AMD is a very richly valued stock....now and in future....very dangerous if / when recession hits
Non-GAAP Trailing P/E Ratios (Dec 2025)
| Company | Ticker | Non-GAAP P/E (TTM) | Why the Adjusted Figure Matters | | AMD | AMD | ~57.9 | Removes ~$2B+ in quarterly non-cash amortization from the Xilinx merger. | | NVIDIA | NVDA | ~45.2 | Earnings are "cleaner" than peers, so the GAAP/Non-GAAP gap is small. | | Broadcom | AVGO | ~28.5 | GAAP P/E is ~69; Non-GAAP removes VMware integration costs. | | Micron | MU | ~27.3 | Strips out inventory write-downs and cyclical accounting adjustments. | | TSMC | TSM | ~29.2 | Trades closer to its GAAP figure as it grows organically without major M&A. | | ASML | ASML | ~38.8 | Reflects core lithography machine sales, excluding minor structural shifts. | | Qualcomm | QCOM | ~34.7 | Reflects strong mobile recovery and expansion into automotive/PC. | | Intel | INTC | ~92.0* | Even non-GAAP remains high as Intel’s margins are still recovering. |
Export to Sheets
Analysis: What these numbers tell us- AMD vs. NVIDIA: While AMD’s GAAP P/E looks "scary" (often over 100), the 57.9 Non-GAAP figure shows it is actually trading at only a slight premium to NVIDIA. Investors are paying more for AMD's potential to gain market share in AI accelerators (MI350 series).
- The Broadcom "Discount": At ~28.5x non-GAAP, Broadcom is significantly cheaper than NVIDIA and AMD. This is because a larger portion of its revenue comes from stable, slower-growing networking and software vs. pure-play AI training chips.
- Intel's Anomaly: Intel’s P/E (even non-GAAP) is essentially "broken" right now. Because their earnings-per-share ($0.23 last quarter) are so much lower than their historical $1.00+ levels, the P/E ratio looks massive. It is a "turnaround play" rather than a "growth play."
- Manufacturing Stability: TSMC and Micron remain the "value" plays of the sector, trading at multiples below 30x despite being the literal foundation of the AI era.
Market Insight: Many institutional investors have shifted focus to Forward P/E (2026 estimates) because 2025 was a "transitional" year for AI infrastructure. For example, AMD’s Forward P/E is currently near 33x, suggesting the market expects its earnings to nearly double next year.
Would you like me to pull the Forward P/E estimates to see which of these looks like the best value for 2026?
2026 Forward P/E Ratios (Non-GAAP)
| Company | Ticker | Forward P/E (2026 Est.) | Projected Earnings Growth | Investment Narrative | | NVIDIA | NVDA | ~24.5 | ~60% | The "value" play of AI? Despite its size, its massive earnings growth is making the stock look historically cheap on a forward basis. | | AMD | AMD | ~33.1 | ~63% | Trading at a premium to NVIDIA. Markets are betting on the MI350/MI400 series becoming a true second-source for AI GPUs. | | TSMC | TSM | ~18.5 | ~20% | The foundation of the industry. Historically trades at a discount because of geopolitical risk, but looks highly attractive below 20x. | | Broadcom | AVGO | ~23.3 | ~38% | Transitioning from networking to custom AI silicon (ASICs). Valuation has compressed as VMWare integration stabilizes. | | ASML | ASML | ~29.0 | ~10-15% | Priced for steady growth. Investors are watching for the "2026 Transition" as foundries move to High-NA EUV machines. | | Micron | MU | ~10.8 | ~100%+ | The cheapest in the sector. Memory is highly cyclical; this low P/E reflects the peak of the projected 2026 earnings boom. | | Intel | INTC | ~61.4 | ~76% | Still the most expensive on a forward basis because its profit margins haven't fully recovered from its massive fab investments. |
Export to Sheets
Key Takeaways from the Forward Data- NVIDIA vs. AMD: Surprisingly, NVIDIA is currently "cheaper" than AMD on a 2026 forward basis (~24.5x vs ~33x). This suggests the market is either skeptical that NVIDIA can maintain its growth or is over-optimistic about AMD's ability to steal market share.
- The "Deep Value" of Memory: Micron trading at 10.8x forward earnings highlights the cyclical nature of the business. Investors are cautious that 2026 might be the "top" of the memory cycle, keeping the multiple low despite massive current profits from HBM (High Bandwidth Memory).
- Broadcom’s Evolution: Broadcom is now trading in the same valuation neighborhood as NVIDIA (~23x). This marks a shift in how the market views them—no longer just a "boring" networking company, but a critical AI infrastructure play.
Why these numbers might change- Rubin & Blackwell (NVIDIA): If the 2026 "Rubin" architecture launch exceeds expectations, that 24.5x P/E will look like a steal.
- Intel 18A: If Intel’s new manufacturing node succeeds in 2026, its earnings will likely spike, causing that high 61.4x P/E to crash toward 20x very quickly.
- HBM Demand: If AI server demand slows, Micron’s "cheap" 10x P/E could become a "value trap" as earnings estimates get slashed.
Would you like me to look into the PEG Ratio (Price/Earnings to Growth) for these companies? It's often the best way to see who is actually the cheapest relative to how fast they are growing. |
