Meta’s newly formed Superintelligence Labs team has produced its first internally developed AI models just six months after the lab’s inception, according to CTO Andrew Bosworth. Speaking at the World Economic Forum in Davos, Bosworth confirmed to Reuters that these initial models, which encompass both text and multimodal (image/video) capabilities, have demonstrated performance “very good” enough for internal use. This milestone follows Meta’s announcements of two flagship projects—Avocado, a text model slated for a first-quarter rollout, and Mango, focused on images and video—and underscores the company’s pivot to building proprietary AI infrastructure rather than relying solely on third-party solutions.

Meta Platforms’ shares have fallen roughly 8.5% so far in 2026 and about 20% since its October earnings report, driven largely by investor concern over aggressive capital outlays. Management raised full-year 2025 capex guidance to $70–72 billion and forecast over $100 billion in 2026 as it scales data-center build-outs for AI compute. Despite this, core fundamentals remain robust: third-quarter revenue grew 26% year-over-year, led by AI-driven ad ranking improvements; operating income rose 18% to $20.5 billion; and free cash flow amounted to nearly $11 billion. CEO Mark Zuckerberg reported the annual revenue run rate from end-to-end AI ad tools has surpassed $60 billion, while the Meta AI assistant now boasts over one billion monthly active users. Quarterly revenue guidance of $56–59 billion for Q4 2025 further supports the bull case that the pullback may represent a buying opportunity for long-term investors.

Nicola Mendelsohn, head of Meta’s global business group, cautioned at Davos that Europe’s stringent AI regulations are making it more difficult for companies to deploy new AI products and services. She noted that compliance requirements under the upcoming EU AI Act—covering risk assessments, data-governance protocols and transparency mandates—are lengthening development cycles and increasing costs. According to Mendelsohn, business leaders have expressed growing frustration as these rules delay the roll-out of AI-driven features across Meta’s advertising, content moderation and consumer-facing tools, potentially affecting revenue growth in key European markets if implementation timelines continue to slip.