blockmachine
SN19Self-optimising blockchain infrastructure that improves itself automatically
RPC endpoints are the unglamorous backbone of every blockchain app. Blockmachine puts a competitive marketplace around them, built on Bittensor's incentive layer.
// Decentralized RPC infrastructure, incentivized
Blockmachine (SN19) is a decentralized marketplace for blockchain RPC infrastructure on Bittensor. Independent node operators run Bittensor subtensor nodes, serve incoming RPC requests, and earn emissions proportional to the work they deliver.
The simple version: Think of it like a decentralized Infura or Alchemy. Instead of one company running the servers that power wallets, block explorers, and dapps, Blockmachine routes those requests across a network of independent node operators who compete on price and quality.
Centralized equivalent: Infura, Alchemy, QuickNode.
How it works:
- Miners run Bittensor subtensor nodes (lite or archive) behind authenticated gateways, set a price per Compute Unit, and serve incoming RPC requests
- Validators pull gateway logs from S3, re-execute queries against independent reference nodes to verify correctness, then submit weighted scores on-chain each epoch (~72 minutes)
- The problem it solves: Blockchain apps depend on RPC endpoints to query balances, submit transactions, and read chain state. A handful of centralized providers control most of this traffic, creating single points of failure and access control.
- The opportunity: Every dapp, wallet, and chain explorer in crypto runs on RPC infrastructure. A decentralized, incentivized alternative removes a quiet chokepoint that most users never see.
- The Bittensor advantage: Bittensor's emission model provides natural incentives for node operators to compete on quality and price. Validators verify responses against reference nodes every epoch, and confirmed bad responses result in a permanent coldkey ban. The network enforces honesty automatically.
- Traction signals: According to early community reporting, Blockmachine launched on mainnet with paying customers from day one, with TaoStats cited as an early customer. It is one of the few Bittensor subnets with live revenue from an external product.
Category: Other (Blockchain RPC Infrastructure) | Centralized Competitor: Infura, Alchemy, QuickNode
Most users interact with blockchains through RPC endpoints without knowing it. Every wallet that shows a balance, every dapp that submits a transaction, most of them are making calls to a centralized server. Blockmachine is building the decentralized alternative on Bittensor's incentive layer, where node operators compete for the same traffic through measurable, verifiable work.
Mechanism:
Miners run Bittensor subtensor nodes: either lite nodes that sync in about 15 minutes, or archive nodes with full historical state that require 4+ TB of storage. These nodes sit behind an nginx gateway that terminates TLS, authenticates requests via bearer tokens, and proxies WebSocket RPC traffic to the underlying chain node.
Miners set their own price in USD per Compute Unit. A CU represents the normalized computational cost of a specific RPC method: a simple balance query costs fewer CUs than a state trace. The protocol gateway routes customer requests to miners based on quality score and price, cheaper miners with good scores get more traffic. Miners only earn on successful responses.
Validators audit the system by pulling request logs from the protocol's S3 store and re-executing those same queries against independent reference nodes. If a miner's response hash doesn't match the reference, the miner's coldkey is permanently banned. Validators submit weighted scores each epoch based on CU_served × target_USD_per_CU, capped at a share of total subnet emissions. Emissions that can't be attributed to real miner work flow to a designated burn sink address rather than being distributed. With customer demand still early relative to the subnet's total emission pool, the burn rate is currently high. As demand grows, more of that flow shifts to active miners.
With 14 registered miners, over 39,000 TAO in the pool, and net TAO inflows exceeding 1,200 TAO in the past week, early momentum is steady. The low root proportion (~16%) indicates the pool is largely driven by organic demand rather than protocol subsidy. Price is up 6.8% over the past week and 4.3% over the past month.
- Codebase is very early: 4 total commits across 2 contributors as of late April 2026. The technical foundation is working but the project is weeks old, and the surface area for things to break is real.
- Demand not yet matched to supply: The burn sink currently absorbs most miner emissions because customer demand hasn't scaled to the pool's capacity. Active miners are earning relatively little until that changes.
- Centralized competition is entrenched: Infura, Alchemy, and QuickNode have years of tooling, SLA guarantees, and developer trust. Winning meaningful share requires sustained reliability and pricing that outcompetes proven alternatives.
- Single chain focus: Blockmachine currently targets Bittensor's own subtensor RPC. Expansion to other chains is a reasonable next step but is not yet live as of this writing.
Into the next one.
