Meta's 1.4 billion dollar Muse overhaul

SPEAKER_1 (0:00)

Thinking about refreshing the carpet in your home? Now's the time to do it. For a limited time at The Home Depot, get 10% off installed carpet projects on trusted brands like LifeProof, LifeProof with Pet Proof Technology, Home Decorators Collection and Traffic Master. Plus, with installation starting at just 49 cents per square foot, upgrading your space is more affordable than ever at The Home Depot. Offer valid September 24th, 2026 through October 4th, 2026 Exclusions apply for licenses. See homedepot.com/licensenumbers.

SPEAKER_2 (0:30)

College football is back, so Hilton called on me, the superstition concierge, to make your fan rituals a reality. Need a room to match your lucky number? We got you. Want to make sure our team doesn't wash your lucky jersey? Oh, that smells lucky.

Hilton's unmatched hospitality can keep up with any superstition. Even a marching band wake up call at 555 and 55 seconds.

SPEAKER_3 (0:51)

Hit it!

SPEAKER_2 (0:53)

When you need a team that will do whatever it takes on game day, it matters where you stay. Hilton, for this day.

SPEAKER_3 (1:30)

Meta spent $1.4 billion to acquire scale AI founder Alexander Wang. They just brought him in, gave him the title of Chief AI Officer, and basically handed him a single objective. He had to execute this nine-month overhaul of their entire artificial intelligence infrastructure.

SPEAKER_4 (1:46)

Right, which internally they were calling Project Avocado.

SPEAKER_3 (1:49)

Exactly, Project Avocado.

SPEAKER_4 (1:51)

It's just, you know, spending over a billion dollars on what is essentially an acquihire is highly unusual, even in this industry.

SPEAKER_3 (1:58)

Yeah, it's a massive premium.

SPEAKER_4 (2:00)

Right. When you see corporate restructuring with that kind of price tag, companies are typically buying a large active user base or getting a completely proprietary hardware platform.

SPEAKER_3 (2:11)

Sure.

SPEAKER_4 (2:12)

Dropping that amount of capital to bring in one single mind and his, you know, his immediate orbit of engineers just to rebuild your internal stack from scratch, it indicates severe organizational panic.

SPEAKER_3 (2:23)

Yeah, they clearly realized something about their trajectory was broken.

SPEAKER_4 (2:26)

Fundamentally broken. Yeah.

SPEAKER_3 (2:28)

So the result of that nine month sprint was the total abandonment of their public parameter stacking strategy.

SPEAKER_4 (2:35)

The llama approach.

SPEAKER_3 (2:36)

Right. The entire approach they use for the llama models. The whole industry basically accepted that as their core identity, and it was scrapped entirely.

SPEAKER_4 (2:44)

Which is wild.

SPEAKER_3 (2:45)

Yeah. They replaced it with a closed source, native reagentic ecosystem called Muse.

SPEAKER_4 (2:50)

They spent years building all this incredible open source goodwill with llama.

SPEAKER_3 (2:54)

They really did.

SPEAKER_4 (2:55)

They positioned themselves purposefully as the open counterweight to proprietary models. They rallied an entire developer community around that ethos.

SPEAKER_3 (3:04)

And then just tossed it.

SPEAKER_4 (3:06)

Yeah. Tossing all of that away requires a stark realization that your previous technical trajectory is a dead end.

SPEAKER_3 (3:12)

We also saw them form Meta super intelligence labs specifically to house this new direction.

SPEAKER_4 (3:17)

Right. Walling off the exact type of research they used to make public.

SPEAKER_3 (3:21)

So if you alienate the developer community you cultivated over all those years, you have to ask what exact capability you are gaining that justifies a $1.4 billion architectural reset.

SPEAKER_4 (3:32)

We need to keep that structural question in mind as we look at what Muse actually does.

SPEAKER_3 (3:37)

Because it's not a small pivot.

SPEAKER_4 (3:38)

No. A pivot with this much financial weight behind it is never just about getting a slightly higher score on a standard coding test.

SPEAKER_3 (3:45)

Right.

SPEAKER_4 (3:46)

Hitting a wall with the old architecture forces you to build a fundamentally different engine, just to keep moving forward.

SPEAKER_3 (3:52)

The specific technical bottleneck that triggered all this was the failure of parameter stacking.

SPEAKER_4 (3:57)

Right.

SPEAKER_3 (3:57)

For years, the industry operated under this assumption that you just scale up the raw parameter count. You throw exponentially more compute at the pre-training phase and the model naturally gets smarter across the board.

SPEAKER_4 (4:07)

That was the playbook.

SPEAKER_3 (4:08)

Yeah. But Meta hit a hard ceiling with that. Scaling raw parameters was no longer yielding broad intelligence. Right. So they built Muse Spark to match their previous flagship reasoning capabilities while using an order of magnitude less pre-training compute.

SPEAKER_4 (4:22)

You see this exact dynamic in physical engineering.

SPEAKER_3 (4:25)

Like with hardware?

SPEAKER_4 (4:26)

Yeah. Well, think about automotive design. When a vehicle is too heavy, the initial solution is just to drop in a bigger engine and burn more fuel to get it moving.

SPEAKER_3 (4:34)

Sure. Just brute force it.