🚨 Every $TAO subnet has been forced to run the same business model. eTAO ends that. It's one of the biggest changes to subnet economics since dTAO. Watch the 80-second explainer 🎬👇 How it worked until now Every 12 seconds, each subnet mints new alpha tokens. Almost every subnet split them the exact same way: • 18% to the owner (the team running it) • 41% to validators and their stakers (checking the work) • 41% to miners (doing the work) A GPU farm, a research lab and a consumer app all lived by one rule. What eTAO (Expressive TAO) changes Each subnet owner can now set their own split, within limits: • Miners: at least 33% • Validators: up to 50% • Owners: up to 33% (18% minimum for lease-created subnets) • Changes about once a week Total emissions don't change. eTAO only changes where they go. Why that's a big deal Different businesses need different budgets: • A compute subnet short on GPUs pays miners more • A research team funds its developers through the owner share • A subnet with hard-to-check work spends more on validators Targon (SN4) is the live example. In the Talking Targon episode with Const, the team said customers want more compute than it has, and most of it is rented. Push miners from 41% to 80% and miners get about 95% more tokens. At the same token price, that nearly doubles the budget to bring on GPUs. More compute → more revenue → bigger buybacks. That's Targon's plan, not a guarantee. The twist Lower yield can be bullish. A high APY tells you how fast new tokens reach stakers. It doesn't tell you what customers pay. Normal companies make this trade every day: reinvest in production instead of paying out. eTAO lets subnets make the same call. The guardrails • Root $TAO stakers are paid their baseline first, calculated from live chain state • Owners can't set validators below that baseline • Miners can never drop below a third • The burn and recycle slots exist but are fixed at zero for now Why? Const explained it: without a floor, a subnet could cut out its outside holders, fake its price signal and pull emissions it hasn't earned. The floor keeps every subnet paying the network that funds it. Everything else that just changed eTAO isn't alone. Bittensor is redesigning itself fast: • Null Consensus: opt-in, up to 2,500 participant slots per subnet, up from 256 • Proof-of-work registration: miners can join with compute, no TAO upfront • Subnet registration cost now halves every 8 days, down from 16 • New-subnet immunity drops from 120 days to 60 • Gamma (in development): subnets buy compute, storage and inference from each other with their emissions Put it together: bigger competitions, easier entry, faster turnover, custom budgets, and soon an internal economy where subnets are each other's customers. The take • eTAO is on testnet now. Mainnet comes with the next runtime upgrade (v481) • An owner can fund the wrong thing. A weak scoring system can pay miners who farm rewards instead of building the product • Alpha staking yields aren't permanent. A subnet's split is now part of what you're investing in • Emissions aren't revenue. The bull case only works if extra output gets paid for Why it matters for AI AI doesn't grow in straight lines. Agents make thousands of requests while a person reviews one dashboard. Workloads shift from compute to storage to verification in weeks. A network funding that can't run on one fixed rule. Now it doesn't have to. 🧠 Sources: Bittensor release notes (v481), Talking Targon Ep. 01 with Const, Novelty Search. $TAO / Not financial advice, always DYOR.