Written by: Route 2 FI

Translated by: Blockchain Simplified

The first time I looked at Bittensor, I closed the webpage right away. Haha, true story. I remember Algod promoted this coin when it was around 50 dollars in the summer of 2023. I refused to research it then, haha. Too complicated. Too niche. Too much like a "research paper." At that time, I held Bitcoin, my main holding was ETH, and a bag of Solana (which I bought for 22 dollars in July 2023). When I looked at it the second time, it really caught my attention. Then I bought the top when it was around 400-500 dollars. I know, what a "fool" (Retardio). But it was the Bittensor subnets that really sparked my interest when I tried them this March. I found out it wasn't "just another L1." It's a machine intelligence market where AI model research gets funded automatically to the best models.

What is Bittensor?

Bittensor is a decentralized intelligence network: an open market where AI models are rewarded based on their usefulness. Think of it as a "trading platform," but instead of trading stock models, analysts are not human winners, but algorithms that score outputs and route rewards. The weak chain is called Subtensor. Initially launched as a Polkadot parachain (laugh), it migrated to its own network in 2023. The token is TAO, with a maximum supply of 21 million, the same as Bitcoin. TAO is the benchmark currency and incentive layer for the entire network.

• Miners run AI models and generate outputs.
• Validators score these outputs.

An algorithm called Yuma Consensus turns these scores into emissions.

TAO will continuously flow to those miners and validators who truly perform.

At the first halving, the network emits 7,200 TAO per day. After the halving in December 2025, this number reaches 3,600 TAO. It's at the same level as Bitcoin: as the network matures, the supply becomes increasingly constrained. This is all continuously happening on-chain. There is no foundation to determine which lab deserves more funding this quarter. The Yuma Consensus almost decides everything to avoid making the system just another ordinary chain. Each validator in the subnet regularly submits a score for miners. The protocol aggregates these scores into weights for distribution and eliminates outliers. If your score is too far above the collective, your influence and rewards will be reduced. The same goes for miners trying to overscore. The result is:

• The cost of lying.
• Collusion is weak.
• Staying fundamentally honest earns more than trying to game the system. It's not perfect, but it’s one of the few attempts on "how to pay models according to practical usage without trusting centralized judges."

Subnets: How this evolves into a platform

Subnets are independent AI markets built on Bittensor. Each subnet:

Defines a specific task (trading signals, embeddings, GPU computing power, drug discovery, etc.).

Has its own miners, validators, and incentive rules.

Competes for TAO emission contributions.

By the end of 2025, there are over 120 active subnets and the growth continues, with a target of 256. By the beginning of 2026, just the top 10 subnets will have a valuation of around 550 million dollars and have now reached 1.5 billion! A few examples of infectious diseases:

• Templar (Subnet 3) — Decentralized cross-training of global computing power.
• Targon (Subnet 4) — Multimodal AI (text, images, audio).
• Infinite Games (Subnet 6) — Prediction markets and real-world prediction.
• Taoshi (Subnet 8) — Trading signals across multiple asset classes.
• TaiFi (Subnet 10) — DEX + liquid staking within the Bittensor ecosystem.
• Nova (Subnet 68) — Molecular modeling and drug discovery.

Anyone can initiate and launch a subnet. If it can attract miners, validators, and stakers, it can get a portion of the emission volume.

Dynamic TAO (dTAO): When the market takes over everything

Okay, wait, the next content will get a bit technical, but I'll try to explain in layman's terms (ELI5). Until early 2025, a group of rooted validators controlled the emission distribution across subnets. 64 validators on subnet 0 held the power, with over half of the control held by just 5 individuals. For a "decentralized AI" network, that is a centralized barrier. The dynamic TAO (dTAO) deployed in February 2025 solved this problem. The market replaced the committee to determine which subnets deserve emissions:

• Each subnet has its own AlphaToken.
• TAO holders can stake TAO into a subnet and receive that subnet's AlphaToken via AMM.
• The Alpha/TAO price and staking liquidity reflect which subnets the market values the most.
• Emissions flow to subnets with higher Alpha value and higher demand.

Miners and validators within the subnet default to that subnet's Alpha. They can hold (leverage stake in that subnet) or redeem back to TAO.

Each AlphaToken also has a maximum limit of 21 million and its own halving schedule. In the months following the launch of dTAO, over 100 subnets went live, shifting from "zombie" subnet emissions to subnets with actual utility. This is the core mental model: TAO is the underlying asset; AlphaToken is your way of expressing a view on a specific AI vertical within the network.

How to purchase subnet tokens (Beginner's guide)

If you can buy spot on a centralized exchange (CEX) and use a browser wallet like Rabby, you have already completed 90%. The primary paths are: simple CEX market and a leveraged Alpha route.

Path 1: CEX Subnet Market (Easiest)

Some trading platforms now directly list specific subnet tokens.

CEX accounts are available for trading TAO and/or subnet tokens. TAO has been listed on platforms like BN, Coinbase, Kraken; some have also listed spot trading pairs for top subnet tokens.

Purchase TAO. Go to the TAO/USDT trading pair, place an order, and complete the transaction.

If available, purchase the subnet token. Search for the subnet code (like TEMPLAR, TAOSHI, etc.). If you see liquidity in the spot pair, buy in just like you would with any other altcoin.

If you are willing to take on the risk of a self-hosted wallet, withdraw to self-custody. Transfer TAO and any subnet tokens to your controlled wallet. You have not yet used the on-chain Alpha AMM, but you have direct price configurability.

Path 2 FI: Gaining Gap Alpha via TAO (dTAO Process)

This is the "real" exposure to subnets: converting TAO to AlphaToken through the Bittensor protocol stack.

