Meta to pour the GDP of Kenya into AI infrastructure push in 2026
Summary
Meta plans a dramatic increase in capital expenditure for 2026, targeting $115bn–$135bn to expand AI infrastructure and support its Superintelligence Labs alongside its core services. That top-end figure is broadly comparable to Kenya’s entire 2025 GDP (about $136bn). The spend is designed to accelerate “personal superintelligence” initiatives, merge large language models with recommendation systems to boost ad relevance, and scale agent-based features and shopping tools. Meta has also launched a Meta Compute initiative to oversee tens to hundreds of gigawatts of datacentre capacity and signed nuclear energy agreements to support the power needs.
Key Points
- Meta expects 2026 capex of $115bn–$135bn, up from $72.22bn in 2025.
- The planned spend is roughly equivalent to Kenya’s 2025 GDP (~$136bn).
- Funds will back Superintelligence Labs, larger AI models, and tighter integration of LLMs with Meta’s recommendation and ad systems.
- Meta doubled GPUs for training its GEM generative ads model in Q4 2025 as an example of the ramp.
- Meta Compute will manage a growing global datacentre footprint measured in tens to potentially hundreds of gigawatts.
- Agreements with nuclear suppliers aim to meet future energy demands from expanded datacentre build-out.
- Despite the capex surge, Meta expects 2026 operating income to be above 2025 levels and has given Q1 2026 revenue guidance of $53.5bn–$56.5bn.
Why should I read this?
Because this is not just another earnings note — it’s the tech world doubling down on AI at planetary scale. If you care about where AI compute, adtech, energy demand or datacentre supply chains are headed, this is the sort of move that changes markets, hiring, and policy. Short version: big money, big consequences. We read it so you don’t have to slog through the call.
Author style
Punchy — this story matters. Meta’s bet could reshape ad targeting, server demand and the energy landscape; if you work in infrastructure, ads, AI product or regulation, dig into the details.
Context and relevance
Hyperscalers are locked in an AI infrastructure arms race. Meta’s announced capex joins similar multi‑billion plans from other cloud and AI players (Amazon, Google, Microsoft) and signals further demand for GPUs, networking gear, power capacity and datacentre real estate. The spend has downstream implications for energy policy, sustainability debates, semiconductor supply, and antitrust/regulatory scrutiny — especially as Meta links LLMs to ad monetisation and personalisation. For businesses and policymakers, this is part of a trend that will shape investment, regulation and competitive dynamics across tech and energy sectors in 2026 and beyond.