Imagine a merry parade of brains stitched from copper, silicon, and a pinch of mischief, gobbling energy like it’s a box of fizzy pops. AI is bumping into a wall labeled Energy, and as models grow plump, training them may drink the same fizzy juice that powers a nuclear reactor, says Greg Osuri, the cheeky founder of Akash Network. 🍬🤖
In a chat with CryptoMoon’s Andrew Fenton at Token2049 in Singapore, Osuri warned that the industry underestimates how fast compute demands are doubling and the environmental price tag that comes with it. He noted data centers already gulp hundreds of megawatts of fossil-fuel power. 🏭⚡
Osuri warned the trend could trigger an energy crisis, raising household power bills and adding millions of tons of new emissions each year. “We’re getting to a point where AI is killing people,” he said, a sentence that makes the room feel a tad like a thundercloud with a silver lining (or a very loud alarm bell).
The Whimsical Plan: Tiny GPUs, Big Smiles
On Sept. 30, Bloomberg reported that AI data centers are sending power costs surging in the US. The piece showed data centers have helped cook up higher energy bills for everyday folks. Wholesale electricity costs surged 267% in five years in areas near data centers. Yikes! 😬
Osuri told CryptoMoon that the antidote is decentralization. Instead of stacking chips in one monstrous mega-data center, imagine a merry flotilla-lots of smaller GPUs, from top-tier enterprise cards to gaming chips in cozy bedrooms-working together like a clockwork toy to unlock efficiency and sustainability. “Once incentives are figured out, this will take off like mining did,” he said, adding that home computers may also earn tokens by lending spare compute power. 🪙
This vision harks back to the early days of Bitcoin mining, when ordinary folks could pitch in their processing power and be rewarded. This time, the treasure would be training AI models instead of solving cryptic puzzles.
Osuri says this could give everyday people a stake in the future of AI while trimming costs for developers. 💰🤝
A few bumps along the bumpy road
Not everything shiny and giggly lies ahead. Training large-scale models across a patchwork of GPUs requires wizardry in software and a bit of tightrope coordination. The industry is just beginning to crack this nut. “About six months ago, several companies started showing bits and bobs of distributed training,” Osuri said. “No one has tied all those bits together and actually run a model.” He thinks this could change “by the end of the year.”
Another goblin in the pantry is fair incentive systems. “The hard part is incentive,” Osuri said. “Why would someone lend their computer to train? What do they get back? That’s a trickier puzzle than the actual algorithm.”
Still, decentralized AI training is pitched as a necessity. By spreading workloads across global networks, AI could ease pressure on energy grids, cut carbon emissions, and cook up a more sustainable AI economy. 🧭🌍
Read More
- USD VES PREDICTION
- FIL PREDICTION. FIL cryptocurrency
- USD PLN PREDICTION
- EUR CNY PREDICTION
- JUP PREDICTION. JUP cryptocurrency
- BTC PREDICTION. BTC cryptocurrency
- STETH PREDICTION. STETH cryptocurrency
- INJ PREDICTION. INJ cryptocurrency
- NEAR PREDICTION. NEAR cryptocurrency
- USD MYR PREDICTION
2025-09-30 15:34