BIG ANNOUNCEMENT FROM HUGGING FACE TODAY: We're unveiling Microduck 🐥🤖 It's a tiny $399 open-source robot you can teach new tricks with reinforcement learning. It can walk, pick things up, get back up when it falls, and even roller-skate. Welcome to the era of open-source affordable robots to democratize physical AI and world models! 🤗🤗🤗

Robotics Starts Here is live. 13 parts. Complete beginner's guide. No gatekeeping, no prerequisites, just clarity on what robotics actually takes and where to start building. From understanding the stack to your first robot at Tnkr.ai

数据收集是物理智能（Physical Intelligence）发展的最大瓶颈，而激励机制则是解决这一瓶颈的最佳方案。 物理智能需要的是在真实环境中同步记录的“感知状态 + 执行动作 + 结果反馈”（如视觉、触觉、力反馈、抓取轨迹等）。这种数据无法从网络“刮取”，必须通过真实硬件一次次生成，导致严重稀缺。具体瓶颈原因如下： 1获取方式根本不同，无法规模化 2成本极高 3速度慢、多样性差、风险高 4现有数据量严重不足
 为什么激励是解决数据瓶颈的最佳方案？ 1去中心化 + 广泛参与 2创造正向飞轮效应 3可持续、可扩展、低壁垒 总结 数据收集之所以是物理智能的底颈，是因为物理世界交互的固有特性（真实、昂贵、慢、多样）与数字世界的“免费海量”完全不同。激励机制则是最佳解决方案，因为它通过经济杠杆和集体智慧，把数据生产从“稀缺瓶颈”转变为“可扩展飞轮”，兼具效率、可持续性和创新性。 未来，随着激励驱动的数据网络成熟，物理智能将像数字智能一样迎来爆发式发展，最终让机器人等系统真正智能地融入并服务物理世界。 谢谢 @grok

Robotics is crypto's most obvious blind spot. In 2016 it wasn't even a category investors tracked. Now it's the 2nd biggest in private markets, ahead of fintech, at $263B. Q1 alone did $16B across almost 500 deals. Here's what almost nobody in crypto is pricing in...all that capital is about to hit a wall the traditional financial stack can't get past. A robot can't open a bank account. It can't hold a legal identity, sign a contract, or get paid through ACH. Every rail in TradFi assumes a human sits behind the transaction. That works when a person signs for a few thousand machines. It breaks when millions of them start transacting on their own. Robotics routes through crypto because crypto is the only stack ever built for non-human actors: • Permissionless identity a machine can hold without a passport or a corporation • Programmable payments that settle machine to machine, no bank in the loop • Machine-readable ownership, so an asset can belong to something that isn't a person • Coordination between autonomous agents with no legal entity in the middle You could always make the argument Robotics doesn't need crypto. But I've always focused on following where the world's attention and hype lies, and Crypto tends to find a way of monetizing that mindshare. Where I'm looking Here's who's building those rails, grouped by the layer of the stack they're solving. Not a buy list btw...just a watchlist of which piece of the machine economy each one is going after. • Coordination. Virtuals (@virtuals_io) runs the biggest AI agent ecosystem and is now pushing those agents into the physical world. Its Eastworld robotics track covers teleop data, simulation, and real robot deployment. 50k+ agents recently got access to 430+ tokenized equities through Ondo / Treasures. • Location. Robots are useless without precise positioning, and GEODNET (@GEODNET) supplies it. 21k+ stations across 150+ countries, roughly $6.7M in annualized revenue and $5.3M in annualized holder revenue through buybacks and burns. • Identity and payments. Fabric (@FabricFND) handles the backend: machine identity, wallets, payments, and task coordination. Plumbing for a world where machines transact onchain without a human in the loop. • Telemetry. IoTeX (@iotex_io) has been grinding on DePIN and machine data for years. Device identity, verifiable offchain compute, real-world machine data. Not flashy, but it's what matters when robots need telemetry anyone can trust. • Spatial. Auki (@AukiLabs) is building Posemesh, a spatial awareness network that helps robots, smart glasses, and other devices understand the physical spaces around them. • Ownership. xMAQUINA (@xmaquina) wraps physical AI exposure in a DAO. Fractional machine ownership and robotics deals that normally sit behind VC rounds, backed by an $18M treasury with $7.6M set aside for robotics investments. • Simulation. Strike (@StrikeRobot_ai) works on humanoid autonomy and sim-to-real training. They recently rebuilt a real Eastworlds lab inside a simulation to test whether behaviors trained in sim survive contact with the real world. • Training data. Caspius (@caspius_ai) is after embodied AI data. Robots learn from real-world task footage, and Caspius just hit an ATH in footage contributed while adding NFT/account pairing. • Tooling. Roba (@Roba_Labs) wants to make robotics easier to build around. No-code tools, simulation, robot templates, a creator marketplace, and compute for training and inference. The catch I'm not going to pretend this is a rotation you front-run next week. Most of these are early, thinly traded, and completely dependent on physical AI actually scaling. If robots stall, these rails don't matter yet. But the capital is already moving, and it's moving toward a category that structurally needs crypto rails to work. I'd rather map this corner early than chase it once it's obvious.