Bittensor Subnets Explained: Templar, Chutes & How to Buy and Stake $TAO
Our previous article covered why Bittensor ($TAO) exists and what the decentralized AI thesis is. This is the next layer — a technical breakdown of what subnets actually are, which ones matter, and exactly how to get exposure through buying TAO and staking alpha tokens.
What Is a Subnet? The Technical Foundation
Bittensor is not one monolithic AI network. It is a network of networks — each called a subnet — where independent teams compete to build the best version of a specific AI task. Think of the Bittensor mainchain as a coordination layer, and subnets as the individual markets that live on top of it.
A subnet is an independent subnetwork within Bittensor, each with its own task definition, scoring mechanism, and participants (miners and validators). Subnets compete for a share of the total TAO emission based on how much staked TAO they attract.
Every subnet has two types of participants:
- M Miners — provide the actual AI compute or data. They run models, generate outputs, or contribute training. They are scored continuously by validators and earn TAO proportional to the quality of their work.
- V Validators — evaluate the work of miners using the subnet's scoring function. They are the quality gatekeepers. Stakers delegate their TAO to validators, earning a share of the validator's rewards.
The quality of any given subnet ultimately comes down to the sophistication of its incentive mechanism — how well the task is defined and how reliably miners are scored. A poorly designed incentive produces garbage output. A well-designed one produces genuine innovation at competitive scale.
Dynamic TAO (dTAO): How Subnet Tokens Work
In February 2025, Bittensor introduced its most important upgrade to date: Dynamic TAO (dTAO). This fundamentally changed how subnet emissions are allocated and introduced a new class of asset — alpha tokens.
Each subnet now has its own alpha token (e.g. α for subnet SN3, named after the subnet). Alpha tokens are not separate coins — they exist natively within the Bittensor ecosystem and represent your staked position inside a specific subnet's liquidity pool.
Here is how it works mechanically:
- 1 Each subnet operates an AMM (Automated Market Maker) — a liquidity pool containing both TAO and that subnet's alpha token.
- 2 When you stake TAO into a subnet, your TAO enters the pool and you receive alpha tokens in return. The exchange rate at that moment determines how much alpha you get.
- 3 The more TAO that flows into a subnet's pool, the higher the alpha token price becomes (basic supply/demand dynamics within the AMM).
- 4 Emissions follow TAO. The protocol distributes daily TAO rewards proportionally to subnets based on how much TAO is staked in them. Validators earn TAO, a portion of which flows to their delegators.
- 5 When you unstake, your alpha is swapped back to TAO at the prevailing market rate. If the subnet's alpha price has increased, you receive more TAO than you put in — and vice versa.
Staking into subnets introduces variable exchange rate risk. You are not just earning yield — you are taking a directional position on that subnet's alpha price. A subnet that loses credibility or emissions will see its alpha price fall, potentially reducing the TAO value of your stake even while you are "earning" rewards.
This design is intentional. dTAO turns TAO holders into market participants who signal their conviction in specific subnets by staking capital into them. The result is a continuous, market-driven allocation mechanism — one that is more dynamic than any committee or governance vote could ever be.
The Subnet Ecosystem: What's Actually Being Built
As of March 2026, Bittensor has over 60 active subnets with more launching regularly. Below are the most significant ones — those commanding the highest emissions, doing the most technically interesting work, or attracting the most community attention.
Use case: Decentralised LLM pre-training. Templar is the most narratively powerful subnet in the ecosystem. It coordinates thousands of miners to collaboratively pre-train large language models using commodity hardware across the internet — no data centre required. In March 2026, Templar completed Covenant-72B, a 72-billion-parameter model trained entirely on-chain across 1.1 trillion tokens. For reference: a 72B model is competitive with open-source leaders like Qwen 2.5 and Llama 3.3. This is the subnet that proved decentralised AI training is not a whitepaper concept — it actually works. Covenant-72B drove a 20%+ TAO price surge when announced and attracted significant institutional attention via a Grayscale ETF filing shortly after.
Use case: Serverless AI inference and GPU compute. Chutes is currently the highest-emission subnet in the entire network, reflecting the market's conviction in its utility. It provides a serverless platform for deploying any AI model via API — reportedly 85% cheaper than AWS or Google Cloud. Developers can spin up inference endpoints without managing servers, containers, or GPU clusters. The demand for cheap, reliable AI compute is enormous and growing; Chutes is the Bittensor ecosystem's answer to centralised cloud providers. Its dominant emission share signals that TAO stakers believe this is where the real-world revenue will flow first.
Use case: Fast AI inference with deterministic verification. Targon focuses on consumer-scale AI usage — fast, cost-effective inference that can be independently verified for correctness. It powers notable products including Dippy (an AI character chat platform with 4M+ users), TAOBOT, and Sybil (an AI-enhanced search tool). Targon competes with centralised inference providers on speed and cost, while adding the cryptographic verification layer that makes it trustless. If decentralised AI ever crosses over into mainstream consumer apps, Targon is one of the most likely pathways.
