Openτensor has launched eTAO, giving subnet owners the ability to decide how alpha emissions are distributed, instead of being locked into the fixed 18% (owner) / 41% (validators) / 41% (miners) split. According to the official announcement, owners can now adjust the allocation within certain limits: owner up to 33% (at least 18% for leased subnets), validators up to 50%, and miners at least 33%, with changes allowed roughly once every 7 days. The system also automatically adjusts the subnet’s effective TAO weight to protect root bittensor:native stakers’ returns. In my view, this effectively decentralizes part of the incentive design to subnet owners, allowing different projects to tailor their reward structures based on their specific needs rather than forcing everyone into the same template. For subnets aiming for differentiated competition, this is a meaningful step forward. That said, greater flexibility also means subnet quality will depend more on the owner’s ability to design incentives well. Done right, it could create a positive flywheel; done poorly, it might hurt long-term participation. It’ll be interesting to see how subnets actually adjust these ratios once it goes live.

The AI dividend has already partially arrived, but it is highly concentrated. NVIDIA and the hyperscalers have captured the most certain infrastructure gains. At the micro level, code output and some knowledge work have seen clear productivity improvements, yet broader macro productivity and the average person’s wallet have lagged, with most value still flowing to capital and the largest companies. Decentralized AI projects are trying to offer another path so that contributors, not just shareholders, can capture part of the value:
 $TAO (Bittensor) turns AI work into open subnet markets where miners, validators, and stakers earn rewards based on output quality, and some subnets are already generating real external customer revenue;
 $RNDR focuses on a distributed GPU marketplace serving rendering and generative workloads;
 $FET (ASI) is centered on AI agents;
 $AKT provides decentralized cloud compute resources. These are not simply trying to replace the big labs. They open up parts of the AI supply side so individuals can participate by providing compute, models, or services. Risks remain high — revenue scale, real adoption, and sustainability still need to be proven. Not financial advice, DYOR. #AIDividend #TAO #RNDR #FET #AKT #Bittensor

bittensor:native is still transitioning from an emissions-driven economy to a customer-driven one. Subnets have started generating revenue from external customers paying for compute and inference, with annualized revenue exceeding $50 million and growing relatively quickly. Emissions still account for the majority of incentives, but real customer demand is beginning to appear — a stage most AI projects haven’t reached yet. In my view, this transition matters more than the revenue number itself. Many projects either stay stuck in pure narrative or treat any fee activity as proof that they’ve already achieved product-market fit. TAO’s current state feels more like this: emissions are still supporting the system, but customer payments are starting to reach meaningful scale and accelerating. If this trend continues, the real question becomes whether subnet revenue can gradually reduce reliance on emissions and eventually create more direct demand for TAO itself. It’s still early, but the direction looks right.

From a centralized AI perspective on bittensor:native : OpenAI, Anthropic, and Google still lead clearly in frontier model quality, product maturity, and enterprise support. Bittensor is not trying to directly compete on “who builds the strongest general-purpose model.” It’s doing something different: turning AI capabilities (inference, compute, data, training, etc.) into an open, competitive market. Subnets have started generating real external revenue, but the network looks more like an infrastructure and long-tail innovation layer than a short-term replacement. Centralized AI currently leads at building the strongest models. Decentralized AI is trying to turn intelligence into a market anyone can participate in. In the near term, the two are more complementary than zero-sum.

From a TAO perspective, @alpharidge_ai is worth paying attention to. It’s not a project stuck in the roadmap stage. It’s a live product layer built on SN45: it has real paying users, a clear Pro plan at $99/month, and its news and events engine is powered by subnet miners processing real data. Compared with many subnets that only talk about concepts and haven’t shipped a product, this kind of closed loop—where people are actually paying to use it—is more meaningful for the ecosystem.

A lot of people use AI every day, but very few actually understand it. A UK survey found that almost everyone has heard of AI, yet only about 17% can explain what it is in detail. Most people can only give a vague description or can’t explain it at all. They use AI to write, research, and analyze things, but they don’t really know how it generates content, and they often overlook the fact that it can hallucinate, carry biases, and has clear limitations. bittensor:native (Bittensor) takes this a step further. Instead of letting one company train and control the models, it uses blockchain incentives to organize participants around the world. Different subnets focus on specific AI-related tasks—some handle inference, some provide compute, others work on data or predictions. Validators score these contributions, and those who perform well earn TAO rewards. TAO has a hard cap of 21 million tokens and a halving mechanism, aiming to decentralize both the production of AI and the incentives around it. The problem is that people who don’t even understand basic AI will find TAO’s subnet system, validator scoring, and dTAO mechanisms even harder to grasp. Many will simply buy in because of the “AI + crypto” narrative and treat it as just another hot token, without knowing what the subnets actually do, how rewards are distributed, or where the risks lie. Tools and concepts are spreading quickly, and more people are getting involved, but the number of people who truly understand the underlying logic hasn’t kept up. The knowledge gap isn’t shrinking—it may even be growing. This isn’t anyone’s fault. It’s just the current reality. AI has become part of daily life, and crypto projects are trying to decentralize it, but there’s still a significant distance between “using it” and “understanding it.”

