SIGNAL · 01The headline is not the story. Electricity is the story.
In recent months, media coverage of AI has continued to focus on models, capabilities, and the competition between companies. But beneath the surface an entirely different layer is moving, one that receives not even half the attention, and that holds most of the strategic risk and opportunity: the electricity layer.
Adopting AI at enterprise scale requires computing power that is growing faster than any previous forecast. Computing power means data centers. Data centers mean electricity. A lot of electricity, delivered continuously. This is no longer a novelty; it is physics.
Whoever holds the electricity, the land and the cooling holds the future of information.
SIGNAL · 02The numbers that change the picture
To see the scale of the shift, you need to hold a few simple numbers in mind at the same time:
- Global data center electricity consumption is expected to at least double by 2030, according to an IEA base-case scenario, and to triple under a rapid-adoption scenario.
- A generative AI query can consume 5 to 10 times as much electricity as a standard search query. At 2026 usage levels, this is not a marginal detail; it is a change of magnitude on the grid.
- Large new server farms planned in the United States and Europe consume electricity on the order of a mid-sized city, and sometimes more.
- In countries with weak grids (parts of Europe, some developing countries), the wait time for a server farm's grid connection already reaches 3 to 7 years.
SIGNAL · 03How it reaches real estate, and why it changes land decisions
Land that until recently was "secondary industrial," or agricultural in less attractive areas, is suddenly becoming strategically valuable, if it has three things: proximity to a grid with available capacity, access to water for cooling, and a short connection time.
This changes several traditional decisions in real estate:
- Land valuation stops being a function of "market location" alone. It becomes a function of access to energy infrastructure.
- Peripheral areas with grid upgrade plans gain new options they did not have 12 months ago.
- "Green real estate" deals are starting to be measured by long-term energy capacity as well, not only by building codes.
- Commercial real estate companies that cannot host AI loads are beginning to feel pricing pressure on assets that, just a year ago, were "winners."
SIGNAL · 04Why this becomes a story about sovereignty
Once governments understand that national AI capability = national electricity infrastructure, the category stops being a matter for private companies. It becomes a matter of sovereignty. Four early signals mark the shift:
- Legislation that enables a regulatory fast-track for national AI infrastructure projects.
- National plans to establish dedicated "energy corridors" for data centers.
- Security partnerships that begin to include electricity supply for data centers, not only intelligence sharing.
- Changes in electricity pricing for industry versus server farms, a sign that the government is starting to set priorities.
The implication: within 2 to 4 years, those who hold land in areas covered by a "computational sovereignty" plan will be in an entirely different position from those who hold "ordinary" land.
SIGNAL · 05Signals worth tracking in the coming months
- Bills or regulations that address connecting data centers to the grid. Every such change moves the price forward.
- New PPAs (Power Purchase Agreements) between large AI companies and electricity producers, a sign of a long-lock strategy.
- Zoning plans that introduce "data center zones" as a dedicated category. A sign of "energy-driven value uplift" zones.
- Changes in electricity tariffs for heavy industry versus compute. A sign of policy focus.
- Investment in SMR nuclear power (Small Modular Reactors). Large contracts being closed mean the market is already building AI into its forecast.
SIGNAL · 06What kind of decision this changes now
This signal is especially relevant to three types of decision-makers:
- Real estate developers. Selecting land for 5- to 10-year projects should start with a check of the regional power grid, not only with price per square meter.
- Institutional investors. Combining energy assets and land assets in the same investment thesis: a multi-layer analysis, not a separate review.
- Public planning bodies. Regional zoning plans that do not account for the AI/electricity layer are quickly becoming outdated.
The real estate decision of 2027 starts with the power grid map of 2026.