Investors traditionally view Meta Platforms as a social media company generating billions in ad revenue from its 3.5 billion daily users across Facebook, Instagram, WhatsApp and Messenger. However, management has redirected focus toward artificial intelligence, building custom AI chips, expanding data centers and developing the open-source Llama language model. Meta’s forecasted capital expenditures for 2025 total $70 billion to $72 billion, up sharply from $39 billion in 2024, and the company anticipates even higher capex in 2026 to support AI infrastructure. CFO Susan Li has indicated that year-over-year capex growth will be significantly larger in 2026, underscoring the firm’s conviction that AI will drive next-generation user engagement, enhanced ad targeting and new revenue streams over the long term.

Meta’s text-based app Threads has rapidly gained traction on mobile devices, reaching 141.5 million daily active users on iOS and Android as of January 7, surpassing the 125 million mobile users of competitor X. While X still leads in web-based visits at around 150 million per day, Threads’ mobile growth reflects sustained user adoption since its 2023 launch. This trend supports Meta’s strategy to diversify engagement beyond its flagship platforms and provides new inventory for advertisers eager to leverage text-based formats. The platform’s growth is attributed to long-term network effects rather than recent controversies at rival services, indicating that Threads may continue to erode X’s mobile audience over the coming quarters.

At the core of Meta’s AI ambitions is Meta Compute, a newly established division charged with centralizing data center design, supply-chain partnerships and strategic capacity planning. Co-led by Santosh Janardhan and Daniel Gross, with guidance from President Dina Powell McCormick, Meta Compute aims to deploy tens of gigawatts of computing capacity this decade and scale toward hundreds of gigawatts over time—energy levels on par with small countries. This initiative positions Meta to catch up with AI leaders by owning more of its compute stack, ensuring reliable energy procurement and reducing dependence on external cloud providers. The combination of internal chip design, purpose-built data centers and aggressive capacity build-out under Meta Compute is expected to lower operating costs per AI token and create a durable competitive advantage in both consumer and enterprise AI markets.