Meta's AI Spending Spree: Can $135B Capex Fuel a Superintelligence Pivot?

Meta's AI spending is a historic capital commitment. The company plans to spend between $115 billion and $135 billion on capital expenditures in 2026, a move that nearly doubles its 2025 capex. This massive outlay is explicitly targeted at building its "Meta Superintelligence Labs," funding data center expansions and custom chip development to train its next-generation models.

This spending spree is entirely dependent on the cash flow from its dominant advertising business. The core ad engine generated $198.8 billion in revenue last year and is on track for $247.7 billion this year. That revenue stream is the only source capable of funding such a capital-intensive pivot without immediate dilution or debt distress.

The first strain from this dual pressure is already visible. While revenue grew, the company's operating margin dipped to 41% in 2025. This compression signals that the rising costs of AI infrastructure and Reality Labs are beginning to erode profitability, placing even greater importance on the advertising business to subsidize the entire strategy.

Muse Spark: A New Model, A New Monetization Play

Muse Spark is the first major AI model from Meta's new Superintelligence Labs, marking a clear strategic pivot. The closed-source model, which outperformed Google's Gemini on some internal tests, is being tested in the Meta AI app and planned for features like Vibes AI video. This shift signals a departure from the open-source Llama strategy, aiming instead to deliver "personal superintelligence" for tasks like planning trips and analyzing images.

The immediate commercial play is a private API preview for select developers. This controlled access is a classic move to build a premium, high-margin revenue stream from the model's underlying technology. It mirrors the monetization paths of OpenAI and Anthropic, allowing Meta to charge for usage while keeping its core AI stack proprietary.

The launch provided a strong near-term catalyst, with Meta's stock rising roughly 4% after the announcement and jumping more than 9% in midday trading. However, the model's success is now tied to the company's massive $135 billion capex plan. For the API to become a meaningful new profit center, Muse Spark must rapidly demonstrate its competitive edge and drive developer adoption, all while the core ad business funds the ongoing infrastructure build-out.

Catalysts, Risks, and What to Watch

The success of Meta's AI pivot hinges on a few critical forward-looking metrics. The most immediate test is the adoption of Muse Spark. Its private API preview is the first monetization step, but for this to become a meaningful new profit center, developer engagement and API usage must accelerate rapidly. This will be the clearest sign that the model has a viable commercial edge and can generate high-margin revenue to offset the massive capex.

The dominant risk is a widening gap between AI spending and revenue growth. Meta's capex plan is set to nearly double in 2026, a move that will further compress its free cash flow. The company's FCF declined 16% last year to $43.6 billion, and its trailing FCF yield has already fallen to 2.6%. With AI infrastructure investments set to reduce FCF further in 2026, the pressure on operating margins-already under strain from Reality Labs-could intensify. The core ad business must continue to grow robustly to subsidize these losses without triggering a valuation reset.

The stock's reaction shows a market caught between optimism and skepticism. The Muse Spark launch sparked a 9% surge in midday trading, reflecting investor enthusiasm for the new model. Yet the stock has still declined 3% year-to-date, highlighting underlying concerns about the spending spree's impact on cash flow and near-term profitability. The path forward will be defined by whether API adoption can validate the AI investment thesis before the cash burn pressures margins and the FCF yield continue to erode.