1/ June 2025. Zuckerberg writes a check for $14.3 billion for 49% of Scale AI.
But the money was never really for the company.
It was for the man who built it. Alexandr Wang. 28 years old. MIT dropout.
Meta's greatest risk. And its greatest profit ๐งต
1
1
49
2/ The risk:
Wang isn't a research scientist. He built a data-labeling business.
Zuckerberg bet Meta's AI turnaround on a business leader, not a researcher.
Yann LeCun ran Meta's AI research for a decade. Turing Award winner.
1
1
29
On his way out, he called Wang "young and inexperienced."
Five months later, he raised over $1B for his own startup.
1
1
12
3/ The collateral damage:
600 people cut first, mostly from the research group Wang inherited.
Then 8,000 more across the company.
Meta wasn't struggling. Revenue up 22%. Profit strong.
This wasn't survival. It was a bet. And everything that didn't serve it got cleared out.
1
1
20
4/ The profit:
April 2026. Meta Superintelligence Labs releases Muse Spark.
First model from Wang's team. Powers Meta AI across Instagram, WhatsApp, Facebook.
Meta says it uses 10x less compute than Llama 4 Maverick.
1
1
23
5/ The benchmarks:
Muse Spark ranks 4th on Artificial Analysis Intelligence Index.
Behind Gemini 3.1 Pro and GPT-5.4. Roughly tied with Claude Opus 4.6.
Not a blowout. But competitive.
Shares rose 9% after announcement.
1
1
36
6/ Why efficiency matters more than benchmarks:
Meta's ad impressions increased 12% last year.
Price per ad increased 9%.
Two numbers that historically move in opposite directions.
Better models = better ad targeting = more revenue.
1
1
13
7/ The real lesson:
None of this was about skill.
LeCun wasn't cut because he stopped being good. He was displaced because someone new redrew the map.
Your value at work was never fixed. It's positional.
And positions move when someone new walks in with more leverage.
1
1
18
8/ Meta bet $14.3 billion on one man's judgment.
So far, it's paying off.
But 8,000 people learned what that bet costs.
#Meta #AI #AlexandrWang
1
19