Google puts 13 billion euros into Finland and extends a nuclear plant's life

Google will invest at least 13 billion euros in Finland over two years. It is the company's largest single investment in Europe and ranks among the largest ever announced in Finnish industrial history. The money goes to AI and cloud capacity: the existing Hamina data centre expands, and three new sites are being built in Kajaani, Vaala and Muhos. Construction runs across 2027 and 2028.

The real story is electricity

Data centre announcements usually talk about floor space and chips. The detail that matters here is on the energy side: Google has signed a 22-year agreement to support extending the life of the Loviisa nuclear plant. On top of that come new onshore wind capacity and a 94-megawatt battery system to stabilise the grid.

So an AI investment is changing a country's nuclear retirement schedule. Now that data centre demand shapes energy policy itself, that line says more than the 13 billion euro figure does.

This is Google's second move on the nuclear side. In the United States the Energy Department approved a $1.9 billion loan to restart a shut Iowa plant, again with Google as the offtaker. The difference in Finland is that this is not reopening a closed plant but postponing the closure of a running one.

The numbers

• 37,000 jobs supported during construction, roughly 16,000 of them in building work.
• 3.6 billion euros a year added to GDP during the initial construction period.
• 31 million euros for local communities, on education and economic growth.
• AI upskilling for more than 4,400 workers, and data centre career training for 100 Finnish students.
• Google has been in Finland since 2009 and put 4.5 billion euros into Hamina over 15 years. The new package is almost three times that.

Hamina was built on the site of a former paper mill and is cooled with seawater, with waste heat distributed to nearby homes and businesses. The same approach is described for the new sites.

What this actually is

Finland was not picked by accident: a cold climate, cheap low-carbon electricity, a stable grid. The AI race now turns less on model quality than on where power and cooling can be found, and northern Europe stands out on exactly that measure.

Keep the scale in view: 13 billion euros is a large figure, but it is spread across one country, two years and four sites. At a moment when the cost of training a single frontier model is a talking point, infrastructure at this size shows AI moving out of software and into heavy industry. The winners include energy companies, not only chipmakers.

Sources: Google announcement, Euronews.