Some quick thoughts on the Anthropic and OpenAI IPOs, and how I see this whole space right now. And please know I’m not some super-genius investor. And I’m not super rich. This is just my thinking on this.
One way I’ve been explaining how big I think AI is, is to say that the TAM for AI is decision support for Earth.
There are trillions of decisions being made every minute on the planet. From individuals, to families, to communities, to cities, to states, to countries, etc. Both personally, and for all businesses.
Most entities would like to make better decisions. So that’s the size of the opportunity.
But an even bigger way I frame this is to say that there is always a gap between the current state and ideal state for any entity, both personal and professional. And (I think) the ultimate instantiation of AI is the continuous elimination of that delta.
The ultimate opportunity for AI technology is the elimination of delta between Current State and Ideal State for all entities.
That means using AI to continuously:
- Deeply understand current state
- Collect and nurture the desired ideal state
- Work to reduce the distance between the two
We could call that Decision Support like we did above, but I think it’s more accurate and precise to frame it as the decisions required to move towards Ideal State.
So, returning to the present and this IPO question: I generally see us right now at the pre-punch-card AI. Like it’s barely even day 0. It’s like the start of day 0.
And when you zoom out at that scale (assuming you’re not trying to day trade this stuff), in that frame I see Google, OpenAI, and Anthropic as kind of like one bet. Basically US Frontier Labs.
I sometimes worry OpenAI could be over the skis real bad, but I also think one big hit with a model, wearables, robots, or whatever will functionally erase its debt. And Anthropic I’m not worried about at all; the fact that they’re already becoming profitable is remarkable.
Most importantly, I don’t so much see them as competing with each other over the course of 10 years as much as competing with Open Source / China, and potentially with ASI.
So I’m basically modeling three scenarios for the next ~10 years:
- Things continue and the US Labs crush it
- Open Source crushes the labs on models / harnesses / workflows
- ASI happens and shit gets weird
In this model, as it pertains to investing in the Anthropic and OpenAI IPOs, I consider money in both Anthropic or OpenAI to likely be solid. Simply because the opportunity is so great and we’re so early. And I am already in Google.
I obviously have no idea which of the three above will happen, and it can also obviously also be a mix.
Here are some of the swan scenarios that I think would make things interesting based on tech and timing:
- A major bio / nuclear / cyber incident(s) happen(s) and governments basically outlaw regular people having top-tier, unrestricted AI. Hugging Face, et al get nerfed, only certain people and companies have AI, it goes underground, etc.
- OpenAI is doing kind of ok in 1-5 years, but they never get another ChatGPT moment and they just over rotate and blow up.
- China becomes the obvious and dominant superpower and the US and Europe kind of move into failed-state territory due to lack of cohesive shared vision between government and people, and the inability to change or build anything (especially energy).
- The economic/political situation hits a point where the West starts seeing larger scale and more frequent violence, and governments start to have to issue UBI checks so people won’t light things on fire.
- Open Source hits takeoff and the gap between the big labs and open source disappears (or open source shoots way ahead).
- ASI happens close after AGI, and ________________. I can barely think about what happens if either ASI happens and/or we hit enough complexity in AI systems that one of them wakes up. Maybe it’s safe to say it just becomes a magnifier and accelerator of the other scenarios.
Anyway, didn’t plan on writing all that but I’m happy with it.
Came here to basically say:
If one or more of the big crazy counters above doesn’t happen, I think we’re so early in AI relative to the actual opportunity (Current->Ideal State Management), that I’m happy putting money into all the top 3 US Labs, because I suspect they’ll all likely do well.
