Nvidia looks interesting again because rising earnings estimates have not been reflected in its valuation. FY28 EPS estimates rose about 25% in 60 days while shares stayed roughly flat; it trades below 14 times FY28 earnings with a mid-single-digit free-cash-flow yield. Exponential change is hard to grasp, and it is hard to see Nvidia not winning if the US does, though there is little capital to absorb market wobbles. x.com  x.com 

Nvidia looks due for a catch-up trade, with the stock near three-year lows relative to AMD. A mid-to-high-teens P/E also seems hard to ignore. x.com 

AMD's stock has outrun its fundamentals, leaving it a bad investment at today's price despite an extremely bullish long-term outlook for the business. From April 2025 to September 2026, AMD rose 8x while revenue grew 55%; Nvidia rose 2.4x while revenue grew 118%. AMD had previously lagged despite improving fundamentals, but the market's catch-up went too far. Its risk/reward is no longer what it was at roughly 20x forward earnings in 2025; a crash could make it attractive again. x.com  x.com 

Nvidia's valuation gives too little credit to its expected long-term earnings growth. Its forward P/E-to-growth ratio is 0.28, the lowest since late 2023; the bullish read depends on those growth expectations proving accurate. x.com  x.com 

Long Nvidia against short AMD could work because investors should always own the best of breed, as CrowdStrike returned 304% after its July 2024 outage while SentinelOne returned 0%. Nvidia is the clear best of breed in semiconductors yet trades at a multiple far below its peer's, with 2-3x the growth and a far stronger bottom line. x.com  x.com 

AMD is priced as if it is winning the ecosystem war when it is fighting uphill, which makes it unattractive. Its chips now compete, but Nvidia built CUDA and the tooling everyone runs on, and switching means rebuilding that foundation, which almost nobody wants to do. AMD's realistic opportunity is newer data centers without that foundation, which is what has been happening. Yet nearly every sell-side firm rates it a buy at about 50x forward earnings. x.com 

AMD and Nvidia both stand to gain from an AI chip market that AMD sizes at $1.4 trillion by 2030, growing 45% a year. Token use rose 158x in two years, and running models now takes 60% of compute, driven by agents, a shift AMD expects to be structural. AMD's Helios racks are in full production, Anthropic is locking in up to 2 GW and OpenAI up to 6 GW, and AMD says it is closing its software gap. Nvidia keeps an edge because CUDA is still better and omnipresent. x.com  x.com  x.com  x.com  x.com  … Show more

NVIDIA's projected earnings growth could support substantial upside without an enormous multiple. At $221.97 it trades at 10.4x estimated FY2029 earnings; even at $350, nearly 58% higher, it would trade near 16.4x. Estimates imply roughly 130% cumulative EPS growth over the next two fiscal years and may be conservative. SpaceX and Meta earnings look bullish for future chip demand. Slower AI spending, margin pressure or competition could undermine the estimates. x.com  x.com 

AMD may be the biggest asymmetry in semiconductors: a more than 2.5x valuation premium to NVIDIA for half the growth and no moat is hard to justify. When the market turns against AMD and AI momentum funds, the drop will come all at once and get nasty. x.com 

Nvidia looks compelling and attractively priced, with CUDA as a big moat. Early in AI's progress the upside is incomprehensible, so a $5 trillion valuation is no longer large, and law-of-large-numbers limits may only bite once AI fully proliferates. Replies to the author note it trades near 17x while growing 75% a year. x.com  x.com  x.com  x.com