Meta Offering $100 Million Salaries to Top OpenAI Researchers Is Peak Silicon Valley Obscenity

This is not exaggeration

Everyone knows the AI industry moves money. Few know how much.

Let’s start with the facts: Meta is losing. CEO Zuck’s latest moves haven’t gone well—like the Llama 4 Maverick PR disaster (training different models to trick the audience is petty) and the forced delay of Llama 4 Behemoth—and now he’s taking desperate steps.

Recently, Meta announced (yet another) restructuring of its AI teams. It’s no longer just two (Yann LeCun’s FAIR and the generative AI branch, whatever their names are now), but three: there’s a new group tasked solely with reaching the mythical superintelligence.

Zuck must’ve realized that chasing artificial general intelligence (AGI) isn’t enough anymore. Not with OpenAI CEO Sam Altman publishing essays like The Intelligence Age and The Gentle Singularity, where he casually alludes to superintelligence as if AGI were a thing of the past. Not with Anthropic CEO Dario Amodei talking about genius-populated countries made of datacenters. Not with SSI CEO Ilya Sutskever naming his company Safe Superintelligence; no commercial products, no strategic roadmap, no wild investment return promises for shareholders.

However, it’s hard to make sensible structural changes if you have no clue what you need them for. That’s why Meta hired a spy. The most expensive spy in history: 28-year-old Alexandr Wang. Meta got Wang and 49% equity on his company, Scale AI, for $14.3 billion. But, who’s Wang? The guy who founded and led Scale AI, the crown jewel of the data-acquisition-for-AI business (which means underpaying people in the Global South to label data that he later sells to AI companies to train AI models).

$14.3 billion is an insane amount of money. And in my view—given that the internet is now exhausted as a data source and AI companies in the know are shifting to synthetic data—it feels like a legal steal on Meta pulled off by Wang. Spy and thief, two in one (it’s a joke, don’t sue me).

So Zuck was falling behind in both results and ambition, but now he’s shown the world that he has more power over his empire than anyone (he’s in full founder mode to destroy the internal structure as he pleases) and more cash than anyone (to make clearly reasonable-priced buyouts). Sorry to say, Zuck—and at this point I’m doing free consulting (you can contact my AI assistant for the invoice)—but neither your power nor your cash is going to save you.

Buying Scale AI is an obvious desperate shot in the dark. (It reminds me of that one time Microsoft bought Mustafa Suleyman in a similar operation, thinking he knew anything about AI.) That’s not going to save you. And neither will a team with the buzzword “superintelligence” in the label. We all know it’s more of a pose than a principled attempt to build whatever you think an actual superintelligence is.

Indeed, Zuck has no principles when it comes to AI. He moves wherever the wind blows. As I wrote in an earlier piece explaining why Altman has the character trait to win that Zuck lacks (it’s focus):

Each day, he decides based on what his rivals are doing, what the public wants, what investors demand, what the President promises, and what his engineers swear by. . . . Despite the relentlessness and the grit, Altman has nothing against Fickle Zuck. And nothing is exactly what he needs. Because if one thing can beat extreme agency toward anywhere, it’s extreme agency toward somewhere: A firm belief beats a careless empire.

“Extreme agency toward anywhere” is, maybe, the best four-word summary of Meta’s last decade (ahem, Metaverse, ahem, $46 billion). Turns out, Zuck’s latest extreme-agency move—Altman himself confessed it in a podcast to his younger brother—is offering $100 million salaries to top OpenAI researchers.

I could go on about how relentless Zuck is. Or how desperate. But I’d rather stop and just emphasize how obscene that number is. It is, in one sentence, the peak of Silicon Valley’s moral and strategic collapse.

Here's the moral part (all are informed estimations): it would take the average writer or artist in the US 2,000 years to make $100 million. So about 25 deals with the devil. A hundred million is enough to pay the annual salary of 1,500 public school teachers. Or to cover the full tuition of 10,000 college students at a state university. It could fund every librarian in New York City for two years. Or provide clean drinking water to a million people for a decade. It’s more than what all the firefighters in San Francisco earn in a year, combined. It could keep a rural hospital running for 20 years. Or endow a public research lab indefinitely. You could give every student in Detroit a new laptop. Every ten years. For life. Or double the arts budget of a small country.

In the hands of Meta, it’s a reckless bid. In the hands of the state, it would be called nation-building. But, of course, no one likes the state in Silicon Valley.

Are top AI researchers worth that much? I guess the market decides. Just like with top actors, singers, and athletes. But Zuck isn’t paying market price. That’s the strategic part. He’s trying to fix a costly mistake—like spending $60 billion on Nvidia GPUs—by bribing the people who might still save him and his declining empire.

An empire erected painstakingly over two long decades, during which Zuck didn’t hesitate for a moment to hijack the psychological vulnerabilities of consumers by taking “as much of [our] time and conscious attention as possible.” Not my words—Sean Parker’s, founding president of Facebook. And now he plans to burn it all in a race he’s already lost because he needs to integrate state-of-the-art AI models into his social media platforms to keep us hooked—to keep the ad money flowing. Gotcha.

The funniest part—if you’re up for a joke after that—is that the best OpenAI researchers have, so far, turned down the offer. I guess they’re not doing this for the money. They’re doing it ‘cause they love it.

---

GENTLE REMINDER: The current Three-Year Birthday Offer gets you a yearly subscription at 20% off forever, and runs from May 30th to July 1st. Lock in your annual subscription now for $80/year. Starting July 1st, The Algorithmic Bridge will move to $120/year. Existing paid subs, including those of you who redeem this offer, will retain their rates indefinitely. If you’ve been thinking about upgrading, now is the time.