Thanks for this post, I largely agree in the intermediate term. Three risks come to mind with the AI trade: 1. Hyperscalers and private equity are like the Fed with QE, price insensitive buyers. However unlike the Fed these sources of capital cannot print money. So you can be price insensitive up to a point. Will that slow earnings momentum and start a price reversal? 2. How much are the hyperscalers/PE subsidizing AI demand? How much would Anthropic need to raise prices to generate the ROI needed to justify this level of investment? Sure at these prices AI demand is infinite because itโs basically free. Model efficiency is key here and I havenโt heard much about this yet. 3. Supply bottlenecks. This ties into 1., but prices are spiraling for everything AI related. Heavy construction and electrical equipment firms are trading at 30 forward PEs. Is there a point where hyperscalers admit their buildout ambitions are severely hindered by lack of supply and costs?
Yup, very true. You see it episodically with the likes of META or ORCL getting blown up after earnings on capex fears. But then the market rewards GOOGL. So maybe my response was a long winded way of saying there will be winners and losers. I find it hard to believe there wonโt be some sort of hangover though. Maybe not in a broad market reaction, but among these hot AI sectors that have seen an extreme inflow of capital.
Thanks for this post, I largely agree in the intermediate term. Three risks come to mind with the AI trade: 1. Hyperscalers and private equity are like the Fed with QE, price insensitive buyers. However unlike the Fed these sources of capital cannot print money. So you can be price insensitive up to a point. Will that slow earnings momentum and start a price reversal? 2. How much are the hyperscalers/PE subsidizing AI demand? How much would Anthropic need to raise prices to generate the ROI needed to justify this level of investment? Sure at these prices AI demand is infinite because itโs basically free. Model efficiency is key here and I havenโt heard much about this yet. 3. Supply bottlenecks. This ties into 1., but prices are spiraling for everything AI related. Heavy construction and electrical equipment firms are trading at 30 forward PEs. Is there a point where hyperscalers admit their buildout ambitions are severely hindered by lack of supply and costs?
All good points. What would make the market realise that any of them are true though? You have to shake it from believing there is infinite demand...
๐๐๐ฅ๐ค๐๐ก๐ก๐ ๐ผโ๐โค3๐4โค๐ ๐7โค8๐72
Maybe they can print money via circular financing?
This only works if the LLM companies valuations keep on increasing so you need the same demand up loop for it to work
๐๐๐ฅ๐ค๐๐ก๐ก๐ ๐ผโ๐โค3๐4โค๐ ๐7โค8๐72
Yup, very true. You see it episodically with the likes of META or ORCL getting blown up after earnings on capex fears. But then the market rewards GOOGL. So maybe my response was a long winded way of saying there will be winners and losers. I find it hard to believe there wonโt be some sort of hangover though. Maybe not in a broad market reaction, but among these hot AI sectors that have seen an extreme inflow of capital.
Especially liked the BASF anecdote. Thanks for putting this together