Akash founder Greg Osuri warns that AI training could lead to a global energy crisis.

Artificial intelligence (AI) is facing an energy crisis as computational demands surge, potentially requiring power equivalent to a nuclear reactor soon, according to Greg Osuri – founder of Akash Network. He warns that data centers currently consume hundreds of megawatts of electricity from fossil fuels, posing a risk of increasing household electricity bills and emitting millions of tons of emissions each year.

The solution, according to Osuri, is decentralized: instead of concentrating power in super data centers, AI training can be distributed across smaller GPU networks, from enterprise chips to household graphics cards. This approach is reminiscent of the early days of Bitcoin, when users contributed processing power to earn rewards.

Despite the ongoing challenges related to technology and incentive mechanisms, Osuri believes that decentralized AI is an inevitable path to reducing energy pressure and building a sustainable AI economy.