The very best research and engineering yields foundational results that create new research directions and whole new markets for startups. Because of my technical expertise, I can see which technologies are foundational, and will drive super-cycles of innovation.

These technologies produce tremendous amounts of value, some of which is captured by the startup and some of which is passed onto the consumer. Because there is tremendous value generated, there is significant opportunity to capture value. The exact mechanism by which value is captured is less important than in traditional businesses, which operate on relatively small margins. Therefore, it’s important to have a high-level understanding of the business model, but less emphasis on detailed financial analysis.

An example of this type of company is OpenAI. They created a new kind of technology by scaling LLMs up with a tremendous amount of compute, data, and parameters. The ChatGPT Plus product sells for $20/month, which is comparable to the cost of a cell phone plan. Can you imagine living without a cell phone plan? When ChatGPT went down, Meta extended their performance review deadline, because many employees used ChatGPT to help write their reviews.

This business model was actually stumbled upon by accident. The tremendous usage of ChatGPT caused the service to crash for all users, so we released a paid option to segment the user base, not primarily to monetize. Now that the value to users is clear, there are many avenues for monetization.

This analysis method only works for truly step-function technological improvements. More traditioanl business analysis is needed for incremental technological gains. Incremental advancements are valuable as well; compounding marginal improvements powers most of the economy.