Skip to content
A $3B Attempt to Build AI With Understanding episode artwork

Dec 22, 2025 7 min

A $3B Attempt to Build AI With Understanding

from Hard Fork AI

Understanding is the missing ingredient. World models aim to supply it. We explore intelligence. Get the top 40+ AI Models for $20 at AI Box: ⁠⁠https://aibox.ai AI Chat YouTube Channel: https://www.youtube.com/@JaedenSchafer Join my AI Hustle Community: https://www.skool.com/aihustle

Episode details come from the podcast's RSS feed.

Embed this episode

Ready to play

A $3B Attempt to Build AI With Understanding

0:00 7:29
of matches

Transcript Auto-generated, may contain errors

Yann LeCun has confirmed his new quote-unquote world model startup is looking for a $5 billion valuation as he is launching his own AI company. He was famous for being the head AI scientist over at Meta and is now going off on his own. He announced this officially, but I think it's been basically Silicon Valley's worst-kept secret. Everyone knows this is what he's been focusing on, and he has made it clear that he is not going to be serving as the company's CEO.

We're going to get into all of the drama, what the company does, who's running it, how much you can expect to make, how much other players in similar positions and situations have made, all of that on the show today. But before we do, if you want to try all of the AI models I talk about on the show, including everything from OpenAI, Meta, Grok, and so many more, there's over 40 different models on AIbox.ai. I'd love for you to try it out. It's my own startup.

I give you access for $20 a month to everything, including 11 labs for audio and a ton of amazing image models. Go check it out. I'll leave a link in the description to AIbox.ai. So let's get into what Yann LeCun is working on.

Of course, he is a scientist who has had a lot of public drama on X. I think like a year ago, he left the platform when Elon Musk took it over. He had a lot of drama with Elon, and sadly, I felt like I didn't really see much from him after that happened. He kind of, I think, got shuffled down at Meta.

Zuckerberg was probably upset that Meta was falling behind in the AI race. The head of Meta AI, the CEO of Meta AI, was acquired from another company, and it felt like Yann LeCun kind of lost some of his shine over there. But no fear, he is a renowned scientist, and he decided to start his own company, and that is called Advanced Machine Intelligence, or AMI. And essentially, went and got Alex LeBrun, who's the co-founder and CEO of the Medical Transcription AI, which is called NABLA.

He is going to be leading the company as the chief executive, and NABLA said that LeCun's new role, they essentially, like, there's a press release about it, and LeCun, Yann LeCun, also confirmed in a LinkedIn post, this is what's happening. Of course, he can't post on X because, you know, he's got drama on X. He hops back on every once in a while. He said, yes, AMI Labs is my new startup.

I am the executive chairman, and Alex LeBrun is transitioning from CEO of NABLA to CEO of AMI Labs. AMI Labs is looking, apparently, to 500 million euros, that's about 586 million dollars, at a $3 billion valuation, or about $3.5 billion before they've even launched. This is according to Financial Times, who cited some people are familiar with all of this. Even how aggressively investors are backing AI startups right now, which were founded by the kind of globally recognized researchers, I think this is not that shocking by today's standards.

Like, for comparison, we can look at the former OpenAI CTO, which is Miriam Moratti. She started Thinking Machine Labs. It was valued at $12 billion in its seed round last year, and Moratti also doesn't have the same kind of level of academic and industry credibility that a lot of people give to Yan. So I think, you know, like Lacan, he's a professor at New York University.

He's a former VP and chief AI scientist at Meta. He's also a recipient of the prestigious AM Turing Award for his work on reinforcement learning. So I think recently there was a whole press release about it. It kind of confirmed what a lot of people in the industry already assumed.

AMI Labs is focused on world model AI. So essentially, it's just a different approach from large language models. They're attempting to give the AI systems an understanding of their entire environment so they can simulate cause and effect, and they can explore what-if scenarios to predict outcomes. A lot of proponents believe that world models are the answer to the structural hallucination problems that a lot of these LLMs have because LLMs, they're not, you know, just make up information.

There's a bunch of other top labs and startups like Google DeepMind and Fifi Lee's startup world labs are also building these world models. And I think, like, if we're comparing it to that, AMI's fundraising ambitions do look pretty bold. Like, these are pretty big. When world labs launched, Lee raised $230 million or $1 million valuation, which was considered, you know, like, insane at the time, or not insane, but it was, you know, quite impressive.

And that was back in August of last year, which in AI terms might well have been, you know, 100 years ago. Meanwhile, I think also NABLA said it's going to begin searching for a new CEO that is going to run it in the interim, or sorry, in the interim that will be run by the co-founder and COO Delphine Grawl, who's not yet named the permanent chief executive. So that company also announced a partnership to use AMI's models as they are developed. So it's kind of interesting because they, you know, he's leaving the company, but they have a great relationship or they're going to have a great relationship with AMI.

So NABLA has raised $120 million in total. There's, you know, they did like a $70 million series in June from a bunch of high profile backers. They have Lacan himself who invested in the company. You had Tony Fadell's Build Collective, HCI Capital, Highland Europe, and Kathy Innovation.

So I think these companies are definitely intertwined. I think Lacan, as an investor in that company, you know, obviously had a great relationship with the CEO and was able to pull him in. So LeBrun, who is going to be the new CEO of AMI. He is a really strong pick, I think, to leave the company.

He's been working on a bunch of different multimodal AI models since before. This was very popular. He started at Nuance Communications back in, like, the early 2010s, back when Nuance powered Apple Siri, and then back then the assistant was still considered this really cutting edge technology. Today's Siri is a piece of garbage, a flaming piece of garbage.

But back then, this was quite amazing stuff. Microsoft later acquired Nuance, and LeBrun went on to found and sell multiple natural language startups, including one that was acquired by Facebook before he was running a Facebook AI division, and then he later founded NABLA back in 2018. So LeBrun said that NABLA is, you know, a favorite within the Paris AI startup scene. It is growing a lot.

He said, we have more than tripled our live ARR this year, up next $1 billion. He said that as he was telling everyone that he was going to be leaving as the CEO. So he's like, look, we did great on the company. We grew a lot, but it is time for me to go.

So it's going to be fascinating to see what happens with this. Thank you so much for tuning into the podcast today. If you enjoyed it, make sure to leave a rating and review wherever you get your podcasts. We are getting close to the Christmas season as the number one Christmas present you could give me.

I would be absolutely thrilled. Thanks so much for tuning in, and I hope you have a fantastic rest of your day. As always, make sure to check out AIbox.ai to get access to over 40 AM models all in one place. All right, I'll see you in the next episode.

Frequently asked questions

How long is this episode of Hard Fork AI?

This episode is 7 minutes long.

When was this Hard Fork AI episode published?

This episode was published on December 22, 2025.

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this Hard Fork AI episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.