IntoTAO
Back to Subnets
Nodexo

Nodexo

SN27

Rent GPUs on demand from a global marketplace of compute providers

Nodexo is a GPU compute marketplace on Bittensor, rebuilding from the ground up with a focus on trustless hardware verification. The idea is simple: let anyone rent or supply GPU capacity, with validators continuously benchmarking providers to keep quality honest.

// GPU compute, verified on-chain.

Price0.00000-5.14% 7d
Holders0
Momentum0.0 / 100Strong
// WHAT_IS_THIS

Nodexo is a decentralized marketplace where GPU providers can list their hardware and buyers can access compute capacity, all without going through AWS, Google Cloud, or any centralized platform. Running on Bittensor's Subnet 27, it uses a network of validators to continuously test GPU performance and ensure providers are delivering what they claim. The subnet is a relaunch of NI Compute, an earlier Neural Internet project on the same netuid.

The simple version: It's like renting a GPU from someone else's machine, except the network independently verifies that machine is real and performs as advertised, before any payment clears.

Centralized equivalent: AWS EC2 GPU instances, Lambda Labs, or CoreWeave.

How it works:

  • Miners supply GPU compute capacity, making their hardware available through the network
  • Validators continuously benchmark and score providers using Proof of GPU mechanisms, verifying hardware quality and real performance
1,686holders|N/Acommits|0social mentions this week
Buy Nodexo on TaoSwap
Research snapshot from April 30, 2026. Live metrics are in the sidebar.
// WHY_THIS_MATTERS
  • The problem it solves: According to the team, access to high-performance compute is bottlenecked by centralized providers, resulting in scarcity, high costs, and single points of failure. Smaller developers get priced out or locked into specific platforms.
  • The opportunity: GPU demand for AI training and inference is not going down. A permissionless marketplace where anyone can supply or consume compute, at market prices and without a platform intermediary, addresses a structural gap the centralized cloud has no incentive to fix.
  • The Bittensor advantage: The trust problem in decentralized GPU rental is hard. Who verifies the hardware is real? Bittensor's incentive layer assigns that job to validators, who get rewarded for accurate scoring. The result is a self-enforcing quality guarantee without a central referee.
  • Traction signals: The pool holds roughly 8,400 TAO with 15.7% root proportion, meaning most of the depth is organic. Staking inflows were positive over the past seven days. The subnet is mid-migration under Nodexo v2.0, and active miners are currently at zero while the new system comes online.

// FULL_ANALYSIS

Category: Inference and Compute | Centralized Competitor: AWS EC2 (GPU instances), Lambda Labs, CoreWeave

The GPU compute market is enormous, and centralized providers control most of it. A handful of hyperscalers and specialty clouds dominate access to NVIDIA H100s and A100s, setting prices and controlling who gets in. Nodexo, formerly the NI Compute subnet from Neural Internet, is attempting to build a decentralized alternative where GPU owners earn by renting capacity and buyers get transparent, benchmarked access to hardware.

Mechanism:

Per the official Nodexo description, the subnet establishes a permissionless GPU marketplace where anyone can supply or consume compute. Validators run continuous benchmarking against registered providers using Proof of GPU, verifying actual hardware performance rather than trusting provider claims. Miners who pass quality checks receive rewards; those who fail are scored accordingly. Community sources reference this framework as "PoGv3," though detailed technical documentation is not publicly available at the time of writing.

The team completed a migration to Nodexo v2.0 earlier this year, with validators now running on the updated system. A v2.1.0 release is in progress but has been delayed. Active miners currently sit at zero, reflecting the transition period rather than a permanent operational state.

The subnet was registered at block 1,727,132, one of the earlier registrations in the Bittensor network, giving the Neural Internet team a long track record on this netuid. At the time of this article, price is 0.00447 TAO, up 18.7% over 30 days. Market cap is roughly 22,229 TAO and pool depth is 8,441 TAO. With root_prop at 15.7%, the pool is mostly organically staked. Twenty-four-hour volume reached 2,173 TAO, high relative to market cap, indicating active price discovery. Net staking flows over seven days were positive at around 290 TAO.

The Gini coefficient for the top 100 addresses is 0.50, one of the more evenly distributed subnets in the ecosystem. Nakamoto coefficient sits at 23, meaning at least 23 independent parties would need to collude to influence consensus. Both figures suggest stake is reasonably spread across participants.


// RISK_FACTORS
Risks assessed as of April 30, 2026. Conditions may have changed.
  • Zero active miners: The subnet currently has no active miners. Nodexo is mid-migration to v2.0, so compute work is not being produced today. Until miner onboarding under the new system is live, emission rewards flow to the pool but no actual compute marketplace is operating.
  • Delayed v2.1.0 release: The team has postponed the v2.1.0 rollout, with no confirmed date. Building a trustless GPU marketplace is technically demanding, and execution delays are a recurring risk in this category.
  • Private codebase: The main repository is not publicly accessible. Community verification of the protocol design and audit by outside developers is limited until the code is opened up.
  • Competition: Chutes (SN64) is the leading compute-layer subnet on Bittensor, with an established position and active emissions. Nodexo needs to differentiate on hardware trust guarantees, price, or specific GPU inventory to carve out a distinct position.
// LIVE_DATA
Price0.00000 TAO
24h+0.36%
7d-5.14%
30d-13.56%
Market Cap0.00 TAO
Emission0.00%
Liquidity8.2K TAO
Holders0