Shyon
09-15
For me, today’s selloff looks more like a repricing than a sign that the AI cycle is over. If frontier model development slows, I think AI spending could simply shift from training the next massive model toward inference and deploying existing models at scale.

I am especially watching AI agents and inference demand. As companies like Microsoft, Google, Amazon and Meta integrate AI deeper into everyday workflows, the demand for GPUs, HBM, networking and data-center power could remain strong. In some ways, broader inference adoption could create an even wider market than frontier training.

That said, I would not ignore valuation risk. If cloud companies start cutting capex while GPU utilization, HBM orders and networking demand weaken together, that would be a much more serious warning. For now, I remain cautiously bullish and see this as a potential rotation within the AI trade rather than the end of it.

@Tiger_comments @TigerStars @TigerClub

Jensen Huang Drops a Number — Why Did AI Hardware Stage a Full Comeback?
Jensen Huang said Nvidia will ship twice as many chips next year as this year, and that AI safety matters but cannot be regulated the way social media was. AI hardware ran: AMD +6.36% to $545.09, Marvell +4.81% to $240.76, Nvidia +2.54% to $219.34, Broadcom +2.29% to $347.30, with the Philadelphia Semiconductor Index up over 3%. That is a third straight up session for chips, a rebound that started in the same week the slowdown argument was loudest. A week of talk finally has a number attached. But the number is the company's own forecast, not signed orders. Is one line enough to hold a rally?
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

  • TigerStars
    09-16 17:01
    TigerStars
    Great perspective on the AI selloff as a potential rotation rather than the end of the cycle 📊 The link between frontier training, inference demand and AI agents is especially clear. It could be even more actionable if you highlighted one or two indicators investors should watch to validate the shift toward inference.
    • Shyon: 
      I would watch GPU utilization and inference workloads, along with HBM orders and networking demand. If inference keeps growing while these indicators remain strong, I would see it as a rotation in AI spending rather than the cycle ending. I will also watch cloud capex for signs of a broader slowdown
  • wobee
    09-15
    wobee
    Inference only stays big if the economics work. Lower ASP chips and better efficiency can cap GPU and HBM upside fast lol
    • Shyon: 
      Thanks for leaving your valuable insights here
Leave a comment
4
3