How to Pick Stocks When Everyone Is Betting on AI | Dan Ha, Karani Asset Management
- What to look for
- This provides an opportunity to distinguish actual growth in the AI industry from expectations priced into stocks, and to examine an investment approach that evaluates abstract characteristics such as corporate culture using observable data, along with the limitations of implementing that approach.
- Why it matters
- This helps investors examine prices, company-specific risk-reward profiles, and sensitivity to discount rates together, rather than relying on a single indicator such as indices, usage, or financial results, and distinguish explanations of an investment strategy from verified performance.
AI summary
The portfolio manager explains that he has adjusted his market assessment based on corrections in individual stocks and changes in risk-reward profiles that are not apparent from looking at indices alone. He emphasizes that judging the actual buildout of AI infrastructure to be at an early stage is separate from determining whether investment prices are still at an early stage, and that stock prices move ahead of the physical buildout. He explains that increased token usage does not guarantee revenue growth at the same pace, and that investors need to examine price multiplied by usage, as well as expectations already priced in. In discussing investment strategy, he introduces an approach that compares corporate cultures using measurable data such as customer and employee reviews and safety statistics, and selects outstanding companies by considering these alongside business competitiveness. He acknowledges that, unlike in theory, a strategy of short-selling companies with poor cultures is difficult to implement because of asymmetric losses and the dilution of signals through diversification. He states that interest rates affect the discount rate used in long-term valuations, but that rather than predicting their precise path, he manages sensitivity to changes in assumptions and diversification. His point is to monitor financial information related to Anthropic and expectations for AI, while managing both upside and downside risks and selecting companies within sectors so that the portfolio does not depend excessively on a single outcome. All directional outlooks are the speaker's conditional judgments, not definitive predictions of short-term prices.