One neocloud alone has $35.1bn in debt.
That’s enough to fund 2 million+ GPU-years of H100 compute at https://t.co/DCp3kvjsTY.
Debt buys hardware.
Access builds companies.
@ionet is built for access. https://t.co/ulwNKgGfZ1
io.net
Decentralized GPU compute network aggregating over 30,000 chips on Solana for AI and machine learning
3.8/ 5
Our score
Good
Services
Developer tooling
Chains supported
1
Free tier
Yes
Pricing from
$0.18/hour
Enterprise SLA
Yes
Token
IO
What is io.net?
io.net is a decentralized physical infrastructure network (DePIN) that aggregates graphics processing units from independent data centers, crypto miners, and decentralized networks like Filecoin and Render. Built on Solana, the platform coordinates compute clusters for artificial intelligence and machine learning teams, providing access to over 30,000 GPUs.
Leadership and origin
The project originated in February 2023 at the Solana Hackathon and the Solana Austin Hacker House. Headquartered in North America, io.net is led by Chief Executive Officer Tory Green and Chief Technology Officer Gaurav Sharma. The team built the network to pool underutilized global compute hardware into unified clusters for distributed computing jobs.
Platform architecture and hardware access
io.net operates two primary offerings: io.net Cloud and IO Intelligence. io.net Cloud handles resource orchestration, cluster scheduling, and automated failover for Python-based workloads, including batch inference, model serving, parallel model training, and reinforcement learning. IO Intelligence provides access to open-source models and agent workflows, with developer tiers receiving $150 in monthly usage credits refreshed every eight hours. Compute clustering spans hardware profiles ranging from consumer chips to enterprise accelerators.
Pricing and economics
Rental pricing on the network depends on hardware availability and tier. Stated hourly rates include RTX 4090 units at $0.18, RTX 3090 units at $0.28, A100 40GB chips at $1.20, A100 80GB chips at $1.49, and H100 PCIe models at $1.49, rising to $2.20 per hour for H100 SXM configurations. New users receive $100 in initial free credits. The platform charges a 10% to 20% network transaction fee. Under its native economic model driven by the IO token, the network introduced the Incentive Dynamic Engine in June 2026, which permanently burns at least 50% of post-payout network revenue.
Milestones and trade-offs
As of June 2026, io.net processed more than 4 billion daily AI inference tokens and secured an $8 million enterprise contract generating roughly $650,000 in monthly on-chain earnings. The network maintains a 99.9% uptime service level agreement using distributed redundancy. Key trade-offs include reliance on distributed, third-party node providers across consumer and enterprise environments, which can introduce latency variations compared to co-located hyperscaler data centers. Workloads also depend on Solana on-chain coordination and token incentives governing cluster supply and hardware operator stability.
Founders
Tory GreenCEO
Gaurav SharmaCTO
Best for:Machine learning teams running batch inferenceDevelopers seeking discounted GPU cluster rentalsAI startups training distributed models
Pros and cons
Pros
- Hourly GPU rental rates start at $0.18 for RTX 4090 chips
- Maintains a 99.9% uptime SLA using automated failover
- Provides new accounts with $100 in trial credits
- High-end hardware access includes NVIDIA H100 and A100 configurations
- Developer plan supplies $150 in monthly IO Intelligence usage credits
Cons
- Platform transaction fees range from 10% to 20%
- Performance relies on decentralized node providers with varied latency
- Network settlements and incentives depend on Solana and the IO token
- Enterprise custom pricing requires a minimum 100 GPU commitment or $10,000 monthly spend
Our review
3.8/5
Good
io.net offers aggressive GPU pricing up to 70% below AWS for machine learning workloads, but relies on a decentralized provider pool rather than dedicated hyperscaler data centers.
io.net operates a decentralized physical infrastructure network (DePIN) on Solana that pools underutilized graphics cards from independent data centers, crypto miners, and partner networks like Render and Filecoin. Originally started in February 2023 at the Solana Hackathon, the platform targets AI and machine learning engineering teams who need on-demand parallel compute without the weeks-long provisioning queues common at major cloud providers.
