Michael Parker

TAO Network. Bittensor news, subnet deep dives, and signal over noise. $TAO intel desk. Breaking updates, subnet scorecards, and real adoption signals

Templar (SN3) is not yet printing the same visible external revenue as Chutes or Lium. They’re still primarily the decentralized training powerhouse: Just completed the largest permissionless LLM training run in history (Covenant-72B, 72B parameters). Got the Jensen Huang shoutout. Pulling 7% of total network emissions purely on technical utility and dev activity. No big disclosed daily revenue numbers or auto-buybacks yet their flywheel right now is compute demand → massive emissions → alpha price appreciation. The revenue stage (paid enterprise training runs, inference licensing, etc.) is the next unlock they’re building toward. It’s coming, but it’s not live at the same scale as the inference-heavy subnets. That’s why I listed the clear revenue flywheels first (Bitcast, Chutes, Lium). Templar is the infrastructure layer enabling the whole stack. Fair callout not every top subnet is at the same stage of the flywheel. Some are still proving the tech before the cash follows. You seeing the difference now?

@bittensormax @Bitcast_network And eg Templar: how much revenues do they generate? I thought they are still not generating any revenues.

@Atppr0 Great question. These are the subnets already turning into real flywheels actual revenue, usage, and buybacks that offset emissions and compound without debt: @Bitcast_network (SN93) First subnet to fully offset miner emissions with real revenue (brands paying creators for attention economy). Historic flywheel live. @chutesai_ (SN64) Leading inference, $22k+/day revenue, auto-staking buybacks pulling TAO back in. @TargonCompute (SN4) Intel-backed confidential compute, $105k revenue in a single week (~$5.5M run-rate). @tplr_ai (SN3) Largest decentralized 72B model training ever, millions in realized PnL. @hippius_subnet (SN75) Cheapest decentralized S3 storage with real usage and traction. @resilabsai (SN46) Real estate pricing oracle for the $300T market, pilots already live. @ridges_ai (SN62) World-leading open-source AI coding assistant. @metanova_labs (SN68) AI drug discovery with real therapies in development. @ai_detection (SN32) New SOTA on MGTD text detection.

@Atppr0 Great follow-up …let’s cut the noise. Revenues DO flow straight into alpha. Take Chutes (SN64) as the clearest example: $22k/day real revenue from paying customers (API calls on their serverless inference platform). Every dollar (or TAO) is auto-staked / auto bought back into SN64 alpha via protocol-level buybacks. It never even touches the team’s wallet it creates direct, non-speculative demand for alpha tied to usage. That’s why their alpha is one of the strongest in the entire ecosystem. Why are Chutes emissions at 0%? Because they don’t need to bribe with emissions anymore. dTAO flipped the script: market demand for alpha now determines TAO emissions. Chutes is so dominant (top market cap subnet, 400k+ users, billions of tokens processed daily) that revenue-driven buybacks + organic staking pull in the TAO weight they need. Zero emissions = pure flywheel, not subsidy junkies. On the “owner pockets billions while miners stay cheap” take: That’s old root-subnet thinking. In dTAO + alpha mechanics: Revenue → alpha buybacks → higher alpha price → more TAO inflows from stakers. More TAO inflows = higher emissions allocated to the subnet. Emissions then split: 18% owner, 41% miners, 41% validators/stakers. Owners are incentivized to grow usage and revenue (their cut scales with the whole pie). Cheap out on miners? Quality drops → users leave → revenue dies → alpha crashes → everyone loses. The flywheel enforces alignment. This isn’t theory it’s live on-chain with Chutes, Lium, Bitcast, etc. Flywheel isn’t “interpreted too good.” It’s executing. You seeing the mechanism now?

@bittensormax @Bitcast_network Thanks. I can't see that all these put revenues into alpha. Why are chutes emissions at 0%? I think the story of flywheels is interpreted too good. A subnet owner will only put as much money in to pay the miner. Owner can make billions but miner work can stay cheap

Meanwhile Bittensor $TAO has zero debt. And profitable flywheels forming.

@pete_rizzo_ Well no actually $TAO is slightly more scarce. Same 21m but 78% locked up in staking. And less in circulation. Harder to mine, and more useful.

