Subscribe now

The large and prolific unwind of Situational Awareness’ public securities book offers opportunities in stocks the fund owned as well in stocks they didn’t.

After the severe sell-off in semiconductors, there are stocks that are pricing in a lower scale of AI CapEx than I believe to be likely over the next 24 months.

Risks include the long-end of the Treasury market which, surging, senses that Fed Chair Warsh doesn’t have the stomach to tighten rates despite his triple urging that this week’s Fed meeting was “not inertial.” I agree with the bond market’s view here. If “we” are right, the yield curve probably steepens. Perhaps I should have puts on bonds but I don’t, yet. It’s moved so fast already.

Another risk is that CapEx underdelivers, I have a hedge I like for this that I’ll include below in the model portfolio for Farley Associated Research, which is effective as of market close July 31, 2026. But first, some notes on select securities and industries:

“High Quality Fins” are among the easiest longs here. Most trade at a discount to historical valuations because of perceived threats. In my view the stablecoin threat to Visa’s credit card franchise is non-existent. MORN and VRSK have lots of proprietary data they own that customers can’t get from AI. MCO and SPGI (alongside MORN) will see large credit issuance that will make their ratings businesses a lot of money (even if the mix shift is lower fee because it is the investment grade hyperscalers doing the issuing). The AI threat to the software businesses of MCO/SPGI/FDS is real… there is Model Context Protocol (MCP) integration (linking to AI, basically) that FDS appears to be a leader in but AI does pose a threat. Nevertheless I actually like the “software factor” embedded in these stocks because they have a low correlation with semiconductors.

Software

If you think AI is fake you should be very long software: CRM and CSU need you now! NOW is included in the model portfolio, as is TOST, but it could use some more additions. I need to do more work on software stocks before I have the confidence I do about other stocks and themes.

Semiconductors

The volume of spending to cure the memory bottleneck in Korea will flow directly to Lam Research and ASML. The level of growth they will experience more than justifies the forward P/Es of ~30 (I think the forward earnings estimates are too low by the way).

I prefer to be a “bottleneck supplier bro” than a “bottleneck bro.” Memory pricing is tough: it could decline on a Chinese increase in capacity. Or it could ramp further and go so high it kills the AI trade. I think owning the suppliers to memory producers, rather than the producers themselves, is the better investment. There are also five stocks with very strong IP that I have dubbed “AI Royalties” which range from the well known-names like ARM to non-obvious names I like very much. Oh, and yeah, Nvidia is cheap as chips.

My Investing Philosophy

Fundamental investors don’t do enough macro and some macro investors don’t know much about single stocks. There is opportunity in the middle.

Trades that are explicitly macro should be done only when they are obvious ie the risk/reward is heavily skewed. Otherwise, it is merely in the background, everywhere and at all times.

I believe in concentration, both in single stocks and in thematics. There don’t need to be 13 High quality financial stocks, $V and $SPGI would in a pinch suffice. Likewise, for many investors, “NVDA and LRCX,” or being long a basket via an ETF such as SMH, may be more suitable than my various picks.

I will make the investment ideas very clear and if it is non-actionable i.e. I’m just writing poetry about the Federal Reserve, I will make that clear too. (It will be good poetry, I promise).

Warning

The model portfolio employs leverage and uses strategies that I have found effective. Leverage is inherently dangerous, as the past few days have shown. Investors who run an unlevered portfolio should feel free to look past my shorts. Even with a VIX at 10, cash is the cheapest option.

Read more