More Than 5,100 AI Data Centers by 2027? A Forecast Meets Local Pushback
Kalshi
A market where people trade views about future events.
data center
A large facility that holds computers for digital services.
prediction market
A market that turns views about future events into trades.
What happened
CNBC reported that traders on Kalshi, a prediction market for future events, see high odds of more than 5,100 U.S. data centers being planned or operating before 2027. This is not an official government count. It is a market forecast. The number may include projects that are still being planned.
That distinction matters. A planned building is not the same as an active building. A forecast can also change when new information arrives. So the headline signals strong expectations. It does not prove that 5,100 facilities will exist.
The background: AI needs physical infrastructure
Kalshi’s own research expects about 5,120 active U.S. data centers by the end of 2026. Its early-June forecast was 4,700. That is a rise of 420 in the expected total. The same research says Amazon, Google, Meta, and Microsoft, four large technology companies, could spend about $725 billion on AI infrastructure in 2026. That would be about 77% more than their roughly $410 billion in 2025.
A data center is a large facility filled with computing equipment. AI services need those machines to process requests and train or run models. This makes the story bigger than a technology-company spending plan. It connects AI growth with buildings, electricity, land, and local approval.
Why the forecast matters
The number suggests that investors and traders still expect AI demand to remain strong. It also shows how quickly expectations can move. A forecast of 4,700 became a forecast of 5,120 within a short period. That change is a signal about confidence in continued construction.
But more buildings do not automatically mean better AI services. Projects need permits, power connections, equipment, and customers. Some projects may be delayed or canceled. Others may be built for future demand and remain partly unused.
Why people are pushing back
The buildout is also facing public resistance. Related CNBC reporting describes concerns about electricity costs and pressure on local power grids. Residents and policymakers are asking who should pay for the extra capacity needed by large facilities. The benefits and costs may not reach the same people.
This creates a real tension. Companies see data centers as necessary for the AI economy. Communities may see them as a claim on scarce power or other local resources. The prediction market does not settle that argument. It only shows that traders expect construction to continue despite the resistance.
What is confirmed
The confirmed facts are limited but useful. CNBC reported the high Kalshi odds. Kalshi’s research gave the 5,120 year-end estimate and the earlier 4,700 forecast. Those figures show a stronger market expectation, not a final count. The original CNBC report and Kalshi research describe the forecasts.
What remains unknown
The exact definition behind the 5,100 figure needs attention. Does it count only active facilities, or also announced projects? How many projects have secured permits? How much power can local grids provide? The sources do not prove that every planned facility will open.
There is also a limit to reading market odds as facts. Kalshi prices reflect the views of traders. They are useful signals, but they are not engineering reports or government statistics.
What to watch next
The next test comes at the end of 2026. Watch the actual number of operating facilities, new permits, project cancellations, and local electricity prices. It will also matter whether the large technology companies keep increasing infrastructure spending. Those facts will show whether the AI buildout is becoming a real, lasting network of facilities, or only a very large set of expectations.
Why AI Data Centers May Pass 5,100
📰 Full story: More Than 5,100 AI Data Centers by 2027? A Forecast Meets Local Pushback
More places for AI computers may be built in the United States. The number is a forecast, not a final count.
forecast
An informed guess about what may happen.
data center
A large building filled with computers.
Kalshi
A market for guesses about future events.
💡 The gist
- Traders expect more than 5,100 data centers by 2027.
- Some of these places may still be only plans.
- Local power concerns are growing too.
Kalshi is a market for guesses about future events. Its traders see a high chance of more than 5,100 U.S. data centers before 2027. A data center is a large building filled with computers. These computers help AI systems process information.
The number is not a government report. It is a forecast. A forecast is an informed guess about what may happen. It can change when new facts appear. It may also count planned buildings. Planned buildings are not operating buildings.
Kalshi’s research gives another number. It expects 5,120 active U.S. data centers by the end of 2026. Its forecast in early June was 4,700. The difference is 420. That change shows that traders became more confident about continued construction.
Why does this happen? AI companies need more computing power. Amazon, Google, Meta, and Microsoft are four large technology companies. Kalshi says they may spend about $725 billion on AI infrastructure in 2026. That is about 77% more than their roughly $410 billion in 2025.
Building these centers takes more than money. Companies need land, equipment, permits, and electricity. Local grids must provide enough power. Some residents worry that large centers could increase pressure on power systems or electricity bills. This has created growing opposition in some communities.
The forecast does not answer every question. It does not show that every project will open. It does not show who will pay for new power supplies. It also does not prove that more buildings will make AI better for everyone.
The next important facts will come at the end of 2026. People will check how many centers are operating. They will also watch permits, cancellations, power prices, and company spending. Those facts will test the forecast. A market guess is useful, but reality gives the final answer.
More Big Computer Houses May Be Coming
📰 Full story: More Than 5,100 AI Data Centers by 2027? A Forecast Meets Local Pushback
AI needs many computers. More computer houses may be built soon.
data center
A big building where many computers stay.
Kalshi
A place where people guess about the future.
A data center is like a big house for computers.
The computers help AI do its work.
Kalshi is a place where people guess about the future.
Many people there think more than 5,100 computer houses may exist or be planned before 2027.
That is a guess.
It is not a finished count.
Some houses may still be only plans.
Building them needs land and electricity.
People nearby may worry about the power supply.
At the end of 2026, we can check what really happened.