Bittensor (TAO)

For anyone trying out @AlphaGapTAO for the first time, you’ll start on the free version, which has limited access to data. However, you can still gain valuable insights from it. For example, when viewing SN14 on @AlphaGapTAO , you can access the following data: ▫️ Market data, including market cap, FDV, TAO in pool, and staked TAO ▫️ Recent signal reports and the weighted score for each signal ▫️ Emissions trajectory chart, as well as the TAO Flow EMA chart ▫️ SN14 social activity data, dev activity charts, flow charts, and other useful metrics These are just a few of the insights available on the free version. As mentioned earlier, there is much more data available. If you need deeper datasets and advanced features, you may want to consider their subscription plan. For now, feel free to explore the free version here: https://t.co/3njcPyRKYw

The Nerds just had an AMA with @knakamor from $TAO SN78 @vocence_bt. I’ll write a longer post tomorrow, but my first reaction is pretty simple. I don’t know if it even matters that much what @knakamor builds. I don’t mean that in a bad way. @vocence_bt is obviously interesting. Voice AI is going to be huge. TTS, voice agents, cloning, emotion, accents, real-time speech, all of that matters. But sometimes the subnet is not just the tech. Sometimes the founder is the signal I think @vocence_bt will do very well.

$TAO long running in profits $BTC short running in profits. Booking partials here from shorts and waiting for New York close. The AMD triggered but I still need a proper MSS on 5min before I add to my winning position. https://t.co/jVGZU1unhs

The market is still treating $TAO's subnets like an experiment. The data doesn't agree for most part subnets are producing More Economic Motion = Higher Internal Velocity. 18% of all $TAO is now locked inside subnets. A year ago, that number was near 2%. Root stake dropped from 70%+ to 47% as capital migrated into the actual product layer. Subnet aggregate market cap sits at $1.08B with $46M in 24h volume. Chutes (SN64), Targon (SN4), Score (SN44), Ridges (SN62) are generating real inference, compute, and dev revenue. Not promises. A Live product market. Subnets are actually buying from subnets. Chutes sells inference to other subnets. Targon sells compute. Data subnets feed agent workflows. Every time the same $TAO recycles inside the ecosystem the velocity multiplier compounds: Where: c is the percentage of value recycled back inside the subnet economy M is the effective velocity multiplier c=0.30 → M=1.43x, c=0.65 → M=2.86x. Inference buys compute. Compute supports training. Training creates models. Models power agents. Agents buy more inference, data, storage, scraping, prediction, VPNs, and tooling. Do you get how powerful this self enforced loop is? So: c = 0.30 gives 1.43x c = 0.65 gives 2.86x That means if Bittensor moves from 30% internal recycling to 65% by 2030, the same base $TAO can create almost double the economic throughput inside the ecosystem. This is not theory. @GSchvey published the on-chain validation. Templar announced training, capital flowed into its pool. Covenant exited, capital fled. The AMM pools are doing exactly what they were designed to do price subnet output in real time. Three independent valuation frameworks @Old_Samster's SOTP via revenue multiples, Yuma's OpEx-replacement floor, monetary throughput thesis all converge on the same conclusion. Current $2.7B market cap is priced as if the operator-revenue transition fails. Important assumption: this is not predicting that c will definitely reach 0.65 by 2030. It models what the system looks like if subnet to subnet spending, internal routing, staking, emissions, and recycled demand keep increasing toward that level. Just keep in mind that the velocity multiplier is only the first layer of this. It does not fully account for shrinking liquid supply, staking, conviction, subnet to subnet revenue loops, external customers, AI agents routing workloads, or the compounding effect of every subnet becoming its own market. This is not failing, folks. It is just getting started. This pattern is not new. Infrastructure that monetizes is always mispriced before it monetizes. We have seen this movie before. $TAO DYOR.

