AI is Becoming Sovereign Power

TL;DR

Governments have started treating AI as a national security asset, with China ordering Meta to unwind its Manus deal and the Pentagon signing deals for AI on classified networks.

  • Big Tech may spend nearly $700bn in 2026
  • Users see little of the cost behind AI
  • Pentagon AI is the same tech people use

Governments have started treating AI as a national security asset. They are deciding who may own it and putting it inside their most sensitive systems, which is how AI is becoming sovereign power. To the people typing into it, the same technology looks cheap.

The simple chat box hides what every answer depends on: chips, data centres, power, cooling, networks, land and financing. In the first quarter of 2026, Alphabet, Amazon, Meta and Microsoft spent more than $130 billion between them on capital investment, driven by data centres and other infrastructure. Their spending for 2026 could approach $700 billion if their investment plans hold, a level that will affect the wider economy. Users see little of that cost because the largest technology companies carry it upfront.

China and the United States have each shown one side of AI as sovereign power. In April 2026, Beijing ordered Meta to unwind its purchase of Manus, a deal worth more than $2 billion, on national security grounds. Manus is an AI agent company founded in China and based in Singapore. The order shows governments starting to treat AI capability as too important to move freely across borders.

Manus builds AI agents, systems that can plan steps and act across software, so they can become part of how a company operates. A research survey describes such agents as having three parts: a model that reasons, a way to read information and a part that carries out the work. Once a system can act inside business software, control over the technology, the talent and the data becomes strategic.

Washington showed the other side. Google signed a classified AI deal with the Pentagon that allows its AI systems to be used for lawful government purposes. Then, on May 1, 2026, the Department of Defense announced agreements with eight companies to deploy AI on classified networks, the secure government systems used for sensitive work. The eight are SpaceX, OpenAI, Google, NVIDIA, Reflection, Microsoft, Amazon Web Services and Oracle.

The core technology the Pentagon is deploying is the same one people use to write emails, summarise documents and find emotional support.

Lead smarter with AI in just 5 minutes a week

Join 50K+ leaders getting our AI For Leaders newsletter every Wednesday: deep analysis, frameworks, and strategies.

No spam ever • Unsubscribe anytime

AI is Becoming Sovereign Power | Chief AI Leader