Meta’s financial performance provides the muscle for its AI push. With nearly 3.5 billion daily active users across Facebook, Instagram, WhatsApp, and Messenger, the company’s ad business generated $46.6 billion in Q2, capitalizing on smarter AI-driven ad targeting. This cash flow—bolstered by a stock surge of nearly 10% after hours—allows Meta to invest heavily without the immediate pressure faced by AI startups. Capital expenditures hit $17 billion this quarter, primarily for AI infrastructure like the Prometheus and Hyperion data centers, which are set to scale up to 5 gigawatts by 2026. CFO Susan Li projects 2025 expenses between $114 billion and $118 billion, with even higher spending expected in 2026, signaling Meta’s long-term commitment to AI dominance.
Unlike its metaverse venture, which has racked up $60 billion in losses since 2020, Meta’s AI investments are already showing returns. AI-powered ad tools are boosting revenue by enabling businesses to create ads from scratch and deploy AI agents for customer service on WhatsApp. Zuckerberg emphasized that these practical applications are just the start, with the real goal being a leap toward superintelligence—AI that surpasses human capabilities across diverse tasks.
The Rise of Meta Superintelligence Labs
In June 2025, Zuckerberg unveiled Meta Superintelligence Labs (MSL), a new division tasked with building AI models that can self-improve and approach human-level intelligence. Led by Alexandr Wang, former CEO of Scale AI, and co-led by ex-GitHub CEO Nat Friedman, MSL represents a $14.3 billion bet, including Meta’s 49% stake in Scale AI. The lab has attracted top talent from OpenAI, Anthropic, Google DeepMind, and Apple, with compensation packages reportedly reaching nine figures for some researchers. Shengjia Zhao, a former OpenAI researcher, was named chief scientist, tasked with steering MSL’s research agenda.
Zuckerberg’s hiring spree has rattled competitors. Offers to OpenAI staff have included up to $300 million over four years, with $100 million in immediate stock vesting, alongside promises of “unparalleled compute” resources. This aggressive recruitment, which included failed attempts to acquire startups like Safe Superintelligence and Perplexity, underscores Meta’s urgency to close the gap with rivals like OpenAI and Google. The lab’s focus is on advancing Meta’s Llama models, though Llama 4’s lukewarm reception among developers has prompted discussions about pivoting from open-source to proprietary models.
Personal Superintelligence: A New Vision
Zuckerberg’s vision for “personal superintelligence” sets Meta apart from competitors who prioritize automation to replace human labor. In a public letter and Instagram Reel, he described AI that “knows us deeply, understands our goals, and can help us achieve them,” emphasizing empowerment over productivity. He envisions this technology integrated into everyday devices like smart glasses, which Meta has sold 2 million pairs of in partnership with Ray-Ban. A prototype AR glasses project, Orion, hints at future AI-driven wearables, though it’s not yet consumer-ready.
This focus on personal empowerment contrasts with OpenAI’s Sam Altman, who has suggested AI could lead to universal basic income by automating jobs. Zuckerberg argues that superintelligence should enhance creativity and personal agency, allowing users to pursue passions rather than just streamlining work. He’s betting that smart glasses, not traditional productivity software, will be the primary interface for this AI, leveraging Meta’s expertise in consumer products to reach billions.
Challenges in the AI Race
Meta’s AI ambitions face hurdles. Llama 4’s struggles highlight technical challenges, as developers found it harder to customize than Llama 3, leading some internal teams to consider rival models like Claude. Zuckerberg’s open-source strategy is also under scrutiny, with reports suggesting MSL may shift toward closed models to better compete with proprietary systems from OpenAI and Google. Open-sourcing raises safety concerns, as models can be manipulated if safeguards are bypassed, prompting Zuckerberg to stress “rigorous” risk mitigation.
The financial stakes are high. Investors, while pleased with Meta’s ad revenue, are wary of the massive AI spending, especially after the metaverse’s costly missteps. Analysts like Emarketer’s Minda Smiley note that while AI is boosting Meta’s core business, the long-term payoff for superintelligence remains uncertain. Zuckerberg himself acknowledged on the earnings call that returns will take time, but he’d rather build capacity early than lag behind.
What’s Next for Meta’s AI Push
Meta’s path to superintelligence hinges on its ability to integrate cutting-edge AI into its vast ecosystem. The Meta AI assistant, projected to be the most-used globally by year-end, is already enhancing user engagement on Instagram and WhatsApp. Future applications could include AI-driven video ad tools or personalized experiences via smart glasses. With data centers like Prometheus coming online in 2026 and a talent-dense team in place, Meta is positioning itself as a serious contender in the AI race.
For users, this could mean smarter, more intuitive tools that blend seamlessly into daily life—think AI assistants that anticipate needs or glasses that overlay helpful information. But the road to superintelligence is fraught with technical and ethical challenges, and Meta’s success depends on balancing innovation with responsibility. As Zuckerberg put it, the next few years will be “decisive” for determining whether superintelligence empowers individuals or reshapes society in less predictable ways.
