AI Models Will Tap the Whole Economy in 2 Years | Aidan Smith artwork

AI Models Will Tap the Whole Economy in 2 Years | Aidan Smith

MTS

October 22, 2026

Aidan Smith, co-founder of Flapping Airplanes, discusses raising $1 billion to build data-efficient AI models by redesigning GPU primitives and model architecture from scratch.

0:00

Aidan Smith

We want to unlock this 90% of the economy, which is untouched right now. That's so much more demand for compute, just waiting to be tapped. I mean, if you think that we're seeing a lot of demand for compute now, imagine what it's gonna look like in two years. Even if we're able to run 1000X more tokens, imagine the demand for compute when you're actually able to tap the whole economy. That's huge.

0:16

SPEAKER_2

Welcome to MTS.

0:17

Aidan Smith

Thank you so much, it's great to be here.

0:18

SPEAKER_2

Great to have you.

0:19

SPEAKER_3

Yeah, this is booked, what, like two hours ago? MTS Speed.

0:23

Aidan Smith

MTS Speed wanted you on the day the announcement came. Better than the airlines, baby, better than the airlines.

0:26

SPEAKER_2

Way better than the airlines. Didn't even charge you up for it. Wow, wow, and you can do two at once. Where did this logo come from? What's the origin story behind the flapping and the airline?

0:35

Aidan Smith

So, the more formal answer is that we really would like to create models that are a bit more like birds than planes, a little bit less rigid and stilted, and a little bit more like natural intelligence, where, I mean, we frankly have been very inspired by the brain and my prior experience in Neuralink and our other experience out in the field. And we basically thought to ourselves, yeah, I mean, why aren't models at the same level of data efficiency as humans? There's this really baffling thing where if you look at some really, really percussion metric like perplexity or something, and you say, okay, how many tokens have you seen Theo? How many tokens has GPT-2 seen? It might actually be kind of similar. And you might even measure similarly on this really bland pre-training eval, and then on a normal post-training eval, it's ridiculous, not even close. The model will just spew word salad and garbage non-stop. And the question is, why? Why are you able to be so much better at such limited data? And so I live a little bit more like birds than planes. The slightly less formal answer is that it's really funny, and that we love the idea that these more traditional financial types would see these kids going out raising a billion dollars or whatever, and pick up the newspaper, read, Flapping Airplanes Raises a Billion Dollars, and say, I don't understand this AI business.

One very famous investor called us Flappy Bird, and thought that we were the app that had pivoted to AI.

I'm here to clear the schedule. This is not the case.

1:54

SPEAKER_3

That's really funny.

1:55

SPEAKER_2

Man, that was crazy.

1:55

SPEAKER_3

Flappy Bird is dead now. It's so sad. I miss it.

1:58

SPEAKER_2

No, they're not. They pivoted to AI.

2:00

Aidan Smith

Exactly.

They're A-Web very well, I hear. OK.

2:04

SPEAKER_3

Wow. So why did you choose the route of data efficiency over any other strategy that a new lab could have taken?

2:11

Aidan Smith

Yeah. I think for us, the biggest reason to grow up data efficiency is that we look at the broader economy, and we look at the parts of the economy that AI can touch today. And it's a tiny fraction of the overall value. And there's this weird paradox where the models are so much smarter than you and I, and also completely inept.

They really can't do anything actually in the real world. And most of the value is frankly stuff in the real world. It's these places where you actually have to be very, very data efficient to learn a new task or to adapt, or to even pick up this knowledge that's germane to a given domain. And so we thought that by investigating data efficient methods, we could go and tackle the other $30 trillion as other people focus on code and what have you, we can go and focus on the long haul. And we really think that there's actually a big advantage here for a Neo Lab, because most Neo Labs are trying to compete on the exact same factors as our folks. This is great, but I really respect our peers and all this, but I don't think that you're as likely to be successful there, versus when you put yourself in a place where you have to innovate, and you can't be on the knife side like everyone else, doing the same thing, the same methods with the same people, you can actually start from scratch.

3:07

SPEAKER_2

Yeah. I mean, two things I'm particularly curious about, I think that what you guys came out with is the primitives for GPUs. You guys can actually have different types of algorithms be really efficient there, and specifically with your background at Neuralink. I mean, it kind of hints to what's coming, and I know you guys didn't particularly release a model quite yet, but when you do, does anything from your background at Neuralink kind of shape what this is going towards?

Read the other 28 minutes of this episode here.

$15 gets you this episode and 99 more. Credits never expire. After checkout, one link brings you back here to read it.

Not what you expected? Full refund within 14 days.

Thousands of transcripts fetched by people building searchable podcast archives

Fetch transcripts with code

The same credits work from a script or an AI agent. The demo key returns a sample episode in full, no card needed:

request
curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.

Using your own key:

request
curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000797599187