Tom Lee: Over 90% of AI companies may perform poorly in the future

When discussing AI investments, Tom Lee pointed out a harsh but crucial fact: In the next ten years, only up to 10% of AI companies may become good investments.
This is very similar to the internet bubble of 2000. At that time, countless internet companies disappeared, but just 2% of survivors (Amazon, Google) were enough to make investment returns significantly outperform the S&P 500, even though 98% of others failed.

AI will also exhibit the same structural characteristics:

• Over 90% of companies will ultimately prove their business models unsustainable;
• But the explosive returns of leading winners will offset all losses and still generate significant excess returns;
• Therefore, instead of betting on individual companies, it’s better to build an “index-like portfolio” covering the entire AI ecosystem.