Meta Platforms: AI Return on Invested Capital Is Uncertain (NASDAQ:META)

The market is likely underpricing the difference between "AI improves product" and "AI becomes a capital-intensity problem." For META, the first effect can support ad monetization quickly, but the second effect matters more for the stock: once capex is large enough to absorb most of EBITDA, incremental spending starts competing with buybacks and raises the bar for multiple expansion. The cloud-leasing signal is especially important because it suggests utilization risk now, not just a long-dated monetization story later.

Near term, the clean beneficiaries are still the infrastructure suppliers, especially NVDA, but this is not a pure bull case for the AI supply chain. If META is overbuilding, suppliers can keep shipping while end-customer economics deteriorate, which eventually shifts pricing power away from platform owners and toward vendors with scarce compute or power access. On the competitive side, stronger ad targeting at META should pressure weaker ad-tech names and smaller performance-marketing budgets first, rather than showing up as broad-based industry growth.

The consensus risk is that investors focus on whether management can articulate a 12-24 month ROIC, when the more relevant question is whether the market will tolerate utility-like capex at a consumer-internet multiple for another year. The thesis breaks if management shows measurable ad ROAS improvement, external cloud demand proves sustained and high-margin, or capex guidance is cut materially. If none of that appears, the next 1-3 month catalyst is another guidance reset around 2026 spending and depreciation, which could compress the multiple before operating leverage rebounds.