< Back to news 
May 22
Private Firms Now Dominate AI Supercomputing: ‘We’re at the mercy of Big Tech’
The global AI arms race is no longer driven by governments but by tech giants. New U.S. research shows that 85% of the world’s most powerful AI supercomputers are now privately owned—up from just 40% in 2019.
At the forefront: Elon Musk’s Colossus, built in just 122 days and powered by 200,000 Nvidia chips making it 50 times more powerful than the top government system from six years ago. The cost? $7 billion. Energy use? Equivalent to 250,000 households. “This paints a grim picture,” says Prof. Cees Snoek, computer science expert at the University of Amsterdam. “The public sector has virtually no computing power left. In the Netherlands, AI researchers share just 650 chips. We’re completely reliant on big tech.”
The outlook is stark: by 2030, the most advanced AI computer is expected to cost $200 billion and require the energy output of nine nuclear reactors.
Vergelijkbaar >
Similar news items
Heartbeat analysis helps unmask deepfakes
Researchers from NFI and UvA present an innovative method to detect deepfakes by analyzing subtle facial color changes that follow the rhythm of the heartbeat.
read more >
June 8
New AI model ‘Aurora’ promises breakthrough in disaster forecasting
An international research team including UvA scientists has introduced Aurora, a powerful AI model that dramatically improves forecasting for air quality, ocean waves, and extreme weather events.
read more >
May 29
Building responsibly on foundation models: practical guide by Utrecht University of Applied Sciences and RAAIT
Researchers from RAAIT have published a practical guide for organisations aiming to develop AI applications using foundation models. The guide supports responsible decision-making.
read more >