It is built up layer by layer into an accelerator that can do real work. Then most of them end up idle, bought for the evenings and dark for the rest of the day.
Install the client, set the hours you don’t need your machine, and it joins the network as a compute node. No new silicon, no new building, no new grid connection.
The job brings its own environment, docks onto the card, runs, and detaches. It reads nothing outside itself and leaves nothing behind when the result is returned.
Cost per kilowatt-hour and carbon intensity, sampled continuously, region by region. Each workload is routed to whichever nodes are cheapest and cleanest at that minute.
A price floor below which nothing runs, hours that stay yours, a thermal ceiling that is never crossed. Earnings settle weekly, itemised by job.
Data centers bring energy to compute. Renti brings compute to energy.
Rent out the machine you already own, or run your workloads on the network it makes. Both sides open in the same week.
Renti turns idle consumer GPUs into a distributed, energy-aware compute network. People rent out unused GPU power and earn automatically; Renti routes each workload to wherever energy is cheapest and cleanest.
Most consumer GPUs spend the overwhelming majority of their life idle.
Install the client, set the hours you don't need your machine, and it joins the network as a compute node.
Training runs, inference and rendering, each bringing its own environment and leaving nothing behind.
A data centre pays to bring energy to one fixed address; a node already stands where the energy is.
Cost per kilowatt-hour and carbon intensity are sampled continuously and workloads routed to the cheapest, cleanest nodes.
A price floor, protected hours and a thermal ceiling. Earnings settle weekly, itemised by job.