Use case: AI-powered financial market prediction. PTN merges deep learning with financial forecasting. Miners build and run predictive models across BTC, ETH, Forex (USD/CAD, EUR/USD), commodities, and equity indices. Validators score signal quality in real time. The network functions as a decentralised quantitative research tournament — the best models earn the most TAO. For traders in communities like Swiss Circle, this is a particularly compelling subnet: it is essentially applying the same analytical rigour we use in our signals, but at scale and fully automated.
Use case: Ultra-low latency AI inference. Nineteen is built for real-time, latency-sensitive inference — the kind of sub-100ms response times needed for autonomous agents, voice assistants, and industrial AI. It positions itself as an outperformer of centralised giants like Mistral and OpenAI on specific real-time workloads. As AI agents become more prevalent in 2026 and beyond, the demand for fast, trustless inference rails will grow significantly.
Use case: General-purpose LLM query and response. The original Bittensor subnet, operated by the Open Tensor Foundation (OTF). SN1 benchmarks miners on the quality of their text generation — effectively an open, incentivised ChatGPT alternative. It is the historical backbone of the network and remains a reference subnet for how incentive mechanisms are designed.
Use case: Multimodal AI and agent infrastructure. Zeus is building agent-native AI infrastructure — models and tooling designed for autonomous AI systems that need to perceive multiple input types (text, image, structured data) and act on them. As the AI agent narrative accelerates in 2026, Zeus is positioning itself as foundational infrastructure for on-chain agents.
The full list of active subnets can be tracked at taostats.io/subnets and taomarketcap.com, both of which show real-time emission rates, alpha token prices, and validator performance.
Emission share is the single most useful on-chain signal for subnet conviction. It reflects where TAO stakers are putting real capital — not speculation, but demonstrated belief with skin in the game. Watching emission shifts week-over-week is one of the most actionable ways to track which narratives within the Bittensor ecosystem are gaining or losing momentum.
How to Buy $TAO: Step-by-Step
TAO is available on all major centralised exchanges. For most traders, a CEX is the simplest and most liquid entry point.
Option 1: Centralised Exchange (Recommended for Beginners)
TAO is listed on Bybit, KuCoin, MEXC, Bitget, and BingX — all of which offer spot trading with good liquidity. Bybit and KuCoin tend to have the deepest TAO order books for larger position sizes. All five are available to register via the Swiss Circle affiliate links on swisscircle.trade.
Register, enable 2FA immediately, and complete identity verification (KYC). Most exchanges require a government-issued ID and a selfie. This typically takes between 5 minutes and a few hours depending on the platform and jurisdiction.
Deposit USDT or USDC via bank transfer or credit card (fees vary). Alternatively, transfer BTC or ETH from another wallet and swap to USDT on the exchange before buying TAO.
Navigate to the TAO/USDT spot pair. Use a limit order at your target price or a market order for immediate execution. For larger sizes, use limit orders to avoid slippage — TAO's liquidity is solid but not as deep as BTC or ETH.
For active trading, holding on the CEX is fine. For longer-term holds or if you want to stake into subnets, withdraw to a non-custodial wallet. TAO runs on its own Substrate-based blockchain (not EVM), so standard MetaMask does not work — you need a TAO-compatible wallet (see below).
Option 2: Wrapped TAO on a DEX
If you already hold ETH and prefer DeFi, the Ethereum-wrapped version of TAO (wTAO) is tradeable on Uniswap. Note that wTAO is not native TAO and carries additional smart contract risk. It is also less liquid than CEX spot. For most traders, the CEX route is simpler and more cost-efficient.
How to Stake TAO Into Subnets (dTAO)
Staking into subnets gives you two types of exposure simultaneously: yield from validator emissions, and price appreciation of the subnet's alpha token. This is more complex than simply holding TAO, but it is how serious participants in the Bittensor ecosystem are positioning.
A non-custodial TAO wallet (SubWallet or the native BTCLI), native TAO tokens withdrawn from a CEX, and a validator address on the subnet you want to stake into.
Step 1: Get a TAO-Compatible Wallet
SubWallet is the most accessible option — a browser extension and mobile app that natively supports all 60+ Bittensor subnets and all 128+ alpha tokens. It is the recommended starting point for non-technical users. Alternatively, advanced users can use BTCLI, the official Bittensor command-line interface, for direct on-chain interactions.
Step 2: Withdraw TAO from Your CEX
In SubWallet, copy your TAO wallet address (it will begin with a 5, as Bittensor uses Substrate address format). On your CEX, initiate a withdrawal of TAO to that address, selecting the Bittensor network. Always send a small test transaction first. Withdrawals typically confirm within 2–5 minutes.
