The cost of running artificial intelligence hardware has opened an unexpected way for AI firms to cut outgoings while allowing individuals to make money from their home technology, according to Titan Network.
The internet infrastructure company said its software pools unused computing resources and rents it out as a “decentralized cloud” to AI companies, who pay less than they would if buying capacity from large, centralized providers.
Powering AI is big business. The software requires massive computing resources, and the data centers consume huge amounts of power to run the machines and cool the buildings. Many bitcoin mining companies have those setups and a number, including MARA Holding (MARA) and Riot Platforms (RIOT), are pivoting to serve the rising demand. On Monday, Alphabet (GOOG) said it planned to raise a whopping $80 billion to spend on AI infrastructure.
“We have two of the top 10 AI companies in the world using our products to realize 75% cost savings on their infrastructure,” Konstantin Tkachuk, founder and chief strategy officer, said in an interview at the Proof of Talk conference in Paris.
The company now has 4 million connected devices worldwide and clients including Tencent, Alibaba, and the AI video platform Kling AI, Titan Network said. About 1 million devices are online at any one time.
Titan is not the first project to try lowering costs by aggregating unused computing capacity in what’s known as a decentralized physical infrastructure network (DePIN) system. Unlike platfoms such as Aethir and Akash Network, which target spare cycles on institutional servers, Titan says it uniquely links to private citizens.
“Titan has broken the code no one else has been to before, enabling regular people to make money from the up-and-coming AI data infrastructure industry,” River Davis, Titan’s creative director, told CoinDesk.
When big companies pay to use the network for data tasks like web scraping, data collection, or content delivery, Titan sends 80% of those corporate earnings directly to the people providing the devices and internet bandwidth, who have downloaded a browser plugin or some specialized software.
The project said it has already captured roughly 5% of the AI data market in Asia.