Supplying a GPU to io.net pays in two streams: hourly block rewards in IO for keeping a verified device online, and job payments whenever a customer rents your hardware for AI work. Block rewards arrive whether or not the GPU is doing paid work, which makes io.net one of the few ways idle silicon earns a yield just for being available.
How much that adds up to depends on your hardware class, uptime, and the network's declining emissions curve, so this guide walks through the mechanics first and the honest economics second.
What a Worker Earns
Block rewards come from the 300 million IO emissions pool and are distributed hourly to devices that pass verification, weighted by hardware model, stake, and reliability, as laid out in io.net's block rewards documentation. They function as an availability payment: the network pays to keep capacity warm.
Job payments land on top when a renter books your device into a cluster for training, fine-tuning, or inference. High-demand cards such as H100s, A100s, and RTX 4090s naturally match more rental queries, so they capture proportionally more job revenue than mid-range hardware sitting in the same marketplace.
Hardware That Qualifies and What It Needs
The network accepts a wide band of devices, from data-center accelerators down to recent consumer GPUs, along with CPUs for lighter workloads. Whatever you connect must clear three bars. First, a 12-hour onboarding stress test that benchmarks the device and its connection. Second, ongoing availability of at least five hours per day to stay reward-eligible. Third, collateral: every chip requires an IO stake, starting at 200 IO and scaling with the model's earning multiplier, which you can fund yourself or share with co-stakers.
Bandwidth and stability matter as much as raw compute. The scheduler favors devices that behave like infrastructure, and the reliability score (0 to 100) that builds from your uptime history follows the device around the marketplace.
Setting Up an IO Worker, Step by Step
- Create an io.net account and open the IO Worker dashboard.
- Install the worker client on your machine (containerized installers cover Windows, Linux, and macOS).
- Let the device run its 12-hour stress test and hardware verification.
- Fund the required stake for your GPU model, alone or through a co-staking offer.
- Bring the device online and confirm it appears as verified in the dashboard.
- Monitor uptime, reliability score, and hourly accruals, and withdraw rewards to your Solana wallet as they settle.
Nothing in the list is technically demanding for anyone comfortable installing software, and a single machine can be onboarded in a day, with the stress test as the longest wait.
The Variables That Decide Your Payout
| Variable | Why it moves your earnings |
|---|---|
| GPU model multiplier | Higher-end cards earn more per hour and attract more jobs |
| Uptime and reliability score | Low scores cut block rewards and repel renters |
| Stake size | Rewards weight toward properly collateralized devices |
| Bandwidth and location | Faster, well-placed connections win cluster selection |
| Network demand | Job payments rise and fall with rented capacity |
| Emissions curve | The hourly pool shrinks roughly 1 percent monthly |
The last row is the one workers most often ignore. Because emissions step down every month on the 20-year schedule described in our IO tokenomics guide, the same device earns fewer IO per hour next year than this year unless demand-driven adjustments or job payments make up the difference.
Is Running a Worker Worth It in 2026?
The sober math: you earn a volatile token against fixed electricity and bandwidth costs. With IO near $0.16 as of July 2026, block rewards on consumer hardware are measured in single-digit dollars per day at best, so the position only makes sense if you either believe in IO's price recovering or already own hardware that would otherwise sit idle. Workers are, functionally, dollar-cost-averaging into IO with compute instead of cash.
Competition also concentrates rewards. Only several thousand devices pass daily verification out of hundreds of thousands registered, per Messari's network overview, and those verified devices split the pool. Serious operators treat it like a small business: enterprise cards, redundant internet, high reliability scores. Casual suppliers with idle rigs still earn, just modestly, and some spread hardware across networks, weighing io.net against alternatives the way our io.net vs Render comparison does.
Frequently Asked Questions
How much can you earn with io.net?
Earnings vary too widely for a single number: hardware model, uptime, stake, and network demand all weight the hourly block reward, and job payments depend on rentals. The marketplace shows projected weekly earnings per device class from trailing seven-day data, which is the most realistic estimate available before you commit.
Do you get paid if your GPU is never rented?
Yes. Block rewards pay hourly for verified availability regardless of rentals, funded by the emissions pool. Job payments are separate and only arrive when a customer books your device, which is why availability alone yields something but rented hardware yields meaningfully more.
Can you run an io.net worker on a gaming PC?
Yes, provided the GPU is a supported model and the machine can meet the five-hour daily uptime and stress-test requirements. Gaming rigs with recent RTX cards are common on the network, though their earnings sit well below data-center hardware.
Turning Spare Compute Into a Position
Running an IO worker converts electricity and idle hardware into a token position, paid hourly, with the network's emissions schedule and your own reliability as the main levers. It will rarely beat a salary, but as a way to accumulate exposure to decentralized AI infrastructure without buying tokens outright, it is one of the more concrete offers in crypto.
However you accumulate, the exit and the hedge live in the same place: sell or add on the IO spot market, manage swings with IO perpetual futures, and keep learning token mechanics in Crypto in a Minute.