Meta Is Paying $100M Signing Bonuses. Its Own Employees Are Miserable.

Less than a year after Mark Zuckerberg declared Meta would go all-in on AI, the mood inside the company has curdled. A recent New York Times report paints a strange picture: nine-figure signing bonuses raining down on a small elite, while everyone else watches their colleagues pack up. “This isn’t the company I joined” has become a refrain.

So let’s talk about what an AI gold rush looks like from the inside — and who pays for it.

A $100 Million Bonus, and the Empty Desk Next Door

The numbers are genuinely unprecedented. To staff up Meta Superintelligence Labs (MSL), Zuckerberg’s new in-house AGI shop, the company reportedly offered signing bonuses north of $100 million to a handful of star researchers poached from OpenAI, Google DeepMind, and Anthropic. Dozens of senior hires followed.

The catch: that money didn’t trickle down. In the same window, Meta executed another round of cuts targeting the bottom 5% of performers. Picture the split-screen — NBA-tier offers landing in one inbox, severance paperwork in another, on the same Tuesday.

One employee told the Times that watching a teammate get walked out while the company handed an outside hire enough money to buy a private island created a specific kind of corporate vertigo. Call it the inequality glitch.

“Performance Culture” Is Doing a Lot of Work

Zuckerberg has been publicly building toward this for over a year. He calls it “performance culture” — a phrase that, like most corporate euphemisms, sounds reasonable until you watch it operate.

In practice, the shift looks like this:

• Bottom 5–10% are auto-flagged for cuts every cycle, regardless of team performance
• Individual stack-rankings now matter more than team-level outcomes
• Anything not feeding the AI roadmap gets quietly deprioritized

That last one is doing the most damage. Engineers in Reality Labs, ad infrastructure, or core Instagram features describe a slow recognition that their work — work that, you know, makes the money — is no longer the kind of work that gets you promoted, protected, or even noticed.

Two Companies Wearing the Same Badge

The most telling detail in the reporting is structural: Meta has effectively split into two organizations.

The first is MSL. Separate building, separate reporting line straight to Zuckerberg, separate comp band that doesn’t show up in the regular calibration. The second is everything else — the ads business that prints the money to pay for MSL’s GPU bill.

Among non-MSL employees, the Times reports a dark joke making the rounds: “We’re the ATM. The real company is over there.” One engineer put it more bluntly: the Meta badge used to mean something. Now it just opens doors.

This is leaking into recruiting too. Top-tier AI candidates, the report notes, won’t even take Meta calls unless the offer is for MSL. The gap between the inside and the outside of that lab keeps widening.

Burnout, With Nowhere to Go

The mental-health picture is grim and consistent across anonymous accounts: people feel they need to pivot to AI work or be next on the list. They feel pressure to ship weekend AI demos. Nobody is giving honest feedback in 1:1s anymore because everyone is calculating how to survive the next cut.

A late-2025 round of additional layoffs froze the ice over completely. One manager told the Times that after watching an adjacent team get vaporized, candor in 1:1s simply ended. That’s not a culture problem you fix with an offsite — that’s a trust collapse.

And leaving isn’t really an option. Big Tech hiring is still soft, senior non-AI roles are thinning out across the industry, and so the prevailing emotion is something like grim endurance. Miserable, but stuck.

Zuckerberg’s Bet, and What It Costs

From the C-suite, none of this is a bug. It’s the price of catching OpenAI, Google, and Anthropic in the AGI race: outbid everyone for talent, cauterize anything that doesn’t compound, and let the org chart rearrange itself around the new center of gravity. Wall Street, for what it’s worth, has applauded.

But the question lurking in the Times piece is the harder one: can a lab built by torching internal trust actually win? Talent that follows money tends to keep following money. The same researchers who left OpenAI for a Meta check can leave Meta for the next check. Loyalty bought is loyalty rented.

There’s also a simpler risk. If the ads business — the one funding all of this — starts to wobble because nobody is minding it, the GPU budget wobbles too. Going all-in on AI while neglecting the cash machine is a strategy with two sharp edges.

The Bottom Line

AI hype is, in the end, a story about people. Who gets the $100 million. Who gets the severance email. Who walks past an empty desk every morning and tries to focus. Meta’s situation is a useful reminder that “all-in on AI” is not a free slogan — it has a human invoice attached.

The same dynamic is almost certainly coming to other Big Tech firms as the race intensifies. We’ll know in a year or two whether Zuckerberg’s bet pays off. The interesting question is what’s getting quietly burned down to fund it.