Urgent.News

What's breaking now, across thousands of outlets.

AI

Could Europe’s start-ups help solve the power problem created by the AI boom?

Euronews Next looked at startups in Europe that are trying to make better use of existing grid capacity, reduce the electricity used for cooling and put waste heat to work.

Europe's burgeoning artificial intelligence (AI) sector is straining the continent's electricity grid, as data centres consume more power to support AI capacity. This growing demand is making it harder to build and connect new facilities. However, European startups are exploring innovative solutions to address the power crunch.

One approach is to rethink cooling methods. UK startup EkkoSense employs wireless sensors to monitor temperature, airflow, and cooling capacity within data centres. By visualising the data in three dimensions, operators can identify equipment receiving improper cooling, leading to a 15% reduction in cooling energy for Virgin Media O2, a major telecommunications company.

Meanwhile, German startup etalytics uses simulation software to optimise chiller, pump, and heat exchanger performance, resulting in a 19.1% decrease in chiller electricity usage during testing.

Spanish company Submer takes a different route by replacing cooling hardware with a non-conductive liquid that directly absorbs heat from servers. This method can increase energy efficiency by up to 50%, reduce carbon emissions from refrigerant gases, and improve the total cost of ownership. Telefónica, a Spanish telecommunications company, tested Submer's system and reported significant energy savings.

Beyond cooling, software from Irish firm GridBeyond helps manage demand by controlling when batteries charge and discharge in response to electricity network conditions. Installed at two Dublin data centres, GridBeyond's software can provide 8 MW of flexible capacity, reducing the need for grid power during peak demand.

Another innovative solution comes from UK startup Deep Green, which places compact computing facilities near establishments that require heat, such as swimming pools and district-heating networks. By transferring waste heat from servers, Deep Green claims swimming pools could cut gas use by 62%, save over £20,000 annually, and reduce carbon emissions by 25.8 tonnes. Deep Green's CEO, Mark Bjornsgaard, emphasizes the potential for "digital boilers" to provide around 30% of industrial and commercial heat needs.

While these startups offer promising solutions, they cannot fully replace the need for physical expansion of data centres and electricity generation infrastructure. Europe will ultimately require more electricity and stronger grids to accommodate AI growth, with new grid projects taking between five to 15 years to plan and implement. Nonetheless, the efforts of these startups may help ease pressure on Europe's power system during the development of this critical infrastructure.

Written by urgent.news from Euronews's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at euronews.com →

More in AI

How I Built an App With No CS Background (Firebase + React, AI-Assisted)

A few months ago I couldn't tell you the difference between a Cloud Function and a Firestore trigger. Today, Veyl — an anonymous accountability partner matching app — is live in production, handling…

  • Anonymous accountability partner app Veyl launched with Firebase and React
  • Author built Veyl by collaborating with AI tools and mastering fundamentals
  • Challenges included Firestore security rules, atomic batch writes, and webhook verification

AI Agent Sandboxing: Practical Guide for Production Safety

AI agent sandboxing is the control plane for safe autonomy The trouble starts when an agent can act faster than your guardrails.

  • AI agent sandboxing provides hard boundaries for agent actions.
  • Containers offer fast execution for low-risk tasks.
  • Ephemeral VMs and microVMs provide stronger isolation for high-risk tasks.

More from Wednesday 23 September →