Cash In on the AI Boom by Renting Out Your Spare Compute

Several companies are building platforms that pay owners of home servers, gaming computers, and laptops for spare computing power used to run AI inference, the process of using a pretrained model to respond to queries.

Ilman Shazhaev, founder and CEO of Abu Dhabi-based Far Labs, described the concept as "Imagine Uber or Airbnb, but for AI-inference computing tasks." His company is launching its platform, Far AI, in the coming weeks. John Federico, founder and CEO of Austin-based Evolving Edge, said his company is currently in open beta. Bless Network, Salad, and Gradient have started providing similar platforms over the last year.

The approach runs inference on smaller, mostly open-source models, distributed across computing power already present in homes and small businesses, with owners compensated. Shazhaev said Far Labs splits a model into pieces distributed across devices, with an orchestrator and load balancer managing task flow. Evolving Edge uses the open-source tool Ray to split inference tasks across multiple GPUs or CPUs.

On security, Federico said Evolving Edge open-sourced its scheduling software so anyone can inspect what it does. Shazhaev said Far Labs designs around "least privilege," running inference as an isolated workload with authenticated, encrypted communication and limits on GPU, CPU, memory, storage, and network resources. Customers do not get arbitrary access to host machines, and providers can inspect resource use, pause a node, revoke access, and remove the software. Workloads are segmented, and sensitive enterprise workloads can be restricted to controlled hardware.

Shazhaev said massive data centers still have advantages, including top-of-the-line GPUs and CPUs, high-speed networking, and sophisticated cooling, and that running comparable tasks on limited user devices is difficult. Federico said large state-of-the-art models are an issue, but that companies fine-tuning open-source models for specific tasks do not need comparable resources.

Shazhaev claims the distributed approach is cheaper because there is no capital expenditure, and more reliable due to decentralization. He said Far Labs claims latency of 100 milliseconds or less. He cited OpenAI's reported US $30 billion in revenue last year and an $8 billion loss at the close of the financial year, attributing the official reason to the high cost of inference.

Federico said the resiliency would benefit applications including smart cities, environmental sensors, and autonomous vehicles, and referenced an Amazon Web Services outage in 2026 that left smart beds stuck upright. He said a network of 250,000 nodes could lose 100 without consequence.