Writing / Reporting
Reporting.
News and explainers on AI, robotics, and Bittensor, written for Intelligence at tao.media. 86 articles since January 2026. Every headline opens the full piece on tao.media.
Explainers
Longer guides to how Bittensor works, from TAO issuance to subnet competition.
- 01How Bittensor Works: A High-Level Guide to the Network's ArchitectureA guide to Bittensor’s architecture, covering subnets, miners, validators, Yuma Consensus, Dynamic TAO, alpha tokens, and why subnet staking works differently from traditional proof-of-stake.
- 02What Is Bittensor?A beginner-friendly guide to Bittensor, TAO, subnets, Dynamic TAO, and how the network is building an open marketplace for artificial intelligence.
- 03How TAO Is MintedLearn how TAO is minted on Bittensor, including block rewards, emissions, halvings, Dynamic TAO, subnet rewards, validators, miners, and the 21 million supply cap.
- 04Bittensor vs Centralized AI (OpenAI, Anthropic, Google): A Complete GuideHow Bittensor stacks up against centralized AI efforts, and why decentralized AI wins long-term.
- 05Is TAO Deflationary?TAO is disinflationary, not deflationary. Its supply keeps rising toward the 21 million cap even as inflation falls with every halving.
- 06How New Subnets Are CreatedA plain-English guide to how new subnets get created on Bittensor, covering registration, lock costs, slot competition, and what goes live on-chain.
- 07How Subnets Compete for EmissionsA guide to how Bittensor subnets compete for TAO emissions under Dynamic TAO.
- 08Subnet Staking vs Root Staking: How TAO Earns on BittensorSubnet staking swaps TAO into subnet alpha, while root staking keeps TAO principal 1:1 and earns claimable validator basket yield after Root Reborn.
- 09What Is Alpha on Bittensor? How Subnet Tokens WorkLearn what alpha is on Bittensor, how subnet tokens are minted and priced against TAO, and how alpha rewards flow to network participants.
- 10What Is a UID on Bittensor? How Neurons Register and Stay AliveLearn what a UID is on Bittensor, how subnet registration works, why immunity matters, and how pruning affects miners and validators.
No articles match.
October 2026
- Agility Picks NVIDIA IGX Thor as Digit 5’s Sole Onboard Compute Platform The humanoid’s compute transition expands Agility’s safety collaboration with NVIDIA ahead of early access in 2027.

- Chutes Plans Open-Source AI Lab on SN64 With Consumer-Hardware Training Chutes’ network will gradually expand training incentives while its inference business continues to support model development.

- OpenAI Models and Codex Head to Google Cloud's Model Garden Enterprises will be able to use existing Google Cloud spending commitments to access OpenAI models alongside Google's governance tools.

- SpaceX Revenue Hits $14.6 Billion as AI Leads Growth and Tesla Merger Talks Begin The company forecasts another jump in sales as compute contracts expand, while independent directors weigh a combination with Tesla.

- Tesla Begins Optimus Production in Fremont, Aims for External Deliveries in Late 2027 Tesla is building its first production-quality humanoids while refining the hands, AI, and durability needed for industrial customers.

- Grok Bot Launches Editable Pages With Scheduled Updates Users can turn Bot tasks into shareable reports, memos, and dashboards that stay current through recurring routines.

- What Is REA? Reverse Engineer Anything and Perplexity Computer’s /reverse-engineer Skill Explained REA is an open-source toolkit that lets AI agents inspect how software works without its source code, and Perplexity Computer now offers it as a built-in skill.

- NVIDIA Adds Isaac OS and Isaac Lab 3.0 as Agile Robots Unveils Research Humanoid Agile ONE SN pairs preconfigured robot-learning software with Jetson Thor as NVIDIA expands its tools for training and deployment.

- OpenAI Acquires Orchestra Bio to Add Drug-Development Coordination to Rosalind The deal brings AI agents for tracking studies and managing drug programs into OpenAI’s life sciences business.

- Affine Launches MiniMax-H3 Video Studio and SN120 Sparse-Attention Competition The Bittensor subnet is rewarding better attention selection to reduce video-generation compute while preserving output quality.

