Meta just walked away from open-source AI. That was its whole identity. Its whole moat story. Its whole "we're the good guys" narrative in a world where OpenAI went corporate and Anthropic went paranoid.

Gone. Llama 4 came out in April 2025 and fell flat. Hard. And now, barely a year later, Mark Zuckerberg has pivoted so hard he just handed $14.3 billion to Alexandr Wang — the Scale AI founder — to build proprietary models at "Meta Superintelligence Labs."

Their first product? Muse Spark, launched last month. It's not open. It's closed source. It's the exact opposite of everything Meta preached for two years.

The Open-Source King Is Dead

Remember 2024? Meta was the champion. Llama 3 was the model everyone built on. Developers loved it. The narrative was: Meta gives away the best AI, OpenAI charges you for it, and Meta wins on ecosystem.

That narrative is dead.

"The AI community largely ignores Meta at this point." — Rob May, CEO of Neurometric

Let that sink in. The company that gave away state-of-the-art models — engineers at Meta reportedly prefer Anthropic's Claude over their own internal tools. That's not a strategy. That's a retreat.

The Numbers Don't Lie (But They Confuse)

Here's what's weird about $META right now. On paper, the business is firing. Revenue is up 33% year-over-year to $215 billion trailing. That's a $1.43 trillion company proving it can still grow like a startup. Profit margins at 32.8%. EPS of $27.49.

And yet the stock is down 13.3% year-to-date at $562.20. That's not a value play. That's a market telling you something's broken.

The P/E is a trailing 20.47 — cheap on paper. Forward P/E of 15.53. PEG ratio of 0.81, which screams "undervalued" if you believe the growth story. But nobody believes the growth story anymore. Because the growth is all ads — 98% of revenue — and AI monetization is a question mark.

The "Fix" Is Expensive and Unproven

So what's Meta doing this week? Three launches in 21 days.

Business Agents (June 3) — AI that automates ad buying and customer service for small businesses. Sounds good. Sounds like Salesforce did five years ago.

$299 Ray-Ban AI glasses (June 23) — Hardware play. Meta's been here before. Remember Portal? Remember Oculus rebrand? Hardware is hard.

Arena prediction markets (June 24, today) — Because Polymarket wasn't enough, now Meta wants in on prediction markets. This feels like a feature, not a business.

It's a lot of throwing stuff at the wall. And it's all happening while Big Tech collectively just watched $2.7 trillion get wiped from Mag 7 valuations in June alone. The AI spending backlash is real. Investors are asking: when does this money come back?

What Meta Bought for $14.3 Billion

Let's talk about Alexandr Wang. He built Scale AI from nothing into the backbone of military and enterprise data labeling. He's 27. He's brilliant. And Meta just paid more for him than most companies' entire market caps.

But here's the thing: Wang's magic was data infrastructure, not foundation models. Building proprietary AI at Meta's scale requires a different playbook. A playbook Anthropic and Google already have.

Meta is playing catch-up. And catch-up, when you've just abandoned your core narrative, is expensive.

The Bottom Line

$META trades at 15.5x forward earnings because the market thinks it's an ad company. If the AI thesis works, that's wrong and it's cheap. If the AI thesis is a $14.3B distraction from a mature ads business, it's fair.

I'm not buying this pivot. Not yet. Open-source was their one structural advantage. They threw it away to compete on the same playing field as OpenAI, Anthropic, and Google — and they're starting behind on all of them.

The stock's down for a reason. The market sees what Zuckerberg won't admit: Meta's AI strategy is a work in progress with no moat, and the open-source king is dead.

Disclosure: This article is for informational and educational purposes only and does not constitute investment advice or a solicitation to buy or sell any securities. The author may hold positions in securities discussed. All financial data sourced from Yahoo Finance as of June 24, 2026. Sources: CNBC, Yahoo Finance. Do your own research.

— The Signal