Meta began companywide layoffs on May 20, cutting approximately 8,000 employees, roughly 10 percent of its 78,865-person workforce, with additional cuts already planned for the second half of 2026. The company has described the cuts as structural rather than performance-based: teams are being reorganized into AI-focused “pods” as Meta redirects spending toward an AI infrastructure buildout expected to cost between $115 billion and $135 billion this year alone.
The new structure sorts AI-facing staff into three tiers: AI Builders, individual contributors who ship models and products; AI Pod Leads, technical managers running small cross-functional teams; and AI Org Heads, senior leaders running entire AI business lines. All of it sits under Alexandr Wang, the 28-year-old founder of Scale AI whom Meta hired as Chief AI Officer in June 2025, after Mark Zuckerberg paid $14.3 billion for a 49 percent stake in Scale AI, a deal that bought Meta half of Scale and, more to the point, bought Wang. His division, Meta Superintelligence Labs, debuted its first major model, Muse Spark, earlier this year.
What the layoffs don’t fully explain is what’s happening underneath Wang’s org chart. In March 2026, Meta CTO Andrew Bosworth created a parallel Applied AI Engineering organization under VP Maher Saba, a Reality Labs veteran, reporting to Bosworth rather than to Wang. Engineers are being reshuffled into Saba’s group instead of Superintelligence Labs, a split insiders describe as deliberate rather than a temporary reorg quirk. Yann LeCun, the Turing Award winner who led Meta FAIR for a decade, left the company in late 2025 after publicly criticizing Wang as “young and inexperienced,” a departure that, in hindsight, reads less like an isolated resignation and more like the first visible crack in a leadership rift that predates this month’s layoffs.
The headline 8,000 figure also understates the actual churn. Meta has separately cancelled roughly 6,000 open requisitions, frozen hiring more broadly, and reassigned about 7,000 existing employees onto AI-focused teams. Combined, that’s tens of thousands of people whose jobs either disappeared, never materialized, or got reassigned in the space of a few months, all in service of a single strategic bet that two of the company’s most senior technical leaders are visibly not aligned on how to execute.
The May cuts are Meta’s largest single round since its 2023 “Year of Efficiency,” when Zuckerberg eliminated roughly 10,000 positions and froze hiring companywide. Counting this year’s round, Zuckerberg has now overseen the elimination of close to 33,000 positions since 2022. The framing is notably different this time: the 2023 cuts were pitched explicitly as belt-tightening during a slowdown, while May 2026’s cuts landed during a quarter of record revenue, with the savings earmarked to help fund AI capital spending rather than to survive a downturn. Reporting from TechCrunch in mid-June, drawing on internal accounts, described conditions inside Saba’s roughly 6,500-person Applied AI Engineering unit in stark terms: engineers and product managers reassigned there with effectively no real choice beyond joining or quitting, some describing their new work as generating puzzles and coding problems to train Meta’s models rather than the frontier research roles they’d originally signed up for. Employees inside the unit began calling themselves “draftees.”
The strain became public enough that both Bosworth and Zuckerberg addressed it directly and on the record. Bosworth told staff in an internal memo that leadership had done an “atrocious job explaining the vision” behind the reorganization and acknowledged it had damaged employee trust. Zuckerberg went further, conceding the changes had “caused distress” and that “we’ve made mistakes and will almost certainly make more,” while promising no further companywide layoffs for the rest of 2026 and capping how many direct reports any one manager could oversee. Coming directly from the company’s own leadership, that’s about as clear a confirmation as exists that the Wang-Bosworth tension described by outside reporting is real, not just outside speculation.
None of this is happening in a vacuum. Google lost roughly $225 billion in market value this year after Gemini 3.5 Pro’s launch was delayed for a full architectural rebuild, and OpenAI and Anthropic are locked in a funding and revenue race that’s consumed close to half of all global venture dollars in 2026. Meta’s layoffs fit the same pattern playing out across the industry: aggressive AI capital spending forcing cost discipline everywhere else in the business, whether or not the internal organization actually running that AI strategy has its own house in order.
Wall Street’s read has been more mixed than the layoff headlines might suggest. Meta stock was down roughly 8.7 percent year-to-date as of the announcement, trailing Alphabet’s 24 percent gain and Amazon’s 12 percent gain over the same stretch, even as sell-side consensus across 55 analysts still averaged a “Strong Buy” with a price target implying more than 35 percent upside. Not everyone buys the cost-savings logic behind the layoffs specifically, though: Morgan Stanley estimated the restructuring would produce an $800 million one-time charge against roughly $2 billion in savings for fiscal 2026, a rounding error next to $135 billion in AI capital spending, suggesting the cuts’ real function may be organizational discipline as much as balance-sheet relief.
Meta has no significant engineering headcount in the Philippines, so the layoffs themselves carry no direct local employment impact. The more relevant thread for Philippine founders and developers runs through Meta’s open-source Llama model family, which remains one of the most widely used free or low-cost AI backbones for local startups that can’t afford to build products on metered Claude or GPT API calls. LeCun was one of the loudest internal champions of open AI research at Meta; his departure, combined with resources visibly being pulled toward Wang’s closed, product-focused Superintelligence Labs, is a real signal worth watching for any Philippine team whose product roadmap quietly assumes Meta will keep shipping competitive open-weight models at the same pace it has for the past two years.
Whether Wang’s Superintelligence Labs or Bosworth’s Applied AI Engineering group ends up running Meta’s actual AI strategy is no longer a matter of outside speculation, both Bosworth’s own memo and Zuckerberg’s public concession confirm the tension is real and unresolved. A company spending $135 billion on AI infrastructure while its own AI leadership structure is still being fought over internally, and while the unit built to absorb that reorganization is being described by its own staff as a “gulag,” is not a stable foundation. The organizations depending on Meta’s AI roadmap, including plenty of them outside the United States, have no real visibility into which side is winning, or what a resolution would even change for the products they rely on.
Share this article