I first heard about the “frontier of knowledge” from Paul Graham and I have heard it repeatedly from many other since: Naval, Elon Musk…

The frontier of knowledge is the edge of what humanity currently understands.

By exploring at the frontier of knowledge you discover problems and solutions before many others. You can apply new solutions to old problems and see if anything previously impossible has now become possible. Or you can use old solutions to solve a new problem people don’t know will exist in the future.

You are in a market of low/no competition.

Moore’s Law - your customers will be price anchored on older, less efficient technology.

It goes without saying that risk and reward go hand in hand, exploring at the frontier of knowledge is free but building at the frontier of knowledge has costs with low likelihood of success. There is almost always asymmetric upside as most humans are loss averse.

There is no single frontier of knowledge, there are many. You can look anywhere to find them. One thing I must add is to learn how much is already known before you dive deep into creating something that already exists.

Some brutal thoughts:

• The extent to which a frontier is valuable is down to the future demand, for example SpaceX will have no shortage of companies and countries wanting to colonise Mars first.
• The value of your role on the frontier may not be remunerated if your contributions are academic and without a sufficient business model for you to get paid.
• You hope your discovery on the frontier is defensible and you build an irreversible moat before the rest of the world catches up. Example is ChatGPT having a good first mover advantage on consumer LLMs. API’s got built using OpenAI’s early models but since the barrier to swap to gemini is low it’s not that sticky.