io.net: a decentralized GPU network for AI compute
Thread1,269 words
Bullish call$IO−96%since listing, moved against the callHow it played out ↓Breakdown of io.net's DePIN of clustered idle GPUs serving AI/ML workloads, its IO Engine and IOG Package, with network stats of 23.5K nodes, 23,094 GPUs, about $3K daily GMV and more than 5x Render's GPU supply.
Key claims
23,094 GPUs, more than 5x Render and 10x Filecoin
$250K GMV paid, about $3K per day
OpenAI spends $700K+ per day running ChatGPT on centralized cloud
AI + DePIN will be the most powerful narrative of the cycle 🤖
With BTC paving the way into the bull, leading AI plays are gearing up for runs that can do 100x
My main bet is the Internet of GPUs. The largest decentralized network of compute power for large-scale AI
Alpha-🧵
1/ In crypto, just one right play can change the game for you entirely.
The key to make that happen is staying on top of the leading narratives.
To identify a major play early is the goal, which means you have to DYOR.
I already did, so let me share some alpha.
2/ AI in conjunction with Decentralized Physical Infrastructure Networks (DePIN) will reshape the digital world.
DePIN will source and render idle compute power globally, thereby enabling highly performant AI/ML models to run inference on decentralized and secure infrastructure.
3/ This means scalable and decentralized AI.
Run on a decentralized backend consisting of app-specific, blockchain-based execution layers, built on high-spec DePIN architecture.
Accessible to anyone and you can be part of it (airdrop confirmed).
Are you ready, anon? 🪂 👀
4/ I'm talking technology that leading infra projects like @rendernetwork and @Filecoin are relying on for AI/ML exposure because they don't have the software / network architecture to do AI / ML workloads.
Let me introduce you to @ionet_official.
5/ Get a coffee and make yourself comfortable anon.
Today, we talk some ground-breaking tech.
As mentioned above, @ionet_official is a decentralized physical infrastructure network that sources GPUs globally.
These are then rented out to AI companies requiring compute power.
6/ So why does that matter?
Well, currently the compute requirements for AI apps are increasing 10x every 18 months (and growing rapidly).
This comes with a 3'100% annual increase in model training and inference cost for AI companies.
7/ This is due to a shortage on the GPU market.
If you can address that supply shortage and serve this hyper-growth market, you will capture a lot of value.
That's what @ionet_official does (and a lot more).
8/ The GPU shortage causes:
✦ Long lead times: Weeks to access GPUs in centralized cloud services
✦ Limited choice: Little choice over GPU hardware, location, security or latency
✦ High costs: OpenAI spends $700K+ per day running ChatGPT on centralized cloud servers
9/ @ionet_official fixes this.
But let's look at the "how" a bit closer.
Most of the GPU supply globally is underutilized.
It sits in sources like independent data centers, crypto mining farms, and consumer GPUs.
x.com/ionet_official/status/1763761145419079917
10/ @ionet_official clusters that idle GPU capacity from different sources and multiple locations.
This allows users (AI/ML companies) to access computing power at a fraction of the cost of traditional cloud providers and benefit from greater customizability.
11/ @ionet_official is purpose-built to support low latency & high demand use cases like AI/ML inference.
Through parallel processing @ionet_official can cluster tens of 1000s of distributed GPUs together instead of just a few (like most competitors).
x.com/ionet_official/status/1758511486408818908
12/ To demonstrate the power of @ionet_official's DePIN infra for AI companies, the team behind IO Network launched BC8.
A generative AI platform built on IO Network, each generated request has a Proof of Compute that proves on-chain which GPU in the network performed the job.
13/ Today @ionet_official products (e.g. Cloud) and BC8 are the only applications that can access the GPU network.
While providing compute to the network is permissionless, accessing the compute capacity and the compute layer (IO Engine) built on top of it, is currently not.
14/ These applications are examples of what is referred to as IO Elements.
IO Elements are products that are built on top of IO's decentralized network of GPUs.
15/ Soon, @ionet_official will release a dev-ready toolkit called IOG Package, which will enable anyone to build IO Elements.
This means that any other apps and businesses can build on top of IO Network's DePIN layer in a permissionless way.
The possibilities are limitless.
16/ So let's take a moment to digest this.
IO Network is the underlying DePIN of globally sourced GPUs.
IO Engine is the Compute Layer built on it.
IOG Package lets you build on that infra.