- SayGM Adds OAuth Login for AI Apps With User-Funded Inference The Bittensor gateway lets users authorize model access with separate permissions and spending limits for each app.

- Kodama Raises $7 Million to Expand Remote-Operated Forestry Robotics The company is scaling robotic skidders that combine remote operators with supervised autonomy in commercial logging operations.

- AGCO Plans 2027 Digit Humanoid Pilot With Agility Robotics at Kansas Factory The agricultural equipment maker will test humanoid automation for repetitive material-handling work at its Hesston plant.

- NVIDIA and Germany Agree to Develop Physical AI Models, With Four Startup Investments Planned The partnership aims to bring AI into German industry, alongside planned backing for robotics, healthcare, audio, and satellite companies.

- OpenAI Releases MultiLangBench as GPT-6 Models Improve COBOL and Fortran Coding New evaluations test whether AI can build database software in less-represented languages and migrate COBOL without changing its behavior.

- Gecko Robotics’ Navy Inspection Contract Ceiling Rises to $300 Million The expansion supports wider use of robots that map ship corrosion before repair work begins.

- CoreWeave Bond Yield Holds Near 15% as Investors Weigh Refinancing Risk Credit investors are demanding higher returns to finance the AI cloud provider’s debt-dependent expansion.

- Recycled vs Burned TAO: Where Registration Costs and Fees Actually Go Learn how recycled and burned TAO differ, where Bittensor registration costs and transaction fees go, and why recycling can delay halvings.

- What Is an Axon on Bittensor? How Neurons Send and Receive Work An axon is the server a neuron advertises, a dendrite is the client that queries it, and a synapse is the message they share.

- Targon Opens Revenue Dashboard as Enterprise Customers Reserve New GPU Capacity Bittensor’s Subnet 4 is making its compute business more visible as customers book capacity on long-term contracts.

- Perplexity Releases Firecracker-Based SPACE Agent Sandbox Under Apache 2.0 The platform behind Perplexity Computer combines isolated code execution with persistent workspaces and host-side security controls.

- Bittensor Publishes Etao Upgrade for Subnet-Controlled Emission Splits The v480 upgrade lets subnet owners rebalance participant rewards while preserving root stakers’ baseline share at each change.

- What Is Chain Consensus on Bittensor? How Blocks Differ from Yuma Learn what chain consensus means on Bittensor, how Proof of Authority works today, and why it is different from Yuma Consensus.

- CERN and Mistral Plan Large Physics Model for LHC Collision Analysis The proposed foundation model would learn from billions of simulated collisions to support multiple particle-physics tasks.

- NVIDIA Launches Nemotron 3.5 Super VL With Open Weights for Multimodal AI Agents The 120B-parameter model gives developers a customizable foundation for agents that reason across documents, images, video, and tools.

- What Are Subnet Hyperparameters on Bittensor? The Settings Behind Every Subnet Learn what Bittensor subnet hyperparameters are, who can change them, how to check live values, and why settings can differ across subnets.

- Agility Robotics Moves Closer to Nasdaq Listing After SEC Clears Churchill XI Merger Filing Churchill Capital XI shareholders will vote Nov. 2 on the SPAC deal, which would take Agility public as AGLT.

- NVIDIA Commits $250M in Compute to UK AI Agent Safety Research The five-year donation will route compute through U.K.-based cloud partners while NVIDIA expands its climate-AI work with The Earthshot Prize.

- Google Rolls Out Gemini 4 Argon Across Gemini App, Search, Antigravity, and API The release puts Google’s newest frontier model in front of Pro and Ultra users, developers, enterprises, researchers, and vetted cyber defenders.

- Meta Launches Muse Wonder Models for Long-Running AI Agents The Muse Wonder series adds Active Memory and Active Focus for long-running agents, while keeping Meta’s latest frontier models behind products and APIs.

- TPN Labs Releases First Production Model From Bittensor Subnet 65 The first TPN v0.2 winner compresses Ministral-3-8B by about 40% while retaining its full 32,000-token context window.

