Decentralized computing network Aethir said the traffic across its GPU network tells it where to put data centers, an approach it calls demand-led capacity planning.
Routing enterprise AI jobs across hundreds of thousands of GPU containers in more than 90 countries shows which regions run short and which requests go unserved, the company said in a blog post. That evidence sits under every siting decision in its ACCELERATE program.
DeAI News reported in August that Aethir had secured access to 10 sites totaling up to 20 megawatts in the U.S. and Europe. The project has also argued that decentralized clouds prop up prices for used GPUs.
The pitch is about sequencing. Most capacity is sited on a forecast and sold afterward, leaving someone to carry the empty racks. Aethir says signing customers first also unlocks prepayments and project-level financing.
"The winners in this market won't be whoever plans the biggest campus," said Mark Rydon, co-founder and chief strategy officer. "They'll be whoever delivers working capacity first."
Aethir projects up to $700 million in contracts by the end of 2026 and more than $2 billion at full buildout, figures it called forward-looking rather than recognized revenue. The first contracts are expected to close in coming months.
HOW AI WAS USED IN THE PRODUCTION OF THIS PIECE: I drafted this story on Claude Desktop using a custom skill designed to generate story drafts based on blog posts. I asked Claude Desktop for some specific edits to the piece, and then I used our Distro Publisher MCP connector to file the story directly to DeAI News on Distro. I used our built-in story editor to make some final edits to the piece, and I used Gemini to generate the story illustration.