The most comprehensive Bittensor subnet analytics. Understand the decentralized AI economy powering the future of machine intelligence.
Bittensor is a decentralized protocol that creates a marketplace for machine intelligence. AI models (miners) compete to produce the best outputs for each subnet's specific task. Validators score their work, and TAO rewards flow to the best performers.
Think of it as Bitcoin's proof-of-work model, but instead of mining blocks, you're mining useful AI intelligence. Each subnet is a specialized AI economy with its own incentive structure.
TAO is the native token that powers the entire ecosystem — used for staking, validator delegation, and subnet registration. Like Bitcoin, TAO has a fixed max supply of 21 million with halvings every ~10.5M blocks.
No VC presale. No investor allocations. All 21M TAO earned through mining — just like Bitcoin.
| Entity | Role | Est. Holdings | Notes |
|---|---|---|---|
| Opentensor Foundation (OTF) | Protocol Steward | ~600k–1M TAO | Manages core development, runs SN0 validators, controls governance proposals via Triumvirate |
| Jacob Steeves (Unconditional) | Co-Founder | Large (est. 500k+ TAO) | Genesis miner. Left active role ~2023 to focus on external projects. Known on-chain alias: "Unconditional" |
| Ala Shaabana | Co-Founder / CEO | Large (est. 500k+ TAO) | Active OTF CEO. Leads strategy and protocol development. Early miner from genesis block |
| Polychain Capital | Market Investor | Est. large position | Open market purchase; publicly disclosed TAO position. One of the earliest large institutional buyers |
| Digital Currency Group (DCG) | Market Investor | Est. significant position | DCG and subsidiary Grayscale have expressed public interest; open market accumulation |
| Subnet Owners (collective) | Ecosystem | ~5–10% of supply | Subnet owners receive 18% of their subnet's emissions as an ongoing reward for managing the incentive mechanism |
| Open Market / Miners / Stakers | Community | ~60–70% of supply | Majority of supply is diffuse — held by miners, validators, delegators, and open market buyers globally |
TAO price history and emission schedule
All active subnets — their purpose, category, and role in the decentralized AI economy
The most important subnets driving TAO value and ecosystem growth
Approximate breakdown of the 21M TAO supply — no VC presale, no investor allocations, all earned through mining.
Ownership, mining mechanics, and competitive context for Bittensor's most significant subnets.
The original Bittensor subnet and protocol reference implementation. Miners run large language models and compete to produce the highest-quality text completions. Validators score responses using automated reward models, and TAO flows to the top performers. SN1 proved the incentive mechanism works at scale and remains one of the largest subnets by delegated stake.
Run an LLM (Llama 3, Mistral, Mixtral, etc.) and respond to prompts issued by validators. Scored on response quality via automated reward models. Typically requires 24GB+ VRAM.
General AI inference marketplace. Anyone can query the best open-source LLMs via Bittensor, with TAO payments routing to the top-performing miners in real time.
OpenAI API, Anthropic API, Together.ai, Replicate. Differentiator: decentralized, censorship-resistant, community-owned inference with no single point of control.
No external VC. Foundation-owned subnet. OTF receives 18% of SN1 emissions as subnet owner — the primary mechanism for foundation revenue. Fully on-chain funded.
A decentralized quant trading signal network. Miners act as trading algorithms, submitting directional signals (long/short/flat) on cryptocurrency pairs. Validators track signal performance over time and miners are rewarded based on risk-adjusted returns. No capital is deployed — it's a pure signal competition with real economic stakes.
Submit trading signals (long/short/flat) on crypto pairs like BTC/USD and ETH/USD. Signals are evaluated over rolling windows and scored on Sharpe ratio and cumulative returns.
Decentralized quant trading signals marketplace. Institutional and retail traders can access the top-performing signals, creating genuine real-world demand for miner output.
Numerai (similar tournament model), traditional quant funds like Two Sigma and Citadel. Differentiator: open permissionless participation, on-chain rewards, real-time public leaderboard.