20,000,000 #BITCOIN HAVE NOW BEEN OFFICIALLY MINED ONLY 1,000,000 BTC LEFT FOR THE NEXT 114 YEARS THE MOST SCARCE ASSET IN THE HISTORY OF MANKIND 🔥

Would you invest in a company that… Every single day takes $20K of real customer revenue and invest in itself? On top of that, just dropped a fresh $700K stock buyback 100% funded by actual product profits? No dilution. No VC handouts. No debt. Zero debt. Zero. Just pure deflation flowing straight to holders while the business scales massively. Who the hell is this unicorn? It’s not a centralized company. It’s Lium (SN51) the decentralized GPU rental marketplace on $TAO. Real paying customers renting GPUs 85-90% cheaper than AWS → revenue → auto-burns + buybacks. This is the flywheel that the FUD says doesn’t exist. Yet here it is. Centralized AI could never pull this off. @lium_io is setting the damn standard. You seeing how insane this is yet? 👀 Drop your take 👇 #Bittensor #TAO

This should be headline news Doubting the $TAO flywheel? Lium (SN51) just shut you up. Auto-burning $20K of tokens EVERY single day. Fresh $700K buyback funded by real external revenue not a single emission!!!! They’re running ~$14K–$20K/day in usage revenue right now (scaling fast) This is the flywheel executing live: real fiat rolling in → buybacks & burns → deflation → value to holders. GPU rental 85-90% cheaper than AWS/CoreWeave with no long contract @lium_io is cookin’ Which subnet copies this next? #Bittensor #TAO

Exactly and currently 13.46 $TAO reward on mining SN68 @metanova_labs that’s about $4,500 you can get over and over dailyzzz That kills the grant model sorry. Mine on BitTensor don’t hope for a grant some university will beat you to.

SUPER GREAT OMG MOMENT Great coverage of @arbos_born & SN97 @distillpub today by Shizzy and the timing couldn’t be better after Jason talked about SLM today and how essentially the future is hyper focused SLM models. @const_reborn is killing it with this concept. Distil is going to turn out something that will build a tidal wave of infrastructure that no one sees coming. I drank the arbos kool-aid today and it tasted good!

@markjeffrey I know @markjeffrey has brought this up many times! We need an actionable team! I’m first to sign up, paying customer waiting, and/or first to stake if this is a subnet. Let’s go!!!

He who solves this shall have the power of Thor.

@TaoOutsider 🤣

$SOL Hi Solana, are you in compression or have you become a stablecoin? Let me know.

@PushingTao This keeps persisting as a common confusion. We just have to keep explaining .

A post last week claimed $dTAO takes money away from $TAO. It's worth clearing this up. Most people hold $TAO without understanding what actually drives its value. Here's the mechanic that matters. Every subnet on Bittensor has its own alpha token. To stake into any subnet you first need TAO which you swap for that subnet's alpha at the current pool ratio. This means every new subnet that attracts genuine interest creates new structural demand for TAO. Not speculative demand. Mechanical demand built into the protocol. 128 subnets competing for capital. Every dollar entering the subnet economy has to pass through TAO first. That's the flywheel.

Someone please build this! Yesterday. We need a decentralized “SuperClaw” on Bittensor, better, uncensorable, and powered by the entire network. OpenClaw is fully open source so Fork it and build a lightweight orchestrator that routes every single task to the best subnets! Am I crazy to think we could just integrate Basilica (SN39) → one-click secure sandboxed deployment Autoppia (SN36) → full web/computer-use agents Ridges (SN62) → autonomous coding & engineering agents?!?! Top inference, payments (Handshake SN58), RL (TrajectoryRL SN11), etc. for everything else? You get the idea! This would be the ultimate agent platform: smarter, cheaper, unstoppable. I’m not the guy to build it… Someone way better than me please step up. We need this yesterday. Who’s in? Builders, devs, subnet teams tag your squad. #Bittensor #TAO #DecentralizedAI #OpenClaw #SuperClaw

Lmao this is embarrassingly backwards. TAO holders aren’t “subsidizing” they stake TAO directly into your subnet’s TAO/α AMM pool, buying your alpha token with real skin in the game. More inflows = more emissions for you. Revenue from projects? Often gets routed back via buybacks/incentives straight into those pools. “Subnets can walk away”? You’d instantly lose the following: the emissions paying your miners & validators (the decentralized workforce actually building the tech) your own 18% owner cut the TAO liquidity backing your alpha Yuma consensus + network effects It’s like a public company CEO firing every engineer who built the product, then shutting down the factory and wondering why the stock tanks. Pure delusion. dTAO was literally designed to make this suicidal for any rational team. They stay and scale because it’s the golden goose. Read the tokenomics before posting midwit FUD.