- ORO Launches AssistantEval Benchmark for Replayable AI Assistant Evaluation The Bittensor Subnet 15 team published a public leaderboard that grades assistants from service logs in resettable synthetic environments.

- Score Launches Detect Football Event on Bittensor Subnet 44 The new SN44 task turns low-resolution youth football footage into an hourly computer vision benchmark with an 85% target score.

- Aleph Alpha Releases Kolibri, a 78B Open-Weight German-English Model for Sovereign AI The Heidelberg lab’s Apache 2.0 mixture-of-experts model targets public administration and industry with German-English reasoning, long-context support, and customer-controlled deployment.

- Coco Robotics and Deliveroo Launch UK Robot Deliveries in Canary Wharf The exclusive partnership marks Coco’s UK debut and Deliveroo’s first autonomous-vehicle delivery rollout in the country.

- Cohere Launches North 2 for Enterprise AI Agents With Memory and Token Controls The upgrade adds persistent agent memory, reusable skills, model-agnostic deployment, and North Admin controls for private enterprise AI rollouts.

- Meta Opens Muse Gadgets With Open-Source Hardware SDKs and Muse Home Link Meta is opening Muse to DIY hardware with open-source ESP32 and Linux SDKs, while shipping a limited USB-C Home Link dongle for U.S. subscribers.

- Grok Bot Adds Proactive Suggestions That Spot Work Without Being Asked The rollout lets a user’s primary Bot surface work it can handle before being asked, with suggestions excluded from usage limits.

- Google Launches Gemini 4 Argon Frontier Model for Coding and Cyber Defense The model is rolling out first to trusted cyber defenders before a wider release to paid API customers and Google AI Ultra subscribers.

September 2026
- Green Compute Eyes $4M UK Data Centre Expansion After Prior $3M Investment The Bittensor SN110 operator said it previously invested $3 million in data centre capacity and is now targeting another UK site by the end of October.

- Cohere Launches Embed 5 Pro and Fast Frontier Embedding Models The new embeddings family pairs a max-quality Pro model with a lower-cost Fast model in one shared embedding space for enterprise RAG, search, and agentic retrieval.

- Targon Launches Bare Metal and Sandboxes for Dedicated and Ephemeral Compute Manifold's platform adds physical servers and disposable Linux environments, broadening Targon's Bittensor compute marketplace beyond confidential containers.

- ChronoLLM Goes With Point-in-Time Models The Subnet 38 product gives quants point-in-time model vintages for backtesting without future information leaking into historical research.

- Opentensor Introduces Gamma Tokens as Proposed Economic Layer for Bittensor Opentensor says gamma would let Bittensor subnets turn emissions into transferable usage credits for compute, inference, and infrastructure inside and beyond the network.

- Tesla Scales Optimus Output as Humanoid Production Line Takes Shape in Fremont The reported ramp moves Optimus into higher internal production, though hands, suppliers, and general-purpose software remain bottlenecks.

- OpenRoboto and Axis Robotics Launch Open Axis Benchmark for Robot Model Evaluation The living benchmark rotates robot-manipulation tasks from Axis Robotics' growing library to test whether models can generalize beyond fixed evaluation sets.

- What Is Stake Weight on Bittensor? Alpha, Root TAO, and Validator Power Learn what stake weight means on Bittensor, how alpha stake and root TAO combine through tao_weight, and why it matters for validator permits and Yuma Consensus.

- What Is Tempo on Bittensor? How Epochs Settle Subnet Rewards Tempo is Bittensor’s payout clock: the block interval between subnet epochs when Yuma Consensus settles accumulated alpha rewards.

- What Is a System One Model in AI? Learn what a System One model is, how it differs from traditional LLMs and reasoning models, and why fast, structured AI decisions matter for automation.

- Who Controls Bittensor? Bittensor has no single owner. Protocol keys, code repos, validators, subnet owners, and TAO holders each control a different layer.

- Swarm SN124 to Begin April 2027 Field Evaluation With European Security Partner The Bittensor SN124 drone-autopilot subnet says Interceptor and Swarm SAR will move from simulation and controlled testing toward field conditions.