Compute pricing is the primary draw. io.net advertises costs up to 70% lower than AWS, Azure, and Google Cloud, charging hourly rates such as $0.18 for an RTX 4090, $1.20 for an A100 40GB, and $1.49 to $2.20 for H100 hardware. The platform takes a 10% to 20% cut on compute transactions while providing a 99.9% uptime SLA supported by automated health monitoring and dynamic failover across nodes. New users receive $100 in initial credits, and developers on IO Intelligence receive $150 in monthly credits refreshed every eight hours.
Operating metrics show measurable commercial adoption. By June 2026, the network processed over 4 billion daily AI inference tokens and secured an $8 million enterprise contract generating approximately $650,000 in monthly on-chain earnings. Under its Incentive Dynamic Engine introduced in June 2026, the project committed to burning at least 50% of post-payout network revenue in native IO tokens, targeting at least 12 million IO burned across twelve months.
The trade-offs stem directly from its decentralized model. While io.net automates scheduling, hyperparameter tuning, and distributed orchestration for Python workloads, individual cluster nodes are geographically distributed across disparate operators. Teams requiring strict data residency guarantees, custom bare-metal physical isolation, or native multi-chain RPC infrastructure will find centralized hyperscalers or specialized node infrastructure better suited to their architecture.
Specs
| Services | Developer tooling · unverified |
|---|---|
| Chains supported | 1 · unverified |
| Free tier | Yes · unverified |
| Pricing from | $0.18/hour · unverified |
| Enterprise SLA | Yes · unverified |
| Open source | Yes · unverified |
| Token | IO · unverified |
| Uptime SLA | 99.9% · unverified |
| Free tier limits | $100 in free credits for new users · unverified |
| Server regions | 130+ countries · unverified |
Data from the provider's website and public sources. Fees and availability change; confirm on the official site before you sign up.
Key features
Distributed GPU clustering
Orchestrates on-demand clusters across more than 30,000 GPUs from independent data centers, crypto mining rigs, and networks like Render and Filecoin.
On-demand model inference
Runs parallel batch inference across distributed GPUs, supporting model exports directly to shared object storage.
Parallel model training
Partitions Python and machine learning workloads across distributed hardware nodes using data and model parallelism libraries.
IO Intelligence platform
Supplies developer APIs, curated open-source AI models, and vector database tools, including $150 in monthly credits refreshed every eight hours.
Dynamic hardware pricing
Offers hourly GPU rates starting at $0.18 for RTX 4090, $1.20 for A100 40GB, and $1.49 for H100 PCIe chips.
Uptime failover orchestration
Maintains a 99.9% uptime SLA through health checks, automated cluster scheduling, and failover across decentralized hardware.
Inside io.net
io.net Cloud
A decentralized computing network that lets developers deploy on-demand GPU and CPU clusters for batch inference, model training, hyperparameter tuning, and reinforcement learning.
IO Intelligence
A developer platform that provides access to open-source AI models, serverless inference workflows, vector databases, and agent orchestration tools.
io.net screenshots
What users say about io.net
Io.net receives praise for providing significantly cheaper GPU compute compared to centralized cloud providers like AWS. On the negative side, privacy-focused developers criticize the platform for content filtering, restrictive terms of service, and potential data privacy risks.
In their own words
“I use IO.net VMs for my model training jobs. I generally opt for US based H100 VMs, and get them for around 2-2.5$ per hour per gpu. For a 4GPU workload for 4-5 days, cost comes out to be around 1000 bucks.”
“If you're using AWS or Google Cloud for H100s, you're likely paying a 5x premium. There's a massive shift happening toward DePIN, where you can rent underutilised GPUs for about 10-20% of the cost of the big providers. It’s basically the 'Airbnb' model but for computer chips.”
“For privacy, on openrouter: Venice.ai, DeepInfra, AtlasCloud, Parasail, Chutes, Groq, Inceptron, Mara, Wafer, AkashML, Decart, SambaNova, and Phala. These all have robust ToS with no mention of content retention or filtering, or explicit note that they don't.”
Based on 112 Reddit posts and comments from the last 12 months, checked Oct 21, 2026. These are posts by users on other sites, not reviews submitted on CryptoReview.