Never send TAO to an EVM address (a 0x... address). Bittensor uses the Substrate address format exclusively. Sending TAO to the wrong address type will result in permanent loss of funds.
Step 3: Choose a Subnet and Validator
Browse available subnets and their validators at taostats.io or directly in SubWallet. Key factors when selecting a validator:
- Validator stake: Higher stake generally means more stable returns — large validators are less likely to be deregistered.
- Validator commission: Validators take a % cut of emissions before distributing to delegators. Lower is better, but very low commission may indicate a validator trying to attract stake quickly — check track record.
- Subnet emission rate: Stake into subnets with meaningful emission (>2%) unless you have strong conviction in a smaller subnet's upside.
- Alpha token price trend: Check the alpha token's TAO-denominated price history. Staking into a subnet whose alpha is in a sustained downtrend means your TAO value erodes even while you earn rewards.
Step 4: Stake via SubWallet
In SubWallet, navigate to Earning → Subnet Staking (dTAO Staking). Select your target subnet, choose a validator, enter your TAO amount, and confirm the transaction. Your TAO is swapped for alpha tokens at the current pool rate. You will see your staked alpha balance and its TAO-equivalent value in real time.
Step 5: Monitor and Unstake
Rewards accrue continuously — approximately every 12 seconds (each Bittensor block). You can check your position's TAO value at any time in SubWallet or taostats.io. When you unstake, your alpha is swapped back to TAO at the then-current rate. There is no lock-up period, but unstaking during a period of low alpha price means you receive less TAO than you put in.
The most sophisticated approach is to treat dTAO staking as both a yield strategy and a leveraged bet on specific subnet narratives. Staking into Templar before a major training milestone, for example, would have resulted in significant alpha price appreciation on top of yield. Monitor subnet development activity, team announcements, and emission changes as leading indicators.
Unstaking timing matters. If you anticipate a subnet's narrative peaking, unstaking before the alpha price corrects protects your TAO denominated returns.
Subnet vs. Simple TAO Hold: Which Makes More Sense?
| Approach | Upside | Risk | Complexity |
|---|---|---|---|
| Hold TAO on CEX | TAO price appreciation | TAO price decline | Low |
| Hold TAO in cold wallet | TAO appreciation + full self-custody | TAO price decline; seed phrase loss | Medium |
| Stake TAO to validator (no subnet) | Base yield (~8–15% APY) + TAO appreciation | Validator performance risk | Medium |
| Stake into high-emission subnet | Yield + alpha price appreciation + TAO | Alpha price decline erodes TAO value | High |
| Stake into early-stage subnet | Highest potential alpha price upside | Subnet fails; alpha loses all TAO value | High |
The right approach depends on your conviction level and risk tolerance. For traders who want clean TAO exposure without operational complexity, a spot position on a CEX with a defined entry/stop structure is entirely valid. The subnet staking layer is for those who want deeper ecosystem participation and are comfortable managing the alpha token dynamic.
Where to Track the Ecosystem
- taostats.io — The definitive Bittensor block explorer. Subnet emissions, validator stats, alpha token prices, network health. This is your primary research tool.
- taomarketcap.com — Clean overview of all subnets ranked by market cap, emission, and validator count.
- subnetalpha.ai — dTAO-focused analytics including historical alpha price charts and staking yield calculators.
- CoinGecko → Bittensor Subnets category — For alpha tokens that have been listed, price and market cap data alongside broader market context.
- tao.media — Community-driven Bittensor news and research aggregation.
The Bottom Line
Bittensor's subnet architecture is one of the most technically interesting structures in crypto. The dTAO upgrade turned it from a monolithic network into a market-driven ecosystem where capital allocates itself to the best AI work — automatically and continuously.
Templar proved that decentralised LLM training at scale is real. Chutes is building the serverless compute layer that developers actually need. PTN is applying machine learning to the same markets we trade every day. These are not vaporware projects — they are operational subnets producing measurable output.
Whether you engage with the ecosystem at the level of a simple spot hold, basic validator staking, or full dTAO subnet positioning, the key is matching your level of complexity to your level of conviction and your operational capacity. Start simple, build familiarity with the mechanics, then go deeper.
This article is for educational and informational purposes only. Nothing in this article constitutes financial advice, investment advice, or a recommendation to buy, sell, or stake any cryptocurrency. Staking and DeFi activities carry significant risks including smart contract risk, exchange rate risk, and total loss of staked capital. Bittensor's subnet economics are complex and can change rapidly with protocol upgrades. Always conduct your own research and consult a qualified financial adviser before making investment decisions. Swiss Circle does not hold any affiliation with Bittensor, the Open Tensor Foundation, or any subnet team mentioned in this article. Past performance of any cryptocurrency is not indicative of future results.