Introduction
A data center is a big building full of computers. These computers run apps, websites, and AI tools. AI means computer programs that can learn and answer questions. Building these places cost much more in 2026. One big study says the cost went up 13% in one year.
What Happened
Turner & Townsend made the study. It looked at more than 500 data center projects. The projects were in 62 markets around the world. It also asked 401 industry leaders what they see. The study found that the average cost per watt rose 13% in 2026. A watt is a small unit of electric power. Builders use cost per watt to compare projects of different sizes.
The rise was biggest in North America and Europe. More than half of the leaders said costs rose 6% to 15% this year. AI data centers cost about 10% more to build than normal ones in the United States. The most advanced AI buildings cost $14.00 to $16.00 per watt. The average in the United States was $12.08 per watt. Tokyo was the most costly city in the study. Tokyo cost $18.28 per watt.
Key Figures
- Cost rise in 2026: 13% in one year.
- Projects studied: more than 500 in 62 markets.
- People asked: 401 industry leaders.
- AI buildings in the US: about 10% more than normal buildings.
- Top AI buildings: $14.00 to $16.00 per watt.
- US average: $12.08 per watt.
- Tokyo: $18.28 per watt, the highest in the study.
Why It Happened
Many companies want more AI power fast. They all want the same land, power, and workers. There is not enough to go around. Skilled workers are hard to find. In the study, 81% of leaders said skilled labor is tight. Think of a school fair with only two helpers. Every class wants help at the same time. The price of help goes up.
Why It Matters
Big tech companies spend a lot of money on AI. Higher building costs mean they get less for each dollar. It can also slow down new projects. This matters for AI stocks too. Stock means a small piece of a company. When costs rise, a company may earn less profit. Profit means money left after costs are paid. Readers who follow AI stocks can read about Anthropic stock falling 9% on Friday on this blog. More stories like this are on our Business News page.
What Happens Next
Paul Barry from Turner & Townsend spoke to CNBC. He said costs may rise 6% to 10% next year. That would be slower than this year. But it would still be a big rise. Companies will watch power and workers closely. They need both before they can build.
Final Thoughts
Data centers are the homes of AI. More people want AI, so more homes are needed. But workers and power are in short supply. That is why building costs jumped 13%. The next year will show if the rise slows down.
Quick Questions
- How much did data center building costs rise in 2026? They rose 13% in one year. The number comes from the Turner & Townsend study.
- Why did the costs go up so much? Many companies want AI power at the same time. There are not enough skilled workers or power to go around.
- Which city cost the most in the study? Tokyo cost the most. It cost $18.28 per watt.
Disclaimer
Disclaimer: This article is for educational and informational purposes only and does not constitute financial advice.
Sources
CNBC and Turner & Townsend. Picture: Rows of computer servers in a data center. Photo by Victorgrigas, CC BY-SA 3.0, via Wikimedia Commons.

Leave a Reply