Meta Accelerates Superintelligence Push Amid Financial Outlays and Leadership Shifts

Meta Platforms, under the direction of Mark Zuckerberg, is rapidly advancing its strategy toward achieving artificial superintelligence, a pivot marked by significant internal restructuring and substantial financial commitment aimed at closing perceived competitive gaps. The company is dedicating tens of billions of dollars toward expanding data centers and procuring specialized Graphics Processing Units (GPUs) to meet aggressive development timelines for this next-generation artificial intelligence.

This strategic acceleration follows an internal acknowledgment of performance shortfalls after the April 2025 debut of the Llama 4 model, which reportedly did not meet expectations, signaling a shift away from its prior open-source emphasis. Zuckerberg’s strategy involves aggressive spending and recruitment to address structural weaknesses, a process that has reportedly resulted in organizational friction and departures among senior leadership. The firm projects its 2025 capital expenditure (CapEx) to be between $70 billion and $72 billion, with this investment earmarked to bring approximately 1.3 million GPUs online by year-end to form the foundation of its compute infrastructure.

To lead this new initiative, Alexandr Wang, co-founder of Scale AI, joined in June 2025 as Chief AI Officer to direct the newly established Meta Superintelligence Labs (MSL). This move followed Meta’s $14.3 billion investment for a 49% stake in Scale AI. Despite the internal focus on AI development, Meta’s core advertising business remains strong, evidenced by a 26% year-over-year revenue growth to $51.24 billion in the third quarter of 2025, driven by AI-powered ad targeting. However, this intense investment is affecting profitability metrics; GAAP net income for Q3 2025 was $2.71 billion, the lowest since Q2 2019, largely due to a one-time $15.93 billion tax charge.

Internal dynamics within MSL appear strained, with reports indicating tension over Zuckerberg’s detailed oversight of the AI strategy. This friction is contextualized by the departure of foundational AI research chief Yann LeCun at the end of the year to establish AMI Labs in Paris, focusing on 'world models' as an alternative to the Large Language Model (LLM) focus favored by the company’s leadership. Furthermore, the internal development roadmap is shifting, with the large language model 'Avocado' reportedly delayed from late 2025 to early 2026 due to training performance issues. Concurrently, the company is developing 'Mango,' an image and video generation model, slated for a first-half 2026 launch to compete with rivals. Zuckerberg’s vision centers on integrating these advanced AI tools into future smart glasses, leveraging Meta's global user base of 3.54 billion Daily Active People.