Meta's Strategic AI Restructuring Coincides With Tougher Performance Standards
In the glass-and-steel corridors of Meta's Menlo Park headquarters, a significant transformation is underway. The tech giant has split its artificial intelligence operations into two specialized divisions while simultaneously raising performance expectations for its workforce—dual moves that signal an aggressive posture in the increasingly competitive AI landscape.
Specialized AI Units Emerge From Restructuring
Meta's AI division has been reconfigured into two distinct teams with separate but complementary mandates. The AI Products team, now under Connor Hayes's leadership, will focus exclusively on integrating AI capabilities across Meta's consumer platforms—Facebook, Instagram, WhatsApp, and the standalone Meta AI application.
"This isn't just another reorganization," remarked a senior AI strategist familiar with Meta's operations. "It's a recognition that consumer-facing AI and foundational research require different approaches, different timelines, and different metrics for success."
The second unit—the AGI Foundations team—will be jointly led by Ahmad Al-Dahle and Amir Frenkel. This group's mission centers on advancing Meta's core AI technologies, including its flagship Llama models and developing more sophisticated reasoning, voice, and multimedia capabilities.
Summary Assessment of Amir Frenkel's Background and Relevance to Meta's Generative AI Challenges
| Category | Details |
|---|---|
| Current Role | VP, Engineering & Product, XR Tech at Reality Labs (Meta); previously Director of Engineering, Head of Computer Vision at Oculus |
| Experience at Meta | Joined in 2016; promoted through roles in XR and computer vision; deep familiarity with Meta’s technical culture |
| Previous Roles | - Google: Head of Wearables Engineering - Amazon Lab126: Director of Software Development (2012–2015) - HP/Palm, TI: Platform Dev Leadership |
| AI Credentials | Holds a Coursera Machine Learning certificate (2013); no advanced specialization in LLMs or generative AI |
| Strengths | - Proven leadership in complex tech initiatives - Strong background in AR/VR and computer vision - Experienced in scaling teams and products |
| Limitations | - Not specialized in generative AI or LLMs - Unlikely to directly solve issues with Llama 4 model |
| Meta's AI Crisis | Llama 4 delayed; underwhelming performance; 11 of 14 original researchers departed; internal frustration and leadership reshuffling under consideration |
| Overall Assessment | Frenkel could help with organizational or cross-functional tech strategy but lacks the domain-specific expertise needed to resolve Meta’s LLM challenges |
Meanwhile, the company's Fundamental Artificial Intelligence Research unit continues its work on longer-horizon research under Yann LeCun, Meta's Chief AI Scientist. One multimedia team previously under FAIR has been integrated into the AGI Foundations structure, suggesting a move toward more practical applications of theoretical research.
Strategic Timing Amid Competitive Pressures
The restructuring arrives at a critical juncture for Meta. Industry insiders note that Llama 4, the company's latest AI model, has faced delays and underperformed against internal benchmarks. This disappointment comes as competitors like OpenAI, Google, and ByteDance continue to gain momentum in both consumer and enterprise AI markets.
"Meta is essentially running two races simultaneously," explained a technology investment analyst who tracks major AI players. "They need to push innovative AI features to users quickly to maintain platform engagement, while simultaneously developing next-generation foundation models that can compete with or surpass GPT-5 and Google's Gemini."
By creating dedicated teams for each race, Meta appears to be seeking greater development velocity and clearer accountability structures. The separation may also allow the company to better prioritize resources between immediate product needs and longer-term technological advancement.
Performance Review Changes Target "Underperformers"
In parallel with its AI restructuring, Meta has implemented more stringent performance evaluation criteria. According to internal documents, managers of teams with 150 or more employees will now need to classify 15% to 20% of their staff as "below expectations" during mid-year reviews—up from the previous 12% to 15% range.
The timing is notable. These reviews begin June 16, with performance conversations scheduled throughout July and August. An internal memo distributed on May 14 through Meta's company forum explicitly framed the process as "an opportunity to make exit decisions," though it specified there would be "no company-wide performance terminations, unlike earlier this year."
This reference acknowledges the approximately 4,000 employees—roughly 5% of Meta's workforce—who were laid off in early 2025, many for performance-related reasons. CEO Mark Zuckerberg had previously signaled his intention to "raise the bar on performance management and move out low-performers faster."
The Human Cost of Strategic Realignment
For Meta employees, these dual announcements create a climate of both opportunity and uncertainty. AI specialists may find new pathways for advancement within the restructured divisions, but the heightened performance expectations cast a shadow across the organization.
"The message is clear—deliver exceptional results or prepare to leave," observed a workforce management consultant who has worked with several major tech companies. "Meta is creating an environment where only those who contribute directly to its AI ambitions will thrive."
The expanded "below expectations" category includes what Meta terms "nonregrettable attrition"—employees deemed noncritical to operations who have already departed. This approach represents a continuation of stricter performance management practices that began in late 2022.
Industry-Wide Recalibration
Meta's moves reflect broader trends across the technology sector. As companies intensify their AI investments, they're simultaneously reassessing their workforce composition and performance standards.
"We're seeing a fundamental shift in how tech companies view their human capital," said a veteran Silicon Valley executive recruiter. "The AI era demands different skills, different mindsets, and different organizational structures. Companies are reconfiguring accordingly, and that inevitably creates winners and losers."
For Meta, the restructuring and performance review changes represent two sides of the same strategic coin—focusing organizational energy and talent on winning what many consider the most consequential technological race of the decade.
As one venture capitalist specializing in AI investments put it: "In the current environment, tech companies must be both visionary and ruthlessly efficient. Meta is clearly trying to be both."
The coming months will reveal whether these organizational changes deliver the accelerated AI development and improved competitive positioning that Meta seeks—and at what human cost these ambitions are pursued.