Centralized AI’s path is clear: the strongest models and compute are concentrated in the hands of a few companies. They train, deploy, and price everything themselves. Ordinary people can only access it through their APIs, paying per use, with the capability ceiling, pricing, and even access itself all controlled by those companies. Bittensor is trying a different path. It breaks the act of “providing intelligence services” into competitive subnets where anyone with compute, developers, and miners from around the world can participate. Whoever provides more useful and recognized services earns emissions and real customer revenue. In the future, whether you’re an individual user, a small team, an independent developer, or a regular company, you can directly call and use AI capabilities on these subnets — without first becoming a big tech customer or getting hired by one. $TAO is the base currency of this open intelligence market. The goal isn’t to build yet another stronger closed model, but to gradually turn AI from “a product exclusive to a few companies” into “a public capability that anyone can participate in, call, and benefit from.” That is the problem Bittensor is actually trying to solve.

Bittensor subnets have always paid miners for supply. @say_gm_ (SN28) is now also paying for demand. SayGTM routes a share of Subnet 28 emissions into an on-chain pool. Affiliates compete for it by bringing real paying customers — not clicks, not signups. Your share of referred revenue = your share of that round’s alpha. Proof of Customer instead of just Proof of Work.

Both are scarce. 1 bitcoin:native : mostly hold or trade.
1 bittensor:native : hold + stake + pay for AI inference/compute + stake into subnets. One is digital gold.
 The other is digital gold that also works as fuel.

Most AI products focus on model power, but what actually shapes daily use is whether memory stays continuous. Ditto V2 works at a deeper level: it unifies your Slack, Telegram, and Discord discussions with what you do in Claude, Cursor, and Codex into one shared memory system. You no longer need to manually copy context — the AI can answer and act directly based on real history. It also supports scheduled workflows and coding agent dispatch, with all sensitive actions requiring your approval. As SN118 on Bittensor, it’s validating the real-world potential of the TAO ecosystem through actual users and commercial revenue. @heydittoai

Bittensor just released Null Consensus, raising the subnet participant limit from 256 to 2,500 — a bigger step than many people expected. A lot of subnets weren’t held back by lack of ambition. They were constrained by hard capacity limits and the coordination costs of multi-validator consensus. Null assigns weight submission to the single highest-stake validator, trading some decentralization for lower friction and higher scalability. For subnets aiming for large-scale, high-intensity competition, this is a real efficiency gain. The protocol doesn’t force the switch. Yuma remains the default, and subnet owners can choose. Different applications can now match different incentive structures — some stick with multi-validator consensus, while others use Null to prioritize scale. Optional proof-of-work registration adds another layer. Miners can join without paying TAO upfront. Imagine a future where some AI agents no longer need to “prepare funds” first. They simply solve a computational challenge with their hardware, enter the subnet, and start competing for rewards. If this path works, the mix of participants inside subnets could become much richer than it is today. Higher capacity, lower barriers, and optional mechanisms — these three changes together should make future subnets more diverse and competitive. We may even see large-scale agent collaboration scenarios that are hard to imagine right now. bittensor:native

Looked into @say_gm_’s Proof of Customer setup. Feels like they’re trying to turn “bringing in paying users” into part of the subnet emissions itself. The SayGTM pool isn’t funded from the company’s own margin. They route a share of Subnet 28 emissions into an on-chain pool, and affiliates earn based on the actual revenue their referrals generate. Signups, clicks, and free credits don’t count — only real top-ups do. Every referred user has a 30-day window from their first payment, and the cap is $1 of alpha per $1 spent. Since the pool size, referred revenue, allocation rate, and final payouts are all recorded on-chain each round, it’s relatively transparent. At least it avoids the usual affiliate problem where you drive a lot of traffic but almost nobody actually pays.

Centralized AI costs aren’t that mysterious. On AWS, a single H100 on-demand runs about $6–7 per hour, and inference APIs carry a clear premium on top of that. Meanwhile, subnets on Bittensor (bittensor:native ) are using open competition to push prices down: Lium (SN51) has H100 rentals listed as low as $1.3–1.4 per hour, and Chutes (SN64) offers inference on some open-source models at 30–50% below major centralized providers. The edge on the decentralized side is openness and fuller competition — it can bring idle compute online and keep pressure on prices. Stability, SLAs, and compliance support are still catching up to centralized clouds, but more developers are willing to try the lower-cost, more open option. Right now the $TAO ecosystem is already competitive on price tags, and it’s gradually moving from narrative to something people can actually use.

