Bittensor (TAO) is shaping up to be the “golden child” of the Crypto x AI narrative At ~$3B, with supply side emissions collapsing post-halving and subnet demand accelerating, the asymmetry is becoming hard to ignore With so much noise around @opentensor, it’s easy to miss what actually matters; here’s a clean segmentation of the ecosystem and the subnets I’m paying attention to: a) Decentralized Compute: Subnet (SN) 64 Chutes: @chutes_ai Distributed Serverless AI Compute Platform. Compute is already becoming a bottleneck in traditional AI, so it's logical decentralized solutions will more than supplement the shortfalls it launched confidential compute last month with its Trusted Execution Environment (TEE); strong candidate to perform as Privacy AI takes off in 2026 SN120 Affine: @affine_io is an incentivized Reinforcement Learning (RL) arena for “reasoning”: miners compete to make measurable improvements on tasks like coding b) AI Training: SN 3 Templar: @tplr_ai providing the decentralized solution to AI training, and recently cited by the Anthropic Co-founder as the largest decentralized model (they've observed) when discussing compute constraints c) AI Agent SN 62 Ridges: @ridges_ai enable AI Software Engineers with the intention of fully replacing human coders. This is inevitable; AI will become more competent than humans at code, when you have AI coders reviewing and iterating on existing coding using reinforcement learning the software creation will evolve at a parabolic rate d) AI Trading: SN 15 Bitquant: @BitQuantAI is the "World’s first crypto AI quant" to automate trading. Most of institutional trading is automated so profitability will come down to who has the best AI quant strategies SN 8 Vanta: @vantaSN8 crowdsources trading strategies and produces market signals, rewarding the best performers SN 50 Synth: @SynthdataCo is a price prediction subnet that iniitally focused on Bitcoin price forecast analysis, and has since bridged to more assets e) Confidential Compute: SN 4 Targon: @TargonCompute provides secure cloud with high-performance GPUs and CPUs; something that's fast becoming in short supply f) Data curation: SN 33 Ready AI: @ReadyAI_ has Harvard incentive designers figuring out the subnet economics for emissions, while making the world's data accessible to AI g) Vision Models / AI: SN 44 Score: @webuildscore uses Vision AI to enable businesses to operate more effectively. Recently partnered with @ReadingFC, has a large commercial contract with a European petroleum supplier and many more in the pipeline SN 72 (NATIX) Streetvison: @NATIXNetwork are partnered with @grabid (Uber of SE Asia) They recently revealed their "World Vision" tech; autonomous vehicles are going to become more prevalent as geospatial awareness improves. SN72 Streetvision leans into Vision Language Action (VLA) models to capture edge case data sets which will be invaluable for autonomous vehicles h) Quantum Computing: qBit labs SN 48 + SN 63: @qBitTensorLabs publicly launched @OpenQuantum_ at Q2B Silicon Valley in front of the leading names in quantum gathering decent attention from big names in quantum I) AI safety: SN 60 Bitsec: @bitsecai AI powered code vulnerability detection; something that'll become increasibly important as cybersecurity becomes an increased attack vector J) DeSci: SN 68 Metanova: @metanova_labs is an AI drug discovery company developing therapies to reprogram body and mind SN 76 Safe Scan: @SAFESCAN_AI recently launched a skin scan app to help battle cancer, and has been accepted into the EU-supported UNICORN HUB accelerator k) Other: - SN 18 Zeus: @zeussubnet weather prediction systems; invaluable for agriculture and many other industries - SN 16 Bit Ad: @bitads_ai Advertising leveraging AI systems; "Social Dilemma" vibes returning to the forefront of attention here - SN 34 Bitmind: @BitMindAI helps with Deep fake detection; something that's going to become increasingly valuable in a world where AI proliferates fake news Ultimately how I think about the inevitability of a subnet comes down to just a few things: 1) Which subnets are solving for the largest problems; data constraints etc? 2) Can the subnet team execute on their vision? 3) What are the sinks for the alpha token (buybacks from existing commercial arrangements / revenue) and will that outperform TAO? If you're looking for a more comprehensive analysis on the ecosystem, follow @KhalaResearch and turn notifications on for a new report in the coming weeks Let me know in the comments on any other subnets tackling lofty problems