User reviews
No user reviews yet. Have you used io.net? Share your experience below.
FAQ
Who founded and runs io.net?
io.net was founded out of the February 2023 Solana Hackathon, led by Chief Executive Officer Tory Green and Chief Technology Officer Gaurav Sharma.
What are the rental rates and platform fees on io.net?
Hardware rentals run at rates such as $0.18 per hour for an RTX 4090, $1.20 for an A100 40GB, and $1.49 to $2.20 for H100 GPUs. io.net retains a 10% to 20% platform fee on transactions.
Does io.net have a native cryptocurrency token?
Yes, io.net uses the native IO token on Solana to facilitate network payments and rewards. Under the Incentive Dynamic Engine, at least 50% of post-payout network revenue in IO is permanently burned.
What chains does io.net support?
io.net is built on the Solana blockchain for its token settlement and infrastructure coordination.
What uptime SLA does io.net provide?
io.net provides a 99.9% uptime SLA, using automated health checks and failover mechanisms across distributed hardware providers.
Latest from io.net on X
You came to rent a GPU.
You ended up funding the whole cloud.
Infrastructure overhead. Managed service premiums. Fees to move your own data.
Every extra dollar is budget you can’t spend building.
Rent the GPU and skip the rest with @ionet.
https://t.co/ypcQzdG1A2 https://t.co/hUBtOVyS6m
You ended up funding the whole cloud.
Infrastructure overhead. Managed service premiums. Fees to move your own data.
Every extra dollar is budget you can’t spend building.
Rent the GPU and skip the rest with @ionet.
https://t.co/ypcQzdG1A2 https://t.co/hUBtOVyS6m
Amazon, @Google and @Meta burned $45 billion in cash last quarter.
A year ago, they generated $36 billion.
Owning AI infrastructure is getting expensive.
Accessing it doesn’t have to be. @ionet
https://t.co/rLiPMgRPiX https://t.co/xNu71gpeiG
A year ago, they generated $36 billion.
Owning AI infrastructure is getting expensive.
Accessing it doesn’t have to be. @ionet
https://t.co/rLiPMgRPiX https://t.co/xNu71gpeiG
AI is about to compete with everyone for compute.
Jack Collier, our Chief Growth Officer, is at Hong Kong FinTech Week on Nov 3 to talk about what that means for finance, builders and infrastructure.
“When AI Sweeps Over the Infrastructure”
1:35 pm - Hall 5FG https://t.co/zMCFiaoAyM
Jack Collier, our Chief Growth Officer, is at Hong Kong FinTech Week on Nov 3 to talk about what that means for finance, builders and infrastructure.
“When AI Sweeps Over the Infrastructure”
1:35 pm - Hall 5FG https://t.co/zMCFiaoAyM
AI lets you build more than you should ship.
@ionet’s Head of Product @Raj_TheBUIDLer joins the @theaisho to talk open vs frontier models, and when an agent is actually worth its cost.
And there's coffee analogy that explains everything.
@ionet’s Head of Product @Raj_TheBUIDLer joins the @theaisho to talk open vs frontier models, and when an agent is actually worth its cost.
And there's coffee analogy that explains everything.
Recent public posts from @ionet, shown as published. Not reviewed or endorsed by CryptoReview.
Alternatives to io.net
Open-source smart contract libraries and enterprise security tooling backing over $38T in transfers
Developer toolingFree tier10 chains supported
Blockchain infrastructure platform supporting 100+ chains with a free 30M-compute-unit tier
RPC / nodes · Data APIs +3Free tier100 chains supported
Managed blockchain infrastructure across 70+ chains with a free 3M-request monthly tier
RPC / nodes · IndexingFree tier70 chains supported
EVM development and simulation platform supporting transaction debugging and testing across 50+ chains
Developer toolingFree tier50 chains supported
Open-source distributed ledger foundation hosting Fabric, Besu, and Cacti under the Linux Foundation
RPC / nodes · Developer toolingFree tier2 chains supported
Blockchain infrastructure platform providing RPC nodes and data streams across 76 networks
RPC / nodes · Data APIsFree tier76 chains supported