Startups Launch 'Airbnb for AI Inference' to Rent Your Idle Gaming PC for Passive Income

Startups Turn Idle Hardware Into AI Infrastructure

Computer hobbyists with gaming rigs and home servers can now monetize spare computing power through emerging platforms that connect idle devices to AI companies seeking resources for AI inference. Abu Dhabi-based Far Labs is launching its Far AI platform in the coming weeks, while Austin-based Evolving Edge is currently in open beta1

1

. Both startups position themselves as an "Airbnb for AI inference," targeting AI workloads on consumer-grade hardware rather than relying solely on massive data centers2

2

.

The concept builds on volunteer-based distributed computing projects like SETI@Home, which ran from 1999 to 2020, but now commercial companies are willing to pay for this decentralized AI infrastructure1

1

. John Federico, founder and CEO of Evolving Edge, argues that traditional data centers extract resources from communities without returning value. "The compute power is out there. If you can orchestrate it, then you're actually adding value to those communities directly," Federico told IEEE Spectrum1

1

.

How Renting Compute Works for Hosts

Signing up to rent your idle gaming PC for AI tasks requires minimal effort. Users install an application, set a schedule for when their machine is available, and the platform handles the rest1

1

. Evolving Edge uses Ray, an open-source framework employed inside conventional data centers, to distribute jobs across the network2

2

. Far Labs employs a proprietary scheduler that splits models into pieces across multiple machines, then reassembles partial outputs into responses2

2

.

Far AI claims latency of 100 ms or less, positioning itself for real-time applications2

2

. These platforms focus on smaller, mostly open-source AI models rather than frontier-scale workloads that still require massive data center infrastructure1

1

. They join established players like Utah-based Salad, which lists more than 60,000 daily active consumer GPUs on its network2

2

.

Security Through Sandboxing and Encryption

Allowing external workloads onto personal machines raises legitimate concerns about malicious code and unauthorized access to local files. Both Far Labs and Evolving Edge address these risks through isolation techniques. AI inference runs as a sandboxed workload with encrypted communication and explicit limits on GPU, CPU, memory, storage, and network resources2

2

.

Ilman Shazhaev, founder and CEO of Far Labs, explains the company's software follows a "least privilege" principle, granting both host and user minimal access needed to complete tasks1

1

. Customers receive no direct access to host machines, and providers can inspect resource use, pause nodes, revoke access, and remove software at any time1

1

. Evolving Edge has open-sourced its node software so hosts can audit exactly what runs on their hardware1

1

2

2

.

Protection works bidirectionally. Workloads are segmented with only minimum required information exposed to individual nodes. Sensitive enterprise workloads can be restricted to controlled hardware rather than routed through consumer devices1

1

.

Profitability Questions Remain Unanswered

While the promise of passive income sounds appealing, the economics of renting out spare compute remain murky. Salad sells consumer-GPU compute starting at $0.02 per hour, but hosts only receive a fraction after the platform takes its cut2

2

. A 2021 analysis estimated Salad generated roughly $3.6 million from users' PCs while distributing $500,000 in rewards—about 14 cents on the dollar for hardware owners2

2

.

Electricity costs further erode potential earnings. An RTX 4090 draws 350W to 450W under sustained load, translating to roughly $40 to $50 monthly at $0.15 per kWh if running continuously2

2

. Systems earning less than their electricity costs effectively pay for the privilege of participating. Neither Far Labs nor Evolving Edge has disclosed payout rates, leaving potential hosts without concrete profitability data2

2

.

Distributed Networks Promise Resilience

Beyond economics, distributed computing offers resilience advantages over centralized clouds. Federico cited the AWS outage last October that left Internet-connected smart beds stuck upright, arguing his network could lose 100 of 250,000 nodes "and it wouldn't matter"2

2

. This redundancy positions decentralized AI infrastructure as potentially more reliable for certain applications.

However, distributed AI compute faces credibility challenges. A June research preprint claimed that Pearl, a blockchain marketed as converting cryptocurrency mining into AI work, ran the equivalent of 320,000 RTX 3090-class GPUs on random matrix math while producing no useful AI computation2

2

. Far Labs and Evolving Edge route real customer inference jobs rather than token rewards, distinguishing them from such schemes, but their cost and low-latency figures remain unsubstantiated until networks operate at scale2

2

.

Watch whether these platforms can prove genuine profitability for hosts, maintain security as networks scale, and deliver the performance claims that differentiate them from traditional data centers. The success of renting out spare compute will ultimately depend on transparent economics and demonstrated reliability at scale.