Q&A with Mark Zuckerberg and Alexandr Wang on the Muse app, Muse Wonder, the risks of committing $100B+ to AI, rebooting Meta's AI team, agentic commerce, more
a truly tender moment 🥹Amit /@amitisinvesting:Zucks: “So at the end of last year, we got quite excited about the progress but hadn't seen the models yet, so it's like we were going to write this check for, I don't know, $100B or $200B and I'm like Alex...you sure this is gonna work?! We are living in unprecedented times.Alex Greenland /@ajrgd:@collision ... I want to know if Mark signed the cheque for $200 billion while having a cheeky pint with AlexAlexander Doria /@dorialexander:Final blogpost
Cheeky Pint
Context & Ripple Effects
Meta had already put Muse at the center of a contentious product push: reporting days earlier described its launch despite internal safety concerns, while a separate interview recorded Zuckerberg’s claim that Meta had reached the AI frontier. Muse Wonder gives that strategy a named model layer for the company’s agent product.
The interview makes the capital-allocation side explicit. Zuckerberg describes weighing a $100 billion to $200 billion commitment before seeing the models, while Meta plans to offer a Wonder-powered Muse agent to subscribers; the payoff therefore depends on turning model capability into a product people will use and pay for.
First-order effects
Meta ties Muse’s initial subscriber offering to Wonder, making the model’s performance directly relevant to the agent product’s commercial appeal.
The disclosed scale of the contemplated investment puts greater pressure on Zuckerberg and Wang’s rebooted AI organization to translate frontier-model progress into an operating product rather than a research claim.
Second-order effects
Meta’s bid to pair a frontier model with a subscriber agent raises the competitive bar for rival AI providers pursuing consumer and professional agent use cases.
Safety concerns surrounding the Muse launch decision become more consequential as Meta moves the agent toward subscriber access, because product adoption and trust become linked to the same model investment.
Third-order effects
If Meta can convert large frontier-model spending into recurring agent subscriptions, public AI labs will be judged less on model announcements alone and more on whether costly compute programs support durable product revenue.
The story belongs to a capital-intensive AI race in which execution risk shifts from training a competitive model to deploying it safely and monetizing it at scale.
The trend: Frontier AI competition is moving toward vertically integrated bets that combine massive model investment, agent distribution, and subscription revenue.
We're back with what I believe may be the first joint Mark-Alex sit-down convo, discussing their new frontier model Muse Wonder on the day it meets the world. https://x.com/...
Alexandr Wang reveals Meta CFO Susan Li pulled him aside “many times” at an exec offsite in Hawaii to warn that if the model didn't work, it would be very, very bad in financial terms ZUCKERBERG:
Zucks: “So at the end of last year, we got quite excited about the progress but hadn't seen the models yet, so it's like we were going to write this check for, I don't know, $100B or $200B and I'm like Alex...you sure this is gonna work?! We are living in unprecedented times.
Cheeky Pint really pulls nuggets out of guests. Zuck had 1.5 pints of Guiness during the podcast, which is notable considering this exchange he had with Theo Von.
assigning @muse to search for uniquely long-tail products and handling the last mile(s) of navigating a maze of web forms to complete the purchase feels like a consumer friendly and exciting future.
In @colossusmag profile of Mark Zuckerberg @finkd ( a must-read by @colossusjeremy), the last quote is: “You can't take this away from him. He wouldn't be him if he stopped.” And somewhere else: “This is how it is for him; there is no stopping”
3 billionaires chatting over beers. People say way more interesting things when they forget they're on a podcast... it feels so real (even if it's a Muse ad). We need more of this!
Mark and Alexandr talk with John re: @muse and agentic commerce, amongst many other AI topics. (We work hourly—often minutely—with the @meta team to push the frontier of use cases. They keep us on our toes!)
If I were young and single I'd try to get Meta to sponsor me for a year to just let Muse plan my life - book flights, hotels, reservations, activities, whatever, and you have to do whatever Muse says, and film the whole thing.
I'm not a huge fan of Guinness (basically liquified pumpernickel), so i get it. But I'd like to think one of the best parts of being worth a quarter trillion dollars is not having to drink stuff you don't like to be accepted by the bros.
Ok, this was a goldmine of insights re: the future of commerce with agents, straight from @finkd + @alexandr_wang 24 key takeaways: 1/ Zuckerberg says agentic commerce is not a 20% conversion lift. It actually creates customers who never existed -> He used Instacart ~2x a year on…
Cheeky Pint is back! Mark Zuckerberg @finkd and Alexandr Wang @alexandr_wang came to the pub to discuss their new frontier model, Muse Wonder. 00:00:24 Muse's breakout use cases 00:07:28 The model is the product 00:14:35 Muse Wonder 00:24:56 How frontier models are built 00:48:26…
The longer version of this clip is hilarious because of the Muse use cases they brag about. Mullet guy talks about sourcing “obscure teas” from random farms in Asia. Zuck talks about feeding Muse footage from the cameras he installed in his home MMA gym. Very relatable.
The face you make on your first Guinness is involuntary. The stuff is roughly 7x more bitter than Bud Light, and bitterness happens to be the one taste humans evolved specifically to spit out. Here's what's going on. Guinness gets its color from barley roasted at around 230°C, wh…
Underdiscussed a CEO who needs to go on a contrived tech podcast in a bar is such a bear signal. Subservient to the narrative, the podcaster in the power position. Much more powerful when the podcaster has to come to the company instead
alexandr wang looks like the product of a generations-long effort by some far-future bene gesserit clan to re-derive the modal attendee of a selective high school in sydney in the year 2018 from first principles.