💰 Money

AI’s next bottleneck may be electricity, not computing

2 min read Tiny Why Newsroom · By Curio, Martian correspondent

Words
International Energy Agency (IEA)

An organization that studies energy use and policy.

Stated Policies Scenario

A forecast based on policies governments have formally proposed.

terawatt-hour

A large unit for measuring electricity use.

What happened

The International Energy Agency (IEA), an organization that studies energy, says global data-center electricity use could double by 2030. Under its Stated Policies Scenario, data centers could consume about 1,000 terawatt-hours each year by then. CNBC reported the estimate on November 19. The forecast ties the increase to the artificial-intelligence boom and its demand for more computing facilities.

Why demand is rising

Data-center electricity use grew 17% in 2025. Global electricity demand grew 3% that year. Data centers therefore expanded much faster than overall power demand.

The IEA points to large project pipelines in Texas, Virginia, Finland, and the United Kingdom. In several regions, data centers waiting for grid connections would require more power than current peak demand. This means construction plans can move faster than the networks that must serve them.

Why grids matter

Building a data center is only one step. The facility also needs a connection to the power network. Without grid upgrades, networks could become more congested and less reliable. New projects could wait longer for connections. Generators could be forced to cut output when the grid cannot meet demand.

The pressure also reaches equipment supply. The IEA says copper, cables, and transformers are already tight. That makes the timing of grid expansion important for data-center construction.

What the numbers show

Across the IEA’s scenarios, peak electricity demand could rise 40% to 50% by 2035. Under the Stated Policies Scenario, grids would add 26 million kilometers of new lines by 2035. About 16 million kilometers of existing lines would need replacement or refurbishment.

Under the IEA’s Current Policies Scenario, grid spending is projected to rise 75% over the next decade. Meanwhile, the world’s largest data-center operators are expected to spend more than $700 billion on infrastructure in 2026, mostly on new facilities. The figures show strong investment. They do not prove that every planned facility will connect on time.

What remains uncertain

The 1,000-terawatt-hour figure is a forecast, not a measured future result. The Stated Policies Scenario includes policies formally put forward but not yet adopted. Actual demand could differ if AI growth changes, projects face local opposition, supply chains tighten, or grids expand more slowly.

The report also does not settle the effect on every community. It describes global demand and network pressure. It does not establish local electricity prices, reliability, or construction dates.

What to watch next

The clearest test will be the connection queue. Watch whether projects in places such as Texas and Virginia receive grid connections, and whether new lines and existing-line upgrades proceed. If demand grows faster than network capacity, facilities may open later. If the network expands, data-center growth may continue more smoothly.

The forecast places the power network at the center of AI’s next growth test. More computing requires more physical infrastructure. Power systems must deliver the needed electricity on schedule.

💰 Money

Why AI is making electricity a bigger problem

📰 Full story: AI’s next bottleneck may be electricity, not computing

The IEA says data centers could use twice as much electricity by 2030.

2 min read Tiny Why Newsroom · By Curio, Martian correspondent

Words
data center

A large facility where computers handle digital work.

power grid

The network that carries electricity to places that use it.

Stated Policies Scenario

A forecast using policies governments have formally announced.

💡 The gist

  • Data centers could use twice as much electricity by 2030.
  • The AI boom is helping push that demand higher.
  • Power grids need more lines and upgrades.

Why demand is growing

The International Energy Agency (IEA), an energy research group, made the forecast. Data centers are large places filled with computers. AI systems use them for calculations.

Data-center electricity use grew 17% in 2025. Global electricity demand grew 3%. Data centers grew much faster than the world’s total demand.

The IEA expects about 1,000 terawatt-hours yearly by 2030. This estimate uses the Stated Policies Scenario. That scenario counts policies governments have formally announced. It is a forecast, not a promise.

Why power grids matter

Power grids carry electricity to places that use it. Large projects are planned in Texas, Virginia, Finland, and the United Kingdom. Some planned centers already wait for grid connections.

In several regions, those waiting centers would need more power than current peak demand. Without upgrades, grids could become more crowded and less reliable. New centers might wait longer. Generators might need to reduce output when the grid cannot meet demand.

What the numbers show

The IEA says peak electricity demand could rise 40% to 50% by 2035. One scenario includes 26 million kilometers of new lines. It also includes replacing or repairing 16 million kilometers of existing lines. The IEA says copper, cables, and transformers are already tight.

The world’s largest data-center operators are expected to spend more than $700 billion on infrastructure in 2026. But spending does not guarantee a quick connection.

What remains uncertain

Actual demand may differ. AI growth, construction delays, local opposition, supplies, and grid capacity all matter. The report gives a global forecast. It does not settle prices or reliability for every community.

What to watch next

Watch connection queues and grid upgrades. If lines arrive late, facilities may open late. If upgrades keep pace, data-center growth may continue more smoothly. The main question is simple: can electricity networks keep up with AI demand?

💰 Money

AI computers may need much more electricity

📰 Full story: AI’s next bottleneck may be electricity, not computing

More AI computers can make electricity roads busier.

1 min read Tiny Why Newsroom · By Curio, Martian correspondent

Words
data center

A big building where many computers work.

power lines

Wires that carry electricity to buildings.

International Energy Agency (IEA)

A group that studies energy.

Big computer buildings

AI helps computers do smart tasks. Data centers are big buildings where many computers work. The International Energy Agency (IEA), a group studying energy, made a future guess. By 2030, data centers could use twice today’s electricity. Twice means adding the same amount again.

Electricity roads

Power lines are like roads for electricity. A power grid is a group of those lines. More data centers mean more places need electricity. Some new centers already wait for grid connections. Their electricity road is not ready yet.

What happens next

The IEA says new lines and old-line repairs will matter. People will watch whether power lines keep up. The future guess is not a promise. Power networks must grow as data centers grow.

Sources