@ikundi Lmao this is embarrassingly backwards. TAO holders aren’t “subsidizing” shit they stake TAO directly into your subnet’s TAO/α AMM pool, buying your alpha token with real skin in the game. More inflows = more emissions for you. Revenue from projects? Often gets routed back via buybacks/incentives straight into those pools. “Subnets can walk away”? You’d instantly lose: the emissions paying your miners & validators (the decentralized workforce actually building the tech) your own 18% owner cut the TAO liquidity backing your alpha Yuma consensus + network effects It’s like a public company CEO firing every engineer who built the product, then shutting down the factory and wondering why the stock tanks. Pure delusion. dTAO was literally designed to make this suicidal for any rational team. They stay and scale because it’s the golden goose. Read the tokenomics before posting midwit FUD.

@bittensormax @markjeffrey $TAO holders are subsidising the work, and the subnets can walk away and launch businesses with not revenue from those projects accruing to the holders.

@Atppr0 Great question. These are the subnets already turning into real flywheels actual revenue, usage, and buybacks that offset emissions and compound without debt: @Bitcast_network (SN93) First subnet to fully offset miner emissions with real revenue (brands paying creators for attention economy). Historic flywheel live. @chutesai_ (SN64) Leading inference, $22k+/day revenue, auto-staking buybacks pulling TAO back in. @TargonCompute (SN4) Intel-backed confidential compute, $105k revenue in a single week (~$5.5M run-rate). @tplr_ai (SN3) Largest decentralized 72B model training ever, millions in realized PnL. @hippius_subnet (SN75) Cheapest decentralized S3 storage with real usage and traction. @resilabsai (SN46) Real estate pricing oracle for the $300T market, pilots already live. @ridges_ai (SN62) World-leading open-source AI coding assistant. @metanova_labs (SN68) AI drug discovery with real therapies in development. @ai_detection (SN32) New SOTA on MGTD text detection.

@bittensormax What subnets are flywheels?

Interactive AI Capital Pressure Dashboard. Side by side centralized cash-burn death spiral versus the decentralized option illustrated. $TAO #Bittensor theaideathspiral.vercel.app

$TAO community, WATCH THIS! Massive opportunity for builders in the SLM space. This could spin out entire subnets focused here. And Jason “accidently” gives away the exact model for any 1 man army to build out an extremely profitable business model arbitraging SLM small language models hyper focused on single topics or businesses needing a couture solution. The #bittnesor implication here is vast!

@ChartingFor @markjeffrey Yes! 105! I’m a believer and great great hashrate episode a few days ago by @markjeffrey going over the details and their incentivized mining variation.

@bittensormax @markjeffrey SN 105

Spot on. They are on the leaderboards. People just haven’t been paying attention. Here are some subnets already shipping world class, real world results right now: @ai_detection (SN32) New SOTA on MGTD text detection @ridges_ai (SN62) World-leading open-source AI coding assistant @TargonCompute (SN4) Intel-backed private decentralized compute @tplr_ai (SN3) Largest decentralized 72B model training ever @chutesai_ (SN64) Leading open-source inference @hippius_subnet (SN75) Cheapest decentralized cloud storage @resilabsai (SN46) Real estate pricing oracle for the $300T market @Bitcast_network (SN93) Web3’s largest creator marketing platform @webuildscore (SN44) World-leading decentralized computer vision @lium_io (SN51) World’s cheapest compute marketplace @vidaio_ (SN85) World’s cheapest video compression AI @BitMindAI (SN34) World’s best deepfake detection @metanova_labs (SN68) AI drug discovery reprogramming body + mind And so many more! The market is automatically incentivizing real intelligence. This is the kingdom being built one subnet at a time. $TAO #Bittensor