TAO’s gap with ETH right now is mainly in scale and maturity. Market cap, liquidity, and order book depth are still dozens to hundreds of times apart. ETH already has a large developer and application ecosystem continuously generating demand, while TAO’s subnets are still early. Institutional tools, derivatives, and historical validation also haven’t caught up yet. That said, $TAO is building on a different track — decentralized AI — rather than simply copying ETH’s path. It’s closer to where $ETH was in 2016–2017: clear direction, early ecosystem forming, and some external demand already emerging, but overall scale and network effects not yet proven at a large scale. The gap is real and will take time to close. Saying the difference is small underestimates both the current gap and TAO’s long-term room to grow as infrastructure in a new sector.

$TAO is still missing network effects and historical validation compared to $BTC and $ETH. BTC has been running since 2009 and built the strongest monetary consensus in crypto. ETH has been around since 2015, survived the DAO hack, multiple hard forks, and full market cycles, and developed a mature smart contract ecosystem and developer network. TAO’s mainnet is much younger. Its subnet ecosystem is still early, and real-world usage and external demand are still building. Network effects take time and actual usage to form; historical validation requires more complete cycles and market stress tests. Right now it looks more like a promising early-stage network than proven infrastructure. The gap is mostly about maturity, not the direction itself.

A lot of people ask how TAO’s liquidity actually is. Among mid-cap projects of similar size, it’s relatively solid. Market cap is around $3.2B, 24h volume stays in the $150–180M range, and the volume-to-market-cap ratio is roughly 5.5–6%. It’s listed on major exchanges (Binance, KuCoin, Coinbase, etc.), and the ±2% order book depth on top pairs is enough for normal trading. Compared to BTC and ETH, the gap is clear. bitcoin:native has a market cap of ~$1.68T and ~$28B in 24h volume. ethereum:native is around $307B market cap with ~$10B in daily volume. On order book depth, BTC and ETH sit in the tens of millions of dollars within ±2% on major exchanges, while bittensor:native is mostly in the hundreds of thousands to low millions. BTC’s volume is roughly 150x TAO’s, ETH’s is about 50–60x. The difference is mainly driven by market cap size (BTC is ~500x larger, ETH ~100x), not because TAO’s liquidity is especially weak for its size. In practice: • For trades of a few thousand to tens of thousands of dollars, slippage feels similar. • Once you get into the six or seven figures, TAO sees noticeably more slippage and price impact, so splitting orders makes more sense. Bottom line: Liquidity is decent-to-good among similar mid-caps, but still a clear step behind the blue chips. Size your positions accordingly.

Why is bittensor:native falling alongside AI stocks? Because the market trades it as “decentralized AI” or “Bitcoin for AI.” When Nvidia-related news is positive, or when Jensen Huang mentions decentralized training, or when overall AI sentiment is strong, TAO often rallies with the sector—and sometimes outperforms many AI stocks. When the AI theme cools off, it gets sold in tandem. Today, Korean AI and chip stocks sold off sharply: Samsung Electronics fell about 7.4%, SK Hynix dropped around 10%, and the KOSPI index declined more than 5%. The main drivers were concerns over increased competition from Chinese memory maker CXMT’s technology progress, plus growing skepticism about AI investment valuations being too high. This sentiment reduced risk appetite for the entire AI theme. As a result, high-beta AI-related tokens in crypto pulled back together. TAO is currently trading around $272–$275, down roughly 6% in the past 24 hours—larger moves than Bitcoin. Leveraged liquidations can amplify the downside further. TAO has no direct business relationship with Samsung or SK Hynix. It doesn’t manufacture chips and doesn’t depend on those companies. The price correlation comes from the same pools of capital trading the AI theme: they buy together when sentiment is strong and sell together when it turns. This isn’t a sudden deterioration in the project’s fundamentals; it’s the normal behavior of a thematic trade. When the AI narrative is hot, TAO tends to outperform; when AI sentiment fades, it tends to underperform.

Bittensor exclusive card concept: spend $TAO directly at Visa merchants worldwide. Instant conversion, TAO cashback, and AI subnet rewards seamlessly landing in daily life. Would you use it? #Bittensor #TAO

Most compute subnets offer spare GPUs. @green_compute_ is building permanent, enterprise-grade infrastructure. UK farm biogas → RTX 5090 clusters → real inference demand. They already have $7M+ in permanent capacity, real revenue, and a flywheel where enterprise money buys back SN110 alpha. This is how Bittensor turns renewable energy into durable AI compute.

If bittensor:native fails, decentralized AI loses its most important experiment. Most AI coins today are building around the edges: compute rental, agent frameworks, data supply, or private inference. These can still exist, but they are essentially selling tools and services. TAO is different. It is trying to build an open market for intelligence production itself — where models compete and get rewarded based on usefulness. If this layer disappears, decentralized AI falls back to infrastructure and applications. There is no longer a mechanism that prices intelligence directly. Centralized labs become even harder to challenge, and the ceiling for the entire sector drops. That’s why TAO is not just another AI coin. It is the core bet of the narrative. It can only succeed.