ByteLeap
SN128Cloud computing and blockchain infrastructure bridging traditional cloud with decentralised networks
Distributed cloud compute on Bittensor's 128th subnet, where hardware providers earn through active leases rather than idle benchmarks.
// Cloud compute, decentralised
ByteLeap is a distributed compute network built on Bittensor that turns hardware into a cloud marketplace. Hardware providers run software on their machines as "workers," which are aggregated by Bittensor miners to offer compute resources for lease to paying customers.
The simple version: It's like a decentralized AWS EC2, where hardware owners make their machines available for cloud compute rentals and earn rewards when those machines are actively leased.
Centralized equivalent: AWS EC2, Google Cloud Compute Engine.
How it works:
- Workers run hardware monitoring software and execute compute tasks via a VM gateway session and Libvirt virtualization, connecting to miners via WebSocket
- Miners aggregate up to 100 workers each, handle resource reporting to the Bittensor network, and route challenges to any unleased machines
- Validators coordinate the network by creating cryptographic challenges, validating scores, and setting weights that determine emission distribution
- The problem it solves: Cloud compute is dominated by a handful of providers. Idle hardware sits underutilized worldwide while buyers pay premium prices to centralized incumbents.
- The opportunity: A decentralized compute marketplace could lower costs for buyers and create income streams for hardware owners who would otherwise leave capacity unused.
- The Bittensor advantage: Emission-based incentives coordinate hardware providers without central orchestration, and the AMM creates transparent price discovery for the subnet's alpha token.
- Traction signals: 42 active miners are registered on the subnet. The lease-based architecture is defined in the public repository, though public evidence of live leasing volume is limited.
Category: Inference and Compute | Centralized Competitor: AWS EC2, Google Cloud, Akash Network
The decentralized cloud compute space is active and competitive. ByteLeap enters with a distinctive three-tier architecture that separates concerns cleanly: validators coordinate, miners aggregate, and workers execute. The defining characteristic is its scoring model: unlike compute subnets that score primarily on benchmark performance, ByteLeap weights active compute leases at 100% of the miner score. Idle hardware earns nothing.
Mechanism:
According to the project's public miner repository, ByteLeap operates a three-tier system. Workers are separate machines running hardware monitoring software and executing compute tasks via a VM gateway (VMGW) session using Libvirt for virtualization. Each miner can aggregate up to 100 workers via WebSocket, serving as the interface between the Bittensor network and the underlying hardware.
Scoring has two components: lease revenue drives the primary score, with workers under active compute rentals earning credit and idle workers scoring zero. An availability multiplier based on 169-hour online presence supplements the base. Validators issue cryptographic challenges using a Merkle tree verification system, but those challenges target only unleased workers, reinforcing active leasing as the main earnings path.
The subnet currently holds a roughly 0.31% share of total network emissions, consistent with its Taoflow position reflecting net staking flows. The pool carries approximately 4,220 TAO in depth, with root proportion near 22%, meaning about 78% of pool depth comes from organic demand rather than protocol subsidy. Price is up around 10% over the past 7 days, though the 30-day trend remains slightly negative. Net TAO inflow over the last week was approximately 197 TAO.
Development activity is limited: one contributor, 12 commits, and the last recorded push was January 2026. A separate worker repository is referenced in the documentation but was not publicly accessible at the time of writing, so the full scope of the codebase is unclear from public sources alone.
- Development Stagnation: The public miner repository has 12 commits from a single contributor, with the last push in January 2026. A compute network needs ongoing maintenance and this pace is a material concern.
- Concentration: A Gini coefficient of 0.85 across the top 100 stake positions and a Nakamoto coefficient of 3 indicate highly concentrated stake. Few positions could significantly influence pool dynamics or validator weight-setting.
- Execution: ByteLeap is early-stage. The public repository covers only the miner component. A separate worker repository is referenced but not publicly visible, limiting outside auditability. Live leasing volume is not verifiable from public sources.
- Competition: Multiple Bittensor subnets operate in decentralized compute, and ByteLeap competes with established centralized providers off-chain as well.