Data centres (2025)

Plug, baby, plug

Last updated: 24 Feb 2025

Market 101

Coming to a field near you: lots and lots of concrete. And electricity.

Blame AI. The AI boom requires greater processing power which, in turn, requires more data centres. Consider this: a query on ChatGPT is around 10x more power intensive than a Google search, according to a 2024 Goldman Sachs report.

To accommodate all these AI chips, data centre operators are planning and constructing facilities with capacity of 200-500 megawatts; Morgan Stanley analysts estimate the cost of building a campus at $10m per megawatt.

The price of admission to this game, then, is a steep one (Forbes is calling it the trillion-dollar tsunami). Tech’s spend on data centres is set to be extravagant. Four companies — Amazon, Meta, Microsoft and Google — promise to spend a combined $300bn on AI infrastructure this year alone. The first data centre associated with the $100bn Stargate venture from OpenAI, SoftBank and Oracle will be the size of Central Park.

Europe is trying to keep up but it will be difficult. France recently announced €109bn to invest more in AI and data centres, with President Emmanuel Macron promising to “plug, baby, plug” nuclear-powered AI. France’s great AI hope — Paris-based Mistral — says it will invest “several billions” in a data centre south of the capital.

Several data-centre-related companies have raised big money. UK-based Ori, which partners with data centres to rent cloud-based services, nabbed $140m this month. Nscale, also based in the UK, is building data centres in countries including Norway; it raised $155m in late 2024. Meanwhile, Swedish startup Evroc, which has raised around $60m, recently announced plans to build an AI data centre in France.

Given how much money is needed to build these facilities, the most interesting angle for VCs could be: how do you make sure they use energy efficiently? These are resource-hungry facilities that use a lot of electricity to keep servers cool and the internet running. In Ireland, data centres consumed over a fifth of the country’s electricity in 2023, overtaking the amount of power used by all urban Irish homes combined.

Sifted recently reported that Berlin-based VC firm Project A is looking into startups working on tech to lower the temperature of data centre servers. VC firm Speedinvest is looking “very seriously” at four companies, including one in optimisation (around faster and more efficient data processing and energy usage) and one focused on cooling.

Geo map

Deals

Funding charts

Benchmarks and investors tracked by Sifted

Sifted take

We're going to see a lot more data centres pop up to accommodate the AI boom but electric grid constraints will complicate capacity. The most compelling data centre-related startups for investors, meanwhile, could well be the ones that help cool these facilities and/or find a use for the excess heat they produce.

Early-stage startups

Total funding

€16.5m

Stockholm, Sweden
2022

A self-described green rival to AWS that aims to provide the largest AI infrastructure in Europe. It plans to build a "hyperscale" cloud centre next to Stockholm airport.

Round

Seed

Valuation

Undisclosed


Date

2023

Size

€16.5m

Total funding

€4m

Cambridge, UK
2022

Semiconductor company building chips for data centres founded by CEO Russell Haggar.

Round

Support programme

Valuation

Undisclosed


Date

2023

Size

Undisclosed

Total funding

€160m

Bruyères-le-Châtel, France
2022

Operator behind the first Al cluster from French AI darling Mistral.

Round

Early

Valuation

Undisclosed


Date

2025

Size

€50m

Total funding

€9.17m

Zurich, Switzerland
2023

The startup aims to reduce the energy consumption of data centres while turbocharging their performance.

Round

Seed

Valuation

Undisclosed


Date

2024

Size

€6.44m

Ones to watch

Alpinovx

Veurey-Voroize, France
2017
Early

15.5m

11m

-

Apheros

Zurich, Switzerland
2023
Early

1.6m

1.6m

-

Atlantic Hub

Derry, United Kingdom
2018
Early

-

-

-

Biomémory

Paris, France
2021
Early

22.4m

17m

-

Cambridge GaN Devices

Cambridge, United Kingdom
2016
Growth

58.8m

30.5m

-

CamGraPhIC

Pisa, Italy
2018
Early

970k

970k

-

Cerabyte

Munich, Germany
2022
Early

6.5m

-

-

Clear Decisions

London, United Kingdom
2024
Early

-

-

-

Comino

London, United Kingdom
2011
Early

920k

920k

-

Coolgradient

Amsterdam, Netherlands
2023
Early

-

-

-

Deep Green

London, United Kingdom
2021
Growth

233.6m

233.6m

-

Eclairion

Bruyères-le-Châtel, France
2022
Early

160m

50m

-

EFFECT Photonics

Eindhoven, Netherlands
2010
Late

119m

35m

-

Ekkosense

Nottingham, United Kingdom
2013
Early

1.2m

1.2m

-

Equal1

Delft, Netherlands
2018
Early

24.6m

-

-

Ethos Engineering

Dublin, Ireland
2005
Late

-

-

-

Evroc

Stockholm, Sweden
2022
Early

45m

45m

-

Kao Data

London, United Kingdom
2019
Growth

399m

240.6m

-

Lightium

Zurich, Switzerland
2023
Early

9.2m

6.4m

-

Lumiphase

Stäfa, Switzerland
2020
Early

2.5m

2.3m

-

Mistral

Paris, France
2023
Growth

1.1b

468m

-

NordicEPOD

Holmestrand, Norway
2022
Early

-

-

-

Phocea DC

Avignon, France
2023
Early

5m

5m

-

QiO Technologies

Farnborough, United Kingdom
2015
Growth

27.1m

9.1m

-

Quantune Technologies

Berlin, Germany
2019
Early

13.5m

13.5m

-

Simplyblock

Teltow, Germany
2022
Early

3m

2.5m

-

SoftIron

London, United Kingdom
2012
Early

38.2m

880k

-

Submer

Barcelona, Spain
2015
Growth

103.7m

51m

-

Swegan

Linköping, Sweden
2014
Growth

27.5m

12m

-

Tiger Technology

Sofia, Bulgaria
2004
Early

4.7m

2m

-

Tloop

Stockholm, Sweden
2021
Early

1.3m

1.3m

-

Upmem

Grenoble, France
2015
Early

15.1m

4.1m

-

Vector Photonics

Glasgow, United Kingdom
2020
Early

7.5m

2m

-

VyperCore

Cambridge, United Kingdom
2022
Early

4m

-

-

Warren

Tallinn, Estonia
2016
Early

1.9m

-

-

Xelera Technologies

Darmstadt, Germany
2018
Early

1.7m

1.7m

-

Yondr

Amsterdam, Netherlands
2019
Growth

138.6m

138.6m

-

ZeroPoint Technologies

Gothenburg, Sweden
2016
Early

15.2m

5m

-

Europe’s scaleups

Who early stage startups are up against

(Pre-)Seed

SeriesA

SeriesB

SeriesC

SeriesD+

IPO/Exit

An AI startup developing advanced language models. At the AI Action Summit in France earlier this month, Mistral announced plans to invest billions of euros in building its own data centre in the country. The company currently relies on infrastructure from US tech giants like AWS and Microsoft to train and run its models.

(Pre-)Seed

SeriesA

SeriesB

SeriesC

SeriesD+

IPO/Exit

A specialist developer and operator of data centers designed for AI and advanced computing. The company recently secured a £206m debt raise through Deutsche Bank, one of the world’s leading lenders to the data center sector.

Sources

Your feedback

How would you rate this briefing?

1
2
3
4
5