Key Takeaways
- Meta gives Muse users 100 million free tokens a week, while OpenAI and Anthropic rate-limit heavy free users.
- New AI models get matched within weeks, so the lasting edge is paying for free usage, and Meta’s advertising business can pay for it.
- The main risk is Meta’s AI infrastructure spending, which looks to be more than $100 billion this year and which could squeeze the cash that funds the giveaway.
- Watch Google, the one rival with the same ad-funded advantage and no Muse-style agent yet.
Meta Platforms (META) is doing something with Muse that OpenAI and Anthropic can’t afford to do:
Giving it away.
Muse, Meta’s three-week-old AI agent, climbed to #1 among free apps in the app stores earlier this month. Writer and investor Jack Raines, author of Young Money, explained how generous that free plan is on this week’s episode of The Prof G Pod:
“Mark Zuckerberg is giving away 100 million tokens per week, which is a ton of compute.”
Anyone who uses Claude and ChatGPT knows that you get rate-limited pretty quickly on the free plans (and, let’s face it, even on the paid plans too).
It comes down to economics: Meta can subsidize the demand because of its enormous ad business, while OpenAI and Anthropic cannot.
The edge that lasts
Raines’s deeper point is about how quickly a model’s lead disappears. OpenAI ships a model that breaks benchmarks, Anthropic tops it, “and then you’ll have a Chinese model that’s 95% as good for 5% of the price.”
“So whatever edge that these companies get with their new model seems to get fully commoditized within just a few weeks of launch.”
If the smartest model only stays on top for a few weeks, the advantage that lasts is being able to keep serving users without cutting them off. Raines puts it in user terms:
“If you’re using Muse and it works nonstop for you all week and you’re using ChatGPT and you keep getting rate limited, sooner or later, you’re just going to choose the AI tool that doesn’t break.”
I think he’s right. Most people won’t know which model tops the leaderboard this week. They will absolutely notice the day their agent stops answering.
Can Meta really afford it?
Start with what the free plan is worth. Meta charges $20 a month for the Power plan’s 500 million tokens a week. The free tier’s 100 million is a fifth of that allowance, so it comes to about $4 a month at Power-plan prices. (That’s what Meta charges for the tokens. What they actually cost Meta to serve isn’t public, so I’m just using it as a proxy.)
Raines says Muse had 3.4 million downloads in its first three weeks. If every one of those users burned the full free allowance, that’s about $13.6 million a month at list price.
Meta also did ~$59 billion in advertising revenue last quarter, plus another $1.5 billion from the rest of the business.
That works out to roughly $20 billion of ad revenue a month, so even a maxed-out free tier for every early download is a rounding error: well under a tenth of a percent.
And that ad engine throws off a lot of cash…
Meta’s operating cash doubled from $57.7 billion in 2021 to $115.8 billion in 2025. Even with capex ramping, there’s plenty of room.
Isn’t the capex eating the cash?
The obvious objection is that someone has to build the compute behind all those free tokens, and Meta is building a lot of it. That is, of course, the same cash the giveaway runs on.
But by the same token (see what I did there?), OpenAI and Anthropic are both unprofitable and raising cash to try and stay afloat.
So in a tight year, Meta’s ad business still pays for the servers. In a tight year for the labs, they have to find new investors, and Raines thinks the private markets are “basically tapped out.”
Investors seem comfortable with the trade for now. Just look at Meta’s price chart over the last three weeks!
The other ad giant
Meta has one competitor that can match this kind of spending: Alphabet (GOOGL). Alphabet has that same kind of cash engine…
Alphabet has produced more than $60 billion of free cash flow in every year on the chart, about $342.5 billion in total from 2021 to 2025. The only dip came in 2022, to $60.0 billion, and it was back up to $69.5 billion a year later.
Growth has flattened lately, though: 2025’s $73.3 billion was only about $500 million more than 2024’s $72.8 billion. That’s still a lot of cash to throw at a Muse rival, if Google decides to build one.
Google already has your Gmail, your calendar, and Android. Raines finds it “actually shocking” that Google didn’t beat Muse to the punch. That gives Meta a head start on the one rival that can afford to give away just as much.
Who can pay for free the longest?
Meta can. Its ad business pays for the giveaway, and OpenAI and Anthropic have to raise the money for theirs. If model edges fade within weeks, the company that never rations its users picks up the ones who get tired of hitting limits. I’d bet on Meta holding that position against the labs.
Of course, Muse is three weeks old, and downloads are a long way from daily habits. The capex bill will keep growing, and Google could still show up with an agent of its own.
So what is Meta stock actually worth?
TIKR lets you forecast the future price of any stock in less than a minute. Just enter a few assumptions into TIKR’s valuation model and see what Meta could be worth. Start from Wall Street consensus estimates, or adjust the inputs to reflect your own view of the business. It’s free to use.
Disclaimer:
Please note that the articles on TIKR are not intended to serve as investment or financial advice from TIKR or our content team, nor are they recommendations to buy or sell any stocks. We create our content based on TIKR Terminal’s investment data and analysts’ estimates. Our analysis might not include recent company news or important updates. TIKR has no position in any of the stocks mentioned. Thank you for reading, and happy investing!
