Meta told staff on Wednesday that roughly 600 people working in its sprawling A.I. operation will be let go — a move that reads like a punctuation mark at the end of a three-year sprint of hiring, splurging and reorganizing as the company tries to catch up in the furious competition around large language models and “superintelligence.”

Meta’s cuts hit legacy research teams, product A.I. squads and parts of the company’s A.I. infrastructure, while the company’s newest, high-priority unit — the small, elite group known internally as TBD that’s building next-generation LLMs and “superintelligence” — is untouched and still hiring. Management framed the layoffs as an efficiency move: fewer people, fewer meetings, and bigger scopes for those who remain.

What happened (in the memo)

The headliner was an internal memo from Meta’s chief A.I. officer, Alexandr Wang, who told employees the goal was to produce smaller, more “load-bearing” teams so decisions move faster and individuals have more impact. That memo — first reported publicly by Axios and confirmed to outlets including TechCrunch — spelled out that cuts would touch FAIR (Meta’s long-running Fundamental A.I. Research group), product teams and infrastructure, while noting TBD would stay intact.

Wang’s line landed like a Silicon Valley cliché and a corporate strategy at once: “By reducing the size of our team, fewer conversations will be required to make a decision, and each person will be more load-bearing and have more scope and impact.” That sentence is now the rhetorical center of this reorg.

Why this matters to Meta — and to the broader A.I. race

Meta hasn’t been idle in A.I. investment. Earlier this year, the company poured $14.3 billion into Scale AI and recruited Scale’s founder to help lead the new push — a dramatic wager meant to accelerate Meta’s ability to train next-generation models and to bring in leadership that could move faster. That headline deal, plus a wave of recruitment from competitor labs, shows how much Meta has bet the company on catching up.

Still, hiring more senior people and piling teams on top of teams creates friction. Multiple accounts of the past 18 months describe overlap, duplicated efforts and a culture that became top-heavy after a hiring spree — the very bloat the company now says it wants to trim. The result: a public show of muscular spending and a private admission that the machinery isn’t humming the way executives hoped.

Who’s affected — and who’s not

Meta’s announcement makes a clear distinction: the cuts hit legacy and supporting groups; TBD, the small unit directly charged with developing the next generation of LLMs and the so-called superintelligence work, is exempt and remains a hiring priority. Meta says it will try to place affected employees into other internal roles where possible.

Executives insist this is not a retreat from A.I. Rather, it’s a surgical reallocation of resources to teams that management believes can move faster and ship products. Whether that faith in “small, talent-dense teams” pays off is the question investors and rivals will be watching.

A new front: WhatsApp, chatbots and platform control

The layoffs arrived amid another aggressive move by Meta: the company also announced that it will cut off non-Meta chatbots (including OpenAI’s ChatGPT) on WhatsApp starting next year, citing misuse of WhatsApp’s business messaging features. OpenAI pushed back on that characterization; Kevin Weil, OpenAI’s vice president of science, posted that it’s “hard to believe Meta is shutting off 1-800-CHATGPT,” and reminded users they can migrate conversations to OpenAI’s own apps and web tools. The episode underscores that Meta’s A.I. fight isn’t just about models and data centers — it’s also about distribution, control over platforms and the rules of who can build what on top of messaging systems used by billions.

Internal friction, morale and the optics of rapid hiring then trimming

The story that’s been playing out in Silicon Valley this year is familiar: dramatic hiring sprees to keep up with competitors, followed by fast reorganizations when product timelines slip or internal coordination breaks down. The optics of recruiting expensive external talent — sometimes with headline pay packages — and then cutting broad swathes of an older internal bench can fracture morale and create recruiting headaches just when Meta most needs the trust of engineers and researchers.

Several outlets reporting on the layoffs mentioned that many of the highest-profile hires (the A-team that came in with Wang) are insulated from these cuts, while bench, infrastructure and some FAIR staff face the brunt. That can, in practice, hollow out institutional knowledge even as the company piles on star hires.

For people impacted, the announcement is immediate and jarring: inboxes with layoff notices, HR processes to navigate, and the uncertainty of a job market that’s still competitive at the top but unpredictable elsewhere. Meta says it will attempt internal placements, but for employees who’ve spent years building research infrastructure or product primitives, the transition is still a major life event.

Meta’s cut of roughly 600 A.I. roles is less an admission of defeat than a blunt management play: reorganize, consolidate, protect the elite unit, and try to shorten decision loops. It’s a reminder that building the next generation of A.I. is not just a technical problem — it’s also deeply organizational. And in this race, where muscle and distribution matter as much as model weights, the winners will be the companies that balance talent, product focus, platform control and the patience to iterate.