We are entering the slowdown, and subsequently, a 3+ year pit of despair for AI applications companies.

Capital has already decided what capabilities will be lucrative in an enterprise context. Think software engineering, legal, CX, sales, foundation models (bio, robotics, etc.) ...

A class of startups have already been chosen in 2023/2024 by tier-1 firms as winners and will continue to be piled into. Their advantage from being first movers is two fold:

• Brand

• If they have users: data

Capabilities for vertical agents in enterprise software have not and will not be ready for anyone during this 3+ year period. The only exception here is in software engineering where there will be enough data for RL to work.

Founders deciding to start an enterprise AI company now in for a rough ride. Unless the founders command capital by right and are kingmade, the only room for value creation is in non-consensus verticals. However, if you're working on something non-consensus right now, you're probably just not going to be able to raise capital. Even if you are successful in raising the capital and are working on something truly important with real upside that venture capital has not aligned itself upon yet, chances are the frontier labs are already exploring it.

If and/or when capabilities become ready, the market will be flooded with a second wave of "chosen" enterprise companies. At this point choosing to bet on this set of companies is even worse than right now because you have to compete with the first-movers wave of mandated companies and frontier labs who have now achieved superintelligence with their army of FDE's coming to eat all enterprise value. GG.

If you started 2 years ago and can't survive the next 3+ year storm, you're cooked. If you're starting now, you're cooked. If you wait for capable LLM's, you're cooked. Everyone except the frontier labs is cooked.

As for consumer companies generally, I think investing in new companies here is over forever.

1. AI is probably not the silver bullet for the "make the world smaller" people in consumer social. It is for the "no human will ever talk to another human and reproduce ever again people" though
2. If we enter a world where the cost of building software amortizes at $0, then consumer software margins will amortize at $0 as well. No money to be made here as an investor.
3. Consumer AI founders, on average, seem the most anti-VC and more likely to seed-strap and run away with your SAFE

If I was a software VC, I'd be fully tweaked. Only places worth putting money into are the frontier labs and physical infrastructure (compute + energy). I also hear that there is $700 Quintillion of rare earth metals on Asteroid 16 Psyche and that the U.S. is $36 Trillion in debt right now so maybe I'd lever up on crypto as well.

If you're a builder, it's truly the best time in the world to start a shitty AI company. Take the biggest SAFE you can get and never convert it. Who needs venture returns when you're on track to clearing $Xmm/yr in 6 months. Make the money while you can.

Or work on something actually important.