AI is driving unprecedented investment in the massive data centers and energy supplies that can support its enormous computational appetite. One potential source of electricity for these facilities is next-generation nuclear power plants, which may be cheaper to build and safer to operate than their predecessors.
We recently held a client-exclusive roundtable discussion on hyperscale AI data centers and next-generation nuclear – two featured technologies on MIT Technology Review’s 10 Breakthrough Technologies for 2026 list. You can watch the conversation here, and don’t forget to subscribe to ensure you get instant access to future discussions.
How social media encourages the worst AI boosterism
Google DeepMind CEO Demis Hassabis summed it up in three words: “This is shameful.”
Hassabis was responding to an overexcited post from Sebastian Bubek, a research scientist at rival firm OpenAI on X, announcing that two mathematicians had used OpenAI’s latest large language model, GPT-5, to find solutions to 10 unsolved problems in mathematics.
Put on your math hat for a minute, and let’s see what this mid-October beef was about. This is a perfect example of what’s wrong with AI right now.
-Will Douglas Haven
This story originally appeared in The Algorithm, our weekly newsletter on AI. To be the first to get stories like this in your inbox, Sign up here.
Paints, coatings and chemicals make the world a colder place
It is becoming difficult to control the heat. During the summer of 2025, heat waves collapse power grids in North America, Europe, and the Middle East. Global warming means more people need air conditioning, which requires more electricity and strains the grid.
But a millennium-old idea (as well as 21st century technology) may provide the answer: radiative cooling. Paint, coatings and textiles can scatter sunlight and dissipate heat – no additional energy required. Read the full story.
