msdak

For the moment, despite the whole AI ecosystem being loss making for the supply chain end-to-end, it is materially juicing earnings in the listed space: *NVDA sells chips to hyperscalers, other cloud cos (Coreweave) and foundation AI model companies at high gross margins; same goes for other suppliers of chips & componentry (AVGO, memory players), fabs (TSMC), and fab equipment suppliers (ASML). *The above book immediate profits on sales, but cloud cos & other buyers like META capitalize spend rather than expense it; it impacts FCF, but earnings only with a lag as depreciation comes through. *Cloud companies sell AI compute resources to AI model cos/start ups on a marked-up basis, booking gross margin & rapid sales growth. *The cost of training and inference compute vastly exceeds the revenues these downstream AI customers are generating so they will be booking huge losses, but these losses are booked primarily in the unlisted/VC space. *These losses are being funded by VC LPs as well as via the growing number of cross-shareholding deals where chip and cloud cos are subbing equity in AI model cos. *However, no one "sees" these losses as VCs mark to cap raise valuation, which have been inflating, so VC LPs see large profits and cloud/chip cos book gains on cross-shareholding investments through other income (AMZN booked US$10bn gain in 3Q25 on its Anthropic stake). So you have a huge profit swell across a vast portion of the listed market space despite the AI ecosystem overall losing money. OpenAI's revenue run-rate atm is only US$13bn. We are going to need to see a 1-2 order of magnitude increase in end AI application monetization to make this sustainable, and relatively quickly. Will it happen? It's not impossible but quite doubtful IMO. During the dot. com bust, S&P earnings fell 40%, which was the primary driver of the c50% fall in the S&P, and a major driver of the NASDAQ's 85% decline. It wasn't just a "P" bubble, it was also an "E" bubble. While it seems like swinging good times at the moment, the whole AI boom & bull market generally is actually quite fragile atm as if AI application & monetization disappoints there is a long way down for both "E" and "P".