Decentralized GPU network io.net has reported more than 30 million compute hours delivered as it marks its third anniversary, alongside more than $24 million in accumulated on-chain revenue. The project’s official blog also placed annualized on-chain revenue at approximately $20 million, pointing to sustained infrastructure demand across its first three years of operation.
The network aggregates GPU capacity from independent data centers, crypto miners and partner providers across more than 130 countries. Its core pitch is on-demand compute for AI and machine learning workloads, positioning io.net as a decentralized alternative to centralized cloud infrastructure.
OpenRouter Integration Adds Inference Scale
Anniversary coverage also highlighted io.net’s integration with OpenRouter, where the network is said to process up to four billion AI inference tokens daily. That activity suggests the platform is gaining traction in live AI inference, although io.net has not broken down how its 30 million compute hours are divided between training, inference and other workload categories.
June 2023: an idea.
June 2024: product-market fit.
June 2025: enterprise scale.
June 2026: sustainable tokenomics.
Three years of building. 30M + compute hours. $24M + on chain revenue. The world's largest decentralized GPU network.
This is what DePIN looks like when it… pic.twitter.com/gFzXzU7cUD
— io.net (@ionet) June 13, 2026
The milestone comes as demand for AI compute continues to expand across both Web2 and crypto-native markets. For io.net, the strategic opportunity is clear: turn fragmented GPU supply into a usable commercial network for developers, enterprises and AI agents that need flexible compute access.
Alongside the usage figures, io.net launched its Incentive Dynamic Engine, or IDE. The mechanism ties $IO token burns directly to verified customer revenue, shifting the project’s tokenomics narrative away from emissions alone and toward measurable commercial activity.
The project said it expects to burn at least 12 million IO tokens over the next year, with the first scheduled burn set for June 11. Network leadership framed the system as a way to connect token supply mechanics with real demand rather than purely speculative incentive cycles.
Enterprise Revenue Becomes Key Test
Enterprise activity is also becoming a larger part of io.net’s growth story. The project highlighted an $8 million commercial agreement that contributes about $650,000 in monthly on-chain network earnings, with additional enterprise deals said to be in advanced negotiation.
Those figures support the case that io.net is moving beyond marketplace experimentation into revenue-producing infrastructure. Still, the timing and scale of pending enterprise agreements have not been fully detailed, leaving future revenue growth dependent on execution and client onboarding.
The platform is also continuing work on its Agent Cloud product, which is designed to let autonomous AI agents procure infrastructure directly. That development would extend io.net’s role from compute marketplace to machine-native infrastructure layer, where agents can source resources without traditional manual procurement workflows.
The main limitation in the latest update is the lack of a short-term activity breakdown. io.net reported strong cumulative metrics, but the operational change over the most recent 48-hour period was not specified.
For now, the anniversary data gives io.net a stronger commercial narrative in the decentralized AI infrastructure sector. The next test is whether compute usage, enterprise revenue and revenue-linked burns continue scaling as the Incentive Dynamic Engine processes more customer-driven activity.
