Uptime Institute Five Data Center Predictions for 2026 - press release

Overview

This document is the press release for the five data center predictions report that Uptime Institute issued in January 2026. The predictions are: concentration of large-model AI compute in a few places, power shortages that onsite generation will not fully solve, carbon capture becoming relevant as gas turbines grow, a continued need for resiliency despite cost pressure, and AI automation moving into production operations.

The most notable number is the projected growth of 75-125 GW in global data center power demand through 2030. The release says this growth will increase reliance on gas turbines for primary power and that carbon capture may become practical. Many developers are proposing onsite power generation, but the long timelines of large power deployments constrain build-outs. On resiliency, the release notes that the cost of redundant capacity in high-density builds is under scrutiny, yet customers and investors are unlikely to tolerate higher availability risk.

Cost pressure therefore does not remove redundancy; it forces designers to justify it more carefully. As a press release, the document does not give the evidence or models behind its predictions, and they are the views of an industry organisation rather than firm forecasts. Its value for designers is the signal that energy supply is no longer just a grid-connection question and that onsite generation scenarios now belong in early planning.

Key points

1. Global data center power demand projected to grow 75-125 GW through 2030.
2. Growth will increase reliance on gas turbines for primary power.
3. Carbon capture may become practical as gas turbines grow.
4. Many developers propose onsite power generation.
5. Long timelines for large power deployments constrain build-outs.
6. Cost of redundant capacity in high-density builds is under scrutiny.
7. Customers and investors are unlikely to tolerate higher availability risk.

Using it in design

• Add an onsite primary generation scenario (gas turbine or engine) to the power module.
• Produce a carbon-intensity output for each power scenario.
• Document the availability risk explicitly before reducing redundancy on cost grounds.
• Reflect long power-delivery lead times in the project schedule.

Who should read it

Designers, investors and developers who set energy and power strategy for data centers.

Publication details

Authors

Uptime Institute

Publisher

Uptime Institute

Year

2026

Document language

English

Source type

Industry body

Access

Free to read

Licence

Copyright retained; citation only — the publisher’s terms apply to the original document

Citation