THE ENERGY IS SHIFTING I can feel it G Communities hyped for UTILITY not memecoins bittensor:native and ethereum:0x4a220e6096b25eadb88358cb44068a3248254675 holders moving different This is what I want to see on my timeline in my community and in my wallet Real projects. Real believers. Real conviction. Someone in my comments said “this is going to be our GameStock time, they’ll make Dumb Money 2 about this group” And honestly? He might be right 💀 Utility supercycle loading LFGGGGGG Tribe -Brondor

The Nerds hosted @matthew_karas from $TAO sn 59 @babelbit . The takeaway was simple. @babelbit is building interpretation infrastructure. That sounds close to translation, but it is a very different problem. Translation systems run a pipeline. Speech to text. Text translated. Text back to speech. @babelbit skips the pipeline and works directly with speech tokens — phonemes, syllables, words, phrases, meaning-bearing audio units. A speech-mode transformer learns the relationships between them and produces interpreted speech in the target language. Speech in. Meaning out. A professional interpreter does not copy words. They preserve meaning. Clean up hesitation. Fix incomplete phrasing. Turn messy speech into clear communication. That is where @babelbit gets interesting. The product becomes more valuable as the domain gets more specific. Legal, medical, diplomatic, enterprise — each has patterns. @matthew_karas said if Babelbit had the communication archives of the top UK legal firms, it could become far better at legal interpretation than a generic tool. The moat is domain-specific speech intelligence. Four people. UK limited company. Matthew on strategy and revenue. Tom on operations. Mica on subnet mechanisms. Josh as chief scientist. Matthew was taught at Cambridge by Tony Robinson, the first person to apply a neural network to speech recognition in 1987. The most interesting new idea was Trans-Modal Distillation Training. Take a large multimodal LLM that translates text across many languages. Systematically reduce its size until it can do little else but translate. Then use it to train a speech-mode transformer by turning text tokens into speech tokens. Text trains speech. That is the trans-modal part. The go-to-market was one of the strongest parts of the AMA. @babelbit is positioning as the engine inside other products. More “Intel Inside” than Apple. SDK. API. Enterprise workflows. Reseller channels. Marketplace distribution. They are working with an AWS Marketplace Partner. Big enterprises have unallocated budget for new tech experiments. That is their door in. Then came the market twist. @babelbit has been approached twice about the paraphrasing alone. Someone speaks in heavy dialect, with hesitation and messy phrasing, and Babelbit outputs polished BBC-style English. One language. Less data. Easier to sell. The big bet is multilingual interpretation. First cheque may be English cleanup. @babelbit also made one of the strongest Bittensor-first commitments I have heard. The company articles include a rule: if they can find a supplier or partner inside Bittensor, they look there first. Confirmed integration with @vidaio_ SN85. Talking to Koyuki from SN78/SN26. That is the ecosystem behavior $TAO needs more of. @const_reborn reached out in December after @babelbit forked Affine code. Matthew spent time with him at the DCG conference in Spain. Signal matters. Babelbit sees Bittensor as critical because the training burden runs for years. They intend to build mechanisms that support the alpha token, but no final structure yet. Can the company value connect back to alpha value? That remains the main investor question. My takeaway: @babelbit is one of the clearer examples of a subnet using Bittensor for a real technical workload instead of wrapping a narrative around emissions. The founder has real background. The architecture is different. The first wedge may be closer than expected. The ecosystem-first posture is strong. Open questions: Can the training scale? Can miners improve the actual product? Can reseller interest become signed customers? Can the alpha token capture value from the company’s success? Babelbit is trying to make machines interpret meaning the way humans actually communicate. That is a much bigger idea than translation. Hosted in The Nerds Telegram. NFA.

Bittensor >> $TAO >> $dTAO Excellent article on SN100 Plateform @platform_tao When architecture → training → datasets → agents → products → revenue ,form a continuous loop powered by Bittensor incentives, things start becoming very interesting. If this vision works, Platform could become far more than an AI challenge: a decentralized research laboratory capable of building real products used outside the network.