The Flattening, In Reverse
Meta is now asking individual contributors inside Applied AI, the roughly 7,000-person division built earlier this year to turn AI research into shipped product, whether they would like to become managers again. Fortune and Forbes both confirm the offer is voluntary, framed internally as part of a routine reorganization rather than a reversal. But several of the employees being asked already held manager titles before Applied AI moved them into individual-contributor roles, which makes this less a routine reorg and more an undo button on a decision the company made in the same division inside the same year.
Applied AI was staffed by pulling roughly 7,000 people out of Meta's existing structure and flattening the reporting lines around them, the same flattening logic Mark Zuckerberg applied company-wide during the 2023 'Year of Efficiency,' when managers and directors were told to become individual contributors or leave. Reintroducing managers into the newest, most AI-native part of the company is Meta quietly admitting that the logic did not survive contact with the work.
The Numbers Behind the Reversal
| Date | Event | People affected |
|---|---|---|
| Early 2026 | Applied AI division formed, structure flattened | ~7,000 reassigned |
| May 2026 | Workforce cuts, management layers disproportionately hit | ~8,000 cut, 6,000 open roles scrapped |
| September 2026 | Voluntary return-to-manager offer inside Applied AI | Subset of the ~7,000 |
Read in sequence, the numbers describe one continuous experiment rather than three separate decisions. Meta flattened a division, cut deep into what remained four months later with managers as the primary target, and is now paying some of the same people to come back into the roles it just eliminated. No public figure exists yet for how many are taking the offer, and Meta has not stated whether the reversal is company-wide policy or confined to Applied AI.
Why the Location of This U-Turn Matters
The significant detail is not that Meta reversed a management decision. Companies reorganize constantly, and reversals rarely make news on their own. What makes this one worth an EU or UK operator's attention is that it happened inside the division built specifically to prove AI could compress the need for coordination overhead, the exact argument several vendors and consultancies are currently using to pitch flatter structures to companies standing up their own AI or product teams. If the division Meta assembled to be the flattest, most AI-native part of the company needed its managers back within months, that argument has just failed its highest-stakes internal test, not a hypothetical one.
Meta has not said publicly that AI failed to replace management coordination. It does not need to. Asking former managers to come back is the admission, whatever the internal messaging calls it.
What an EU or UK Company Should Take From This
Any organization currently being pitched a 'flatten management, let AI coordinate the rest' restructuring should ask the vendor making that pitch a direct question: has your own product team tried this, and for how long before management came back? Meta's timeline gives a rough answer on its own turf, roughly eight months from flattening Applied AI to quietly rebuilding its manager layer. That is not proof the model can never work, but it is a real data point from the company with the deepest AI investment and the clearest incentive to make flat AI-native teams succeed.
The practical takeaway for a European operator evaluating its own structure is to treat 'AI reduces the need for managers' as a claim to test in a small team over a bounded period, with a defined checkpoint, rather than a company-wide restructuring to adopt on a vendor's word.
Read next: Meta Paired an AI Manifesto With a Data Center Pitch | Meta Taught Its Agent to Ask Before It Acts