Taoshi raised undisclosed seed funding. The team operates taomarkets.ai as a commercial product. VC involvement is limited relative to typical DeFi projects.
The most ambitious subnet in Bittensor: miners train language models from scratch and compete for the highest benchmark scores. Validators continuously evaluate submitted models on MMLU, ARC, HellaSwag, and other standard evals. The miner with the best model earns the most TAO — creating a permanently running open-source model training race funded by the network.
Train a language model from scratch (or start from a permissive base), upload it to Hugging Face, and register with validators. Must continuously improve to maintain ranking as competitors submit better models.
Open-source foundation model training funded by TAO emissions. All models produced are publicly available — a community-owned alternative to closed AI labs with economic skin in the game.
OpenAI GPT series, Meta Llama, Mistral, EleutherAI. Differentiator: economically incentivized open-source model development with decentralized on-chain reward distribution.
No external VC. Community-operated with Macrocosmos as primary maintainer. Funded entirely through SN9 emissions. One of the most decentralized and open subnets in the ecosystem.
The API monetization layer for Bittensor. Miners route and serve developer requests across all subnets via a unified REST API. Think of it as AWS Bedrock built on Bittensor — a single API key gives developers access to the entire ecosystem. Corcel.io is the consumer-facing product built on top of SN18, the most commercially developed subnet.
Route and fulfill API requests from the Corcel gateway layer. Scored on response quality, latency, uptime, and throughput. Requires reliable high-bandwidth infrastructure and low-latency routing.
Developer API gateway to all Bittensor subnets. Enables app builders to integrate Bittensor AI capabilities without running their own nodes or understanding subnet-level complexity.
AWS Bedrock, Azure OpenAI Service, Google Vertex AI. Differentiator: decentralized backend, censorship-resistant, TAO-powered economics with no per-seat pricing lock-in or vendor dependency.
Corcel team has undisclosed external funding. Generates subscription revenue via corcel.io alongside TAO emissions. The most commercially mature and externally funded subnet team.
Decentralized encrypted data storage incentivized by TAO. Miners store encrypted data chunks across distributed nodes. Validators periodically challenge miners to prove data is intact and retrievable. Scored on retrieval speed, uptime, and data integrity. Designed specifically for AI dataset and model checkpoint storage at scale.
Run a storage node that accepts encrypted data chunks from the network. Validators issue random retrieval challenges — miners must return data quickly and correctly. Scored on uptime and retrieval latency.
Decentralized AI data storage layer. Paired with SN9 (pretraining) and SN13 (data curation), it completes a full decentralized AI data pipeline from collection → storage → training.
Filecoin, Arweave, IPFS, Storj. Differentiator: AI-native storage with Bittensor's incentive scoring, designed for large model checkpoints and AI dataset storage specifically.
No known external VC. Neural Internet team is community-funded via subnet emissions. Originally incubated with OTF support — now fully self-sustaining through on-chain rewards.
A decentralized GPU compute marketplace on Bittensor. Miners register GPU resources (A100s, H100s, consumer RTX cards) and complete compute tasks assigned by validators. Scored against performance benchmarks including FLOPS, memory bandwidth, and task completion time. Creates a permissionless alternative to centralized cloud GPU providers at a time when GPU scarcity is acute.
Register GPU hardware with the subnet and complete assigned compute workloads (matrix ops, ML training tasks). Validators benchmark hardware and score on raw performance, reliability, and sustained uptime.
Decentralized GPU marketplace for AI workloads. Demand side: researchers, small labs, and developers needing affordable GPU access without cloud lock-in or waitlists.
Akash Network, Render Network, CoreWeave, Lambda Labs, AWS EC2 GPU. Differentiator: TAO-incentivized, permissionless, natively integrated with other Bittensor subnets.
No known VC funding. Community-operated by Neurotensor team. Emissions-funded. High strategic value as AI compute demand grows — one of the most watched subnets for future external investment.
Three ways to earn TAO