8 AI Stocks Are Crashing — I’m Buying These First

The speaker's core thesis is that today’s selloff is not a normal across-the-board crash, but a concentrated repricing of the AI and semiconductor trade. He repeatedly stresses that the damage is clustered in AI infrastructure names—Nvidia, Broadcom, Micron, AMD, Intel, ASML, Lam Research, Applied Materials, Qualcomm—while other areas like healthcare, consumer staples, and some financials are holding up. In his framing, the market is finally testing whether the earnings and cash flow of the AI leaders can justify the amount of enthusiasm, capex, and momentum that pushed the group higher for the past year. He supports that view with market breadth and geography: the selloff began in Asia, where South Korea’s Kospi fell around 10%, then spread into Europe and U.S. technology futures. …

1. This is presented as an AI/semiconductor unwind, not a market-wide crash.
2. Crowding, stretched technicals, and rising skepticism around AI capex are the key drivers.
3. Debt financing and higher-for-longer rates are becoming a bigger concern for mega-cap tech.
4. Cheaper, more competitive AI models may compress returns on the buildout.
5. He prefers selective buying over broad dip-buying.
6. Nvidia, Meta, and Microsoft are his favored risk/reward names.
7. Palantir is his most cautious pick because valuation remains demanding.
8. The broader market may be rotating, not breaking.

• The repeated use of DCF, reverse DCF, and margin-of-safety language gives a veneer of precision, but the underlying assumptions are not fully disclosed in the transcript.
• He presents the AI selloff as a healthy reset, but that view depends on earnings continuing to outrun capex; the evidence shown is more descriptive than conclusive.
• The ranking blends very different businesses together, which makes cross-name comparisons less clean than the video suggests.
• Some named price targets and valuation figures are cited quickly without full methodology, so the confidence in exact upside comparisons is limited.
• He treats lower-cost AI models as a headwind to returns, but does not quantify how much pricing power would need to compress for the thesis to weaken.