China has ordered Meta to unwind its roughly $2-billion acquisition of Manus, an AI startup headquartered in Singapore with Chinese roots. Beijing framed the move as a national security matter, but the ripple effects go beyond geopolitics and venture capital.
Because when governments treat AI like a strategic asset, they also end up shaping where the world builds power-hungry data centers, how those facilities get electricity, and how much cooling water they pull from already stressed regions.
You can feel that shift in the background noise of daily life, from the electric bill to summer heat waves that make cooling infrastructure work overtime.
What China’s order actually does
China’s National Development and Reform Commission (NDRC) published a security review decision on April 27, 2026 that prohibits foreign investment in the acquisition of the Manus project and requires the parties to revoke the transaction. The statement is short, but the message is loud.
Earlier this year, China’s Ministry of Commerce also said it would assess the deal’s consistency with rules covering areas like technology import and export, export controls, outbound investment, and cross-border data transfers. That matters because it signals the review was not only about who owns what, but also about how tech and data might move across borders.
Manus drew attention because it markets a general-purpose AI agent that can handle complex work like coding and market research, and it publicly claimed it crossed $100 million in annual recurring revenue within months of launch. Meta’s bet was straightforward, with AI agents seen as the next big platform layer.
AI agents are not “just software” anymore
There’s a part that gets overlooked in the deal chatter. AI agents do not live on laptops, they live in data centers, and those buildings run like industrial facilities with servers, cooling systems, backup power, and round-the-clock electricity demand.
The International Energy Agency says data centers’ global electricity demand grew 17% in 2025, while AI-focused data centers grew even faster, at 50% in 2025. That kind of growth does not stay inside the tech industry, it spills into power markets and infrastructure planning.
The IEA also projects electricity generation needed to supply data centers could rise from about 460 terawatt-hours in 2024 to more than 1,000 terawatt-hours by 2030 in its base case. It expects renewables to meet nearly half of the additional demand, but natural gas and coal are still in the mix, with nuclear playing a bigger role later in the decade.
The climate squeeze shows up in power grids and prices
When demand rises fast, grids feel it first. In the United States, Reuters reports that rising power demand tied to AI and data center expansion is one factor pushing electricity prices up for consumers, alongside infrastructure and fuel costs. It is the kind of macro trend that hits households indirectly, month after month.
Even the U.S. Department of Energy has framed data center growth as a major load story. A DOE release tied to a Lawrence Berkeley National Laboratory report says data center load growth has tripled over the past decade and is projected to double or triple again by 2028.
And when grid interconnections take too long, some developers go around the system instead of through it. Reuters noted that transmission bottlenecks can push data center builders toward gas-fired, on-site generation, which can lock in emissions even as companies talk up clean energy.
Cooling water is becoming the next flashpoint
Electricity is only half the environmental story. Cooling is the other half, and in many designs that means water.
The Environmental and Energy Study Institute notes that a medium-sized data center can consume up to roughly 110 million gallons of water per year for cooling, while some large facilities can reach millions of gallons per day.
The exact number depends on location, weather, and cooling approach, but the direction is the same–more computing usually means more heat that has to go somewhere.
This is where local politics gets real. In a drought-prone region, water used for cooling can become a community issue fast, especially when residents are being asked to cut back on lawns, showers, or irrigation.
There are technical ways to lower the footprint, and some are surprisingly practical. European Commission coverage of new research points to using data center waste heat for water purification and carbon capture, a reminder that infrastructure can be designed to give something back rather than only consume.
Defense logic is shaping the environmental outcome
Why does this belong in military and defense coverage? Because the Manus decision shows AI is being treated as dual-use and strategic, which changes where capital flows and where computation gets built.
China’s intervention sends a warning that relocating or re-incorporating abroad does not automatically remove a project from Beijing’s reach when the founders, talent, or technology are Chinese-linked. Reuters and other coverage have described this as part of a broader push to keep frontier AI capabilities from being transferred to foreign owners.
On the U.S. side, outbound investment restrictions are also tightening. The U.S. Treasury says its outbound investment program rules were issued with an effective date of January 2, 2025, targeting certain investments tied to sensitive technologies.
Meanwhile, Reuters reported China has considered limiting U.S. capital access to top tech firms without approval, showing how quickly the walls can rise from both directions.
Then there’s the physical footprint of the AI race. One recent example approved in Utah is a proposed 9-gigawatt data center campus designed to operate off-grid on on-site natural gas, including land connected to a military installation authority.
Projects like this sit at the intersection of energy security thinking and industrial-scale computing, and they carry real consequences for emissions, air quality, and local resource use.
At the end of the day, this is what readers should keep in mind: AI policy is no longer only about chips, models, and market share, it is also about land, power plants, water rights, and the environmental trade-offs that come with building strategic infrastructure at speed.
The official statement was published on China’s National Development and Reform Commission.
