# semiconductor guidance — X 热门讨论 (2026-09-15 21:32 UTC)
## @MikeLongTerm (Mike) · 09-15 18:41 · ♥39 ↻3 💬3 $AMD will outperform most stocks in the next 3 months 🧵
I’ve said for months that if we get rate hikes or financial conditions somehow tighten, capital would concentrate in the best revenue and earnings stock and that is @AMD
The Agentic AI/CPU trade (according to many trading desks) is very much alive and not even in the ramping phase. AMD has the highest allocation increase from TSMC among all semiconductor customers, and the biggest CPU ramp (EPYC Venice) starts in Q4 2026.
In short, nobody has “significant” supply until next year, and AMD has the supply advantage. Dr. Su said AMD is asking for more supply from TSMC, so if we see TSMC increase CapEx by 10–15% in the coming months and quarters, that would be even more bullish. I believe Dr. Wei will, IMO.
Price action speaks louder than words. AMD has been the most stable in term of price action among all semi stocks, because it has the best CPUs to scale Agentic to 100s-1000s of Agents like @OpenAI and @AnthropicAI discuss in various forums/podcasts.
I’m still sticking to my $800 PT by year-end instead of my old $620 PT, because in November 2026 we will have the Q3 earnings report which nobody really cares about at this point; and that is when Q4 2026 guidance will be issued. I believe the guidance will be sandbagged, and that would still shock analysts and investors :)).
I'm still sticking to my $800 PT by year end instead of my old PT $620, because in November 2026, we will have Q3 ER, which nobody really cared at this point, that is when Q4 2026 guidance to be issued. I believe, the guidance will be sandbagged, and that would still shock all analysts and investors :)).
So what are you trying to say, Mike? Capital concentration in probably fewer than 20 tickers, which could mean very expensive valuations next year until financial conditions ease. I have no idea whether financial conditions will ease next year. But that is the fun part of owning a high quality business like AMD: I get to enjoy this kind of FOMO from institutions. Months ago, AMD was given $50–$150 PTs, by the way, and declared uninvestable, or people asked, “Do we even need AMD?” AMD was also the least owned among all semiconductor stocks by active funds, according to BofA, and the whole narrative flipped in less than 12 months. And I was right. It was certainly 4+ painful years in which analysts refused to admit they were wrong on revenue and EPS growth and then issued $50–$150 PTs. I called them out nonstop.
AMD will be a much higher quality company from Q4 2026. I don’t know what kind of valuation Wall Street is going to assign it, but I do know it is going to be worth much more than it is now, with three digit growth on both the top and bottom lines. AMD is well on track to become TSMC’s #2 largest customer (currently 3rd) in terms of wafer starts and CoWoS.
We will find out if i'm right again :)).
Not Financial Advice! DYOR! I do own AMD shares! > 引用 @MikeLongTerm: BREAKING $AMD @BankofAmerica $3.2T TAM 2030🆕🚨
America raised its semiconductor TAM outlook to $3.2T by 2030, from a prior $2.7T estimate, citing memory and data center demand.
Companies BofA thinks are winner in this $3.2T TAM: $AMD $NVDA $TSM $MU $LRCX $AVGO $INTC $ARM $MRVL $AMAT $KLAC $ASML $ADI $ON $TXN
The industry took about 50 years to reach $1T in annual sales. BofA now sees the market nearly doubling from an estimated $1.7T over the next four years. It models 18% CAGR for 2026–2030, up from 14%. Memory products and data-center components are the main drivers; auto and industrial demand are expected to recover as well
Capex / equipment (also raised): ~2026 wafer fab equipment (WFE): $156B (from $144B), about +33% YoY ~2027 WFE: $210B (from $190B) ~2030 WFE: as high as $360B ~DRAM equipment: about $61B in 2026 and $85B in 2027
This is a further upgrade from earlier 2026 BofA work that had the 2030 industry closer to $2.0–2.7T, with AI data center systems, HBM/memory tightness, WFE, analog power, and agentic CPU demand as the core themes. Forecasts in this space have been moving up quickly as hyperscaler capex and HBM attach stay stronger than prior models. https://x.com/MikeLongTerm/status/2099931532806730166