Startups Race to Power AI with Brain-Based Chips, Wave-Driven Data Centers and Micro-Nuclear Reactors
Innovative firms are tackling AI's soaring energy needs by using rat-brain-derived adapters, ocean-wave-powered floating nodes and compact nuclear microreactors.
Founded in 2022, Biological Computing Company grows rat-brain or stem-cell neurons, stimulates them electrically and translates the responses into mathematical adapters that sit atop existing AI models, reporting efficiency improvements as high as 500% for a generative video system. Across the country, Panthalassa constructs 300-foot floating nodes that ride ocean waves, driving turbines to generate near-continuous electricity and processing tasks locally before relaying results via satellite.
In the nuclear arena, Valar Atomics built its first microreactor, Ward 250, which reached a zero-power criticality test in Utah and is slated to produce 100 kilowatts, while Deployable Energy also hit criticality, joining a handful of U.S. startups with operational microreactors. Both companies have secured major venture funding—Valar with a $1 billion Series B led by Sequoia Capital—and plan to mass-produce reactors for deployment at AI-intensive facilities. These parallel efforts illustrate a multi-pronged push to meet the massive power appetite of next-generation artificial-intelligence workloads.
Why it matters
AI's exploding compute demand requires new, low-carbon power sources; these breakthroughs could reshape energy supply for tech and industry.
In this story