Ultimately allowing any app or business to build an IO Element on top of IO Network.
17/ But wait, what is IO Engine?
The IO Engine is the programmable compute layer built on IO Network.
Soon, this will turn into a fully decentralized blockchain network (a.k.a. IO Network), bootstrapped on top of the decentralized GPU network.
Governed by the IO token. 🪂 👀
18/ But that's not all.
The network of high performance nodes that is what underlies this stack, can be used to bootstrap independent decentralized networks that run arbitrary IO Elements on top of them.
A whole ecosystem of scalable, app-specific AI chains anon. 🤖 ✨
19/ The nodes that run these chains are the same GPUs that originally joined the network to earn revenue from running AI and ML jobs.
They can now also participate in the validation of the IO blockchain and other blockchains in the IO network.
20/ This will earn nodes validator incentives with layered economics on top of the revenue opportunity arising from providing compute to AI/ML companies.
21/ The prospect of revenue from serving the high-growth AI/ML market in addition to validator rewards is huge.
It increases the return on hardware investment and will catalyse adoption and enable @ionet_official's network of GPUs to grow rapidly.
x.com/ionet_official/status/1761886827571564799
22/ Already today there are over 23.5K nodes in the network.
Aside the decentralization, another benefit is that nodes in the the IO blockchain network will maintain comparably high hardware specs.
These hardware specs allow the IO-based chains to achieve millions of TPS.
23/ Do you see what my thesis amounts to here anon?
And this is not just your average AI "vaporware".
@ionet_official has found real-world adoption and is generating revenue as the market leader in a high-growth market already.
24/ @ionet_official in numbers means:
✦ $250K in GMV paid, ~$3K / day paid
✦ 28'221 compute hours served, about 325 compute hours served / day
✦ 1'762 clusters
✦ 23'500 worker nodes in the network
25/ It is by far the largest network in terms of GPU supply with:
✦ 23'094 total GPUs
✦ 1'627 total CPUs
This is >5x the amount of GPU on @rendernetwork.
This is >10x the amount of GPUs on @Filecoin.
26/ The global GPU market is inefficient.
High-value use cases in need of resources can't access those cost efficiently.
@ionet_official solves this and will become the DePIN backbone of an entire ecosystem of highly performant chains optimized for decentralized AI.
27/ This is pretty much as big as opportunities get anon.
And yes, there will be an airdrop. 🪂
Just connect your hardware to @ionet_official and provide GPU capacity to earn IO tokens and stablecoin revenue from companies renting your compute power.
28/ For Apple Chips, @ionet_official distributes up to 4x the rewards, so it might really be worth it anon.
Also, this is probably the only opportunity out there to farm a high-profile airdrop without having to put any capital at risk.
Kind of a no-brainer really. 🪂
29/ The age of AI is upon anon and it is probably the largest investment opportunity in human history.
NFA but at least don't fade the AI + DePIN narrative and DYOR anon.
I will definitely keep an eye on @ionet_official. 👀
That's a wrap! I hope you've found this thread helpful! 🫡
Make sure to follow me @zerokn0wledge_ and @ionet_official to stay up to date! 🤖 ✨
Like/Repost of the below post is also highly appreciated! 🙏👇
21 images on X
How the call played out · since 06 Mar 2024
Prices as of 23 Oct 2026This piece takes a view on $IO, so that move is read against the call.
Bullish on IO as the main AI and DePIN bet, with far more GPUs than Render and Filecoin.
−96%
Moved against the call · since it listed on 12 Jun 2024, 3 months after this piece
- At listing
- $4.17
- Now
- $0.158
- vs BTC
- −97%
- After 1 month
- −41%
- After 3 months
- −60%
- After 1 year
- −81%
- High since
- +36% 13 Jun 2024
- Low since
- −98% 06 Apr 2026
Since then
Zero has written 51 more pieces on io.net and $IO since this one.
- io.net IDE tokenomics: #1 revenue DePIN at 5.3x MC/ARRBullish call on $IO
- DeAI as crypto's endgame: compute, agents and autonomous economiesOn io.net
- Why decentralized compute solves the startup GPU cost squeezeOn io.net
- Big compute is cracking: summary of the DeAI case against centralized cloudOn io.net
Daily closes in USD from DefiLlama and Yahoo Finance. Only a bullish or bearish call in a thesis or market take is tracked; explainers, lists, maps and mentions are neutral context. Not financial advice.