Meta Platforms Inc. (META) is tearing up its AI playbook. After spending a year championing open-source models as the future, CEO Mark Zuckerberg has executed a sharp philosophical reversal that's reshaping the company's technology strategy, leadership structure, and spending priorities.
The clearest signal? Llama, Meta's open-source AI model that Zuckerberg touted relentlessly, barely got a mention during the company's October earnings call. That silence speaks volumes about how the company views its AI position after Llama 4's disappointing performance and the rise of formidable Chinese competitors like DeepSeek.
The Avocado Pivot
Meta is now building a new frontier model codenamed Avocado, according to a CNBC report Tuesday. Originally scheduled for late 2025, the launch has slipped to early 2026 due to training performance challenges. More importantly, Avocado may debut as a proprietary system rather than an open-source release, marking a complete about-face from Meta's previous strategy.
MarketDash reached out to Meta investor relations for comment and is awaiting a response.
This isn't just a technology shift. It's triggering a cultural earthquake inside Meta that's upending long-established power structures and working norms.
The 28-Year-Old Running Meta's AI Future
Zuckerberg removed Chris Cox, Meta's longtime Chief Product Officer, from the AI division and replaced him with Alexandr Wang, the 28-year-old founder of Scale AI. Wang now serves as chief AI officer, working alongside former GitHub CEO Nat Friedman to lead elite teams that operate with startup-level secrecy and speed.
Meta paid $14.3 billion to acquire Wang's team, a massive bet that Wall Street is watching closely. The new leadership has imposed a "demo, don't memo" culture that leans heavily on modern AI agents and demands 70-hour workweeks. That's a jarring departure from Meta's historically open internal culture, and it's creating friction as the company simultaneously conducts layoffs and restructuring.
The tension reached a breaking point when Yann LeCun, Meta's Chief AI Scientist and a towering figure in the field, left to launch his own startup. That's the kind of departure that signals just how disruptive these changes have been.
The $70 Billion Question
Meta has raised its 2025 capital expenditure outlook to more than $70 billion as it races to catch up in AI. That's an enormous sum, and investors are debating whether the company can generate adequate returns on such massive spending.
To help fund this AI push, Meta is making hard choices elsewhere. The company is slashing up to 30% of its 2026 metaverse budget, delaying its Phoenix mixed-reality glasses to early 2027, and shifting resources toward AI wearables and data center infrastructure.
Hardware Realignment and Strategic Acquisitions
Meta recently acquired Limitless, a startup specializing in AI-powered wearable technology, to accelerate what Zuckerberg calls "personal superintelligence." The company is ending commercial sales of the Limitless Pendant while supporting existing users for at least a year, eliminating subscription fees and providing free access to premium features. Meta is also retiring non-core apps like Rewind and adding tools that help users manage or delete their data.
On the infrastructure side, Meta is diversifying beyond its own data centers through cloud partnerships with CoreWeave Inc. (CRWV) and Oracle Corp. (ORCL). At the same time, the company is building its massive Hyperion facility in Louisiana to support next-generation AI workloads.
Analysts at JPMorgan and Bank of America Securities believe the tighter spending discipline and strategic reprioritization could free up billions in cash flow and support earnings growth as Meta doubles down on AI-driven products and infrastructure.
Market Reaction
Meta shares are up 12% year-to-date as investors weigh the tension between soaring capital expenditures and the company's long-term AI ambitions. On Tuesday at publication, Meta shares were down 1.16% at $659.08, according to market data.
The big question now is whether Zuckerberg's aggressive bet on proprietary AI, combined with the leadership overhaul and massive infrastructure spending, can deliver the breakthrough Meta needs to compete with the industry's AI leaders. The delays, departures, and dramatic strategy shifts suggest the company is still figuring that out.