Purchase TAO on CEX.

Set up a wallet compatible with Bittensor. Use the official TAO wallet or other wallets that collect Subtensor (Talisman, Taostats, etc.). Create the wallet and back up your mnemonic.

Withdraw TAO from CEX to your wallet. Select the TAO native network, paste your wallet address, and send.

Connect to the subnet UI (like Taostats). Go to taostats.io, connect your wallet, and open the Dynamic TAO / Alpha view.

Select a subnet and conduct basic research. Check the TVL, Alpha/TAO price, emission contributions, and the specific business of the subnet (trading, computing power, embeddings, etc.).

Convert TAO to AlphaToken for that subnet. Click "Stake" or "Buy Alpha" for that subnet. The AMM will convert your TAO to Alpha at the current price. Confirm in your wallet.

Decide whether to simply hold Alpha or to stake/lock. For most users, holding Alpha gives you increased leveraged variable exposure to the subnet. Some subnets have extra staking or locking features to earn more rewards or governance rights.

Be aware of liquidity and volatility. Alpha Token liquidity is thin, reflexive, and extremely volatile. Always check depth and slippage before placing an order.

What TAO holders are actually doing

You can play a clear main role:

• Hold original TAO. Simple network staking: 21 million hard cap, halving schedule, more subnets, and gradually more institutional channels over time.
• Delegate staking in the Root Subnet. Delegate your TAO to validators. They verify across subnets; you are rewarded according to your staking ratio, with a historical annual percentage rate (APR) around 15%.
• Rotate into specific AlphaToken. Choose subnets you believe are important long-term, and stake TAO in their Alpha, betting on the performance of these specific markets and narrative dividends.

What can drive TAO (and subnets) up?

Nothing is absolutely guaranteed, but some clear catalysts must be in play. Additionally, I state this is likely a new top. Basically all owners are now talking about TAO. So you have been warned.

Emission schedule and future halvings

As mentioned earlier, TAO has a fixed supply of 21 million and a predictable halving schedule. The first halving at the end of 2025 cuts the emissions from 7,200 TAO per day to 3,600 TAO, which at the start of the narrative and institutional access helps tighten supply. If this process continues, the next halving to 1,800 TAO/day will be the next major structural catalyst: less sell pressure, more competition for existing circulating chips. The first halving has already occurred in mid-December 2025, so we are now in the era of 3,600 TAO per day. The next TAO halving (from 3,600 to 1,800 TAO/day) is based on the supply indicator date for mining: it will automatically trigger when the emitted supply reaches the next programmed threshold, similar to how the first halving was triggered at 10.5 million TAO issued. Most trackers and analytical models estimate this is about a four-year cadence, meaning the second halving will be around the end of 2029, but it will depend on network activity and the speed of TAO emissions.

Covenant-72B and network scaling

Covenant-72B is a 72B ecological parameter model aimed to operate within the Bittensor ecosystem, bound to the Templar subnet. When it is launched in March 2026, TAO surged approximately 46%, and top subnets like Templar and Taoshi also broke the subnet valuation of around 550 million dollars. Just three weeks later, it has now reached 1.5 billion (tripled). If Covenant-72B gains funding, or real laboratory use, it means continuous demand for subnet output, further translating into demand for TAO and AlphaToken.

Subnet growth and composability

Every successful subnet is its own mini bull market. As more vertical domains join: trading, computing, scientific research, infrastructure, you will see:

• More reasons for developers and users to access Bittensor.
• More demand to use TAO as the underlying asset to enter these subnets.

As funds flow into the "hottest" subnets of the month, more rotation potential is created. Importantly: all this activity will ultimately be routed through TAO.

Institutional channels and listings

TAO already has a Grayscale trust and is listed on six trading platforms with staking ETPs, which broadens its visibility to more conservative funds. The next catalyst looks quite promising:

• TAO and major subnet tokens listing on larger CEXs.
• TAO perpetual contracts and options on frontline trading platforms.

More ETPs or funds being built around the "AI infrastructure basket," containing TAO and subnet composability. Once TAO becomes easier to buy on e-commerce screens rather than on DEX, the buyer pool will widen.

Why Jason suddenly supports TAO wholeheartedly on Twitter

Jason and a leading macro/AI key opinion leader (KOL) are vigorously promoting Bittensor. Why?

TAO is the absolute center of AI + cryptocurrency, and this is the most powerful dual narrative in the market. Commenting on it gives you insights into both fields simultaneously.

It has transformed from a "niche science project" into a "serious infrastructure with subnets, halvings, institutional channels, and live 72B models." This is excellent KOL material.

Some of these KOLs now have exposure through funds or products counter to TAO or subnets, so controlling the narrative on X is not charity.

Game theory. When a strange, illiquid thing starts to break into strict AIToken lists with real infrastructure, you can choose to ignore it and risk missing the trade, or you sprint ahead of the crowd and become a "Bittensor ambassador." Be careful. They might hold substantial VC/angel investment shares. They will throw chips over your head. Just like ZEC. So be very cautious!

TAO is extremely large. Alpha Token liquidity is scarce. The Subtensor underlying layer still runs on a proof of authority (PoA) node set controlled by the Opentensor Foundation, and if mishandled, this poses a real centralization risk. But structurally speaking, there are few:

The token truly affects the routing function of the network.

AI models directly serve competitive capital for the network.

Its architecture allows anyone to build dedicated AI markets and access a shared incentive layer.

Currently, I only hold TAO, Templar, and Targon. Maybe I won't achieve greatness (NGMI), but I'm starting to feel fearful; this might be the top.