- Citrini Research Says Robotics Has Reached a Physical AI Tipping Point Citrini’s latest field report argues robotics progress is moving from demos toward useful autonomy, while warning that commercialization remains uneven.

- Watney Robotics Raises $80 Million to Scale Data Center Robots The funding comes as the startup says its 24/7 robot fleet is already operating across U.S. data centers for major hyperscalers.

- Figure Index Crosses 86,000 Weekly Active Users Uploading Robotics Data The company says Index is now drawing 86,000 weekly contributors as it expands a real-world training pipeline for general-purpose robots.

- Agility Robotics Unveils Digit 5 Humanoid for Safer Factory Work The fifth-generation Digit adds a 50-pound payload, faster charging, and a safety architecture designed for work closer to people.

- Refinery Turns Bittensor Subnet 125 Into an Optimizer Research Market The subnet rewards miners for Python optimizer code that improves a standardized 360M-parameter training run, with confirmation required before emissions are paid.

- XPeng Opens IRON Humanoid Robot Production Line in Guangzhou The Guangzhou line moves IRON from prototype theater toward automotive-style manufacturing, though XPeng has not disclosed price, capacity, or real-world deployment results.

- Chutes Names Christian Hyland CEO as AI Compute Platform Expands The leadership change gives Chutes a named executive as the Bittensor-linked compute provider scales serverless infrastructure for open-source AI models.

- Leadpoet Prepares Sales Agent Upgrade With LinkedIn and Email Integrations The Bittensor SN71 team says the release will pair a revised incentive mechanism with direct lead engagement tools.

- Lium Tops LocalMaxxing Qwen3.6 35B Board With 14,499 Tokens Per Second The Bittensor SN51 team says the benchmark shows how rented GPU nodes can serve open models at lower per-token costs than hosted API providers.

- Bittensor vs Other Decentralized AI Networks The real differences emerge in what each network sells, how it coordinates supply, and whether demand comes from customers or token incentives.

- Why Bittensor Exists: The Open-Source AI Thesis The world of open-source AI needs markets, incentives, and contributors who get paid to keep building in public.

August 2026
- Pollen Robotics Makes a Humanoid Hand Cheap Enough to Build at Home Amazing Hand Enhanced brings a compact, open-source humanoid hand for less than $200.

- Beni Camera Robot Wants to Be a Moving Camera Operator Mondo Robotics turns the action camera into a small all-terrain robot that can follow, film, and jump.

- Photon Matrix Built an Iron Dome for Mosquitoes LiDAR, radar, and AI vision spot mosquitoes in flight and turn backyard pest control into active interception.

- Chutes Proposes Conviction Lock for Federated AI Compute Access Chutes proposes conviction lock, linking time-weighted stake to GPU access while building a federated network of independent AI service providers.

- TAO and the dTAO Market: August 2026 Update TAO trades from $219 to $229 as tracked dTAO subnet tokens approach $1 billion. Review emissions, Root Reborn, ETF filings, and market risks.

- What Are Bittensor Subnets? Bittensor subnets are specialized markets where the network's AI gets built, each with its own token and underlying product or solution.

- TAO Supply and Emissions: How Bittensor's 21 Million Cap and Halving Schedule Work How Bittensor issues TAO under a fixed 21 million cap, why halvings trigger on supply milestones rather than the calendar, and where supply stands now.

July 2026
- How TAO Rewards Are Distributed How TAO rewards flow through Bittensor: block emissions, the two-track split into subnets and alpha, and the 41/41/18 payout to miners, validators, and owners.

- TAO Tokenomics Explained Simply How TAO tokenomics work, explained simply: a fixed 21 million cap, Bitcoin-style halvings, token recycling, and why most of the supply stays staked.

June 2026
May 2026
February 2026
January 2026
- Bittensor Launches Global Subnet Ideathon Bittensor has announced a global initiative aimed at onboarding developers and miners to its ecosystem.

- GTAO Performance Analysis Grayscale’s Bittensor Trust (GTAO) is now publicly trading, giving traditional investors regulated exposure to TAO and the decentralized AI network behind Bittensor.



