Meta

A $135B-$145B 2026 capex forecast and new nonvoting shares underscore Meta’s shift toward financing AI at scale while child-safety and privacy liabilities mount.

Who they are

Meta is the corporate parent behind Facebook, Instagram, WhatsApp and Threads, appearing in coverage as a consumer-platform operator that is increasingly using its vast social reach to distribute AI products, agents and hardware. Mark Zuckerberg is central to both its product and capital-allocation story.

The recent arc

Coverage reached its all-time quarterly high in 2026 Q3, driven above all by litigation over alleged social-media harms and the resulting state settlement. A Los Angeles jury found Meta and YouTube negligent in the social-media addiction trial in March; by August, Meta had agreed to pay up to $18B to settle claims by US states over Facebook and Instagram, with daily teen limits and feed controls among the measures described in the related settlement coverage. A New Mexico court’s October order for $25B in Cambridge Analytica-related penalties kept data-practices scrutiny at the forefront.

At the same time, the stories show a decisive AI-infrastructure and product phase. Meta acquired Manus to bring agents to Meta products and Meta AI while continuing to sell the Manus service; it released the open-weight Muse Glimmer model and planned an open-weight Muse Spark 1.2 release. Its October results paired 26% year-over-year Q3 revenue growth to $64.6B and 3.64B family daily active people with a sharply lower operating-income figure because of a settlement charge, while the company lifted the low end of its 2026 capex outlook to $135B and introduced plans for nonvoting Class C shares to fund AI-model development.

The tension

The coverage centers on a costly two-front challenge: Meta is spending aggressively to compete in AI with Google, Microsoft, Amazon and other Big Tech peers while its Facebook and Instagram businesses face legal and policy pressure over youth safety, addictive design and data practices. The financing structure itself has become part of that tension, including reported Hyperion data-center financing, $30B of data-center debt moved through SPVs, and a broader debt build-up since 2022.

Why it matters

If Meta can turn AI models, agents and devices such as Meta Ray-Ban Display into products used across its social platforms, its distribution scale could make the infrastructure outlay strategically consequential. But the legal settlements, penalties and new safety controls may raise the cost and constraints of operating those same platforms; whether AI-led growth offsets that burden depends on execution, regulatory outcomes and sustained user and advertiser demand.

Related: Facebook · Instagram · Mark Zuckerberg · AI · Meta agrees to pay up to $18B to settle US states' claims that it desi · Social media addiction trial: an LA jury finds Meta and YouTube were n