A year after spending $14.3 billion to hire Alexandr Wang and overhaul its AI strategy, Meta has a new model but an old problem: Mark Zuckerberg has to convince Wall Street and developers it was worth it.
The CEO acknowledged in an internal memo reviewed by Reuters that Meta has "made mistakes" during the workforce restructuring tied to its AI push, warning that more errors are likely.
"Given the complexity of these changes, we've made mistakes and will almost certainly make more," Zuckerberg wrote, adding that he is focused on providing as much stability as possible.
The admission came after Meta laid off roughly 8,000 employees in May (10% of its workforce) and reassigned 7,000 more to AI-focused roles. Combined with earlier cuts, the restructuring is expected to affect about 20% of Meta's 78,000-person workforce.
Zuckerberg said the company does not expect additional company-wide layoffs this year and will try to find internal roles for displaced employees. He wrote that creating new roles allowed Meta to shrink teams while keeping the option to transfer people back if needed.
The workforce turmoil is the backdrop to a bigger strategic question. Wang's team delivered the Muse Spark AI model in April, Meta's first proprietary foundation model and a departure from its open-source Llama strategy. The model was designed to plug into Meta's apps and Ray-Ban glasses rather than court third-party developers.
That pivot carries risk. Meta's Llama 4 release fell flat last year, and developers have not forgotten.
Rob May, CEO of startup Neurometric, told CNBC the AI community "largely ignores Meta at this point," adding that Muse Spark generated a "yawn" because the technology is not widely accessible. Krish Subramanian, CEO of KOI AI, said the lack of developer trust will hurt Meta if it does not refocus on third-party builders.
Wall Street is also waiting. Meta's stock is down 18% over the past 12 months, the worst performer among megacap tech alongside Microsoft, despite 33% revenue growth in the first quarter.
The company still relies on ads for 98% of revenue. Ralph Schackart, an analyst at William Blair, said investors want tangible evidence of new AI-first products from Muse Spark.
Meta's capex guidance for 2026 is $125 billion to $145 billion, much of it going to AI infrastructure. The company added $107 billion in contractual commitments for cloud and data center deals in a single quarter.
Wang has called Muse Spark an "appetizer" for larger models to come. A Meta spokesperson said the company still plans to offer developers API access this month.
But Howard Yu, a business professor at IMD, pointed out that Zuckerberg's metaverse bets have generated over $80 billion in total losses since 2020, making the AI pitch a tougher sell.
"He's running out of the space for his credibility to last," Yu said.
