2025; AI infrastructure spending     │
│ PRIMARY RISK: Nvidia ecosystem lock-in; valuation compression   │
│ SELL TRIGGER: Data center GPU growth <30% YoY                   │
└─────────────────────────────────────────────────────────────────┘

Quality Assessment

Quality Grade: A-

Exceptional management, strong product execution, growing moat in data center. Deducted for:

• Low current ROE (purchase accounting distortion)
• Dependency on TSMC
• Narrow moat vs. Nvidia

Tier: T2 Resilient

Strong growth profile with meaningful competitive position, but not a fortress business.

---

Appendix: Sources Used

Primary Documents Downloaded

MCP Data Retrieved

Data Validation