Thesis on $HAN @HumanoidNetwork, launched just over a day ago --Introduction-- The first Robotics project on @RobinhoodCrypto launched through Virtuals. They’re building a decentralized data network for training robots. The premise ; physical AI is bottlenecked not by algorithms but by motion data (robots don’t know ‘how’ to move) & since robot demonstrations are expensive, and collecting human demonstrations globally means dealing with payroll and employment law in every country. Their answer is to replace hiring with a token-coordinated protocol, so a contributor in Lagos or Manila submits through the same pipeline as one in San Francisco. “As global labor shortages accelerate, robotics is becoming the inevitable future of work. But while robot hardware has advanced rapidly, the real bottleneck lies in software: very few are building the intelligence that runs these machines.” It’s easy to train an LLM because the ‘complete’ data is instantly available all over the internet (just text). It’s hard to train a robot exactly how to move or act in certain situations because for example a warehouse robot needs tens of thousand of hours of high quality data from actual warehouse workers. Humanoid is building a solution for exactly that & through the token-coordinated protocol they can reach a much bigger & variational target dataset than any other competing company can. --Team-- There’s still little to find about the team online but they announced their first advisor today ; Mark Desantis : Mark is a serial tech entrepreneur who has built and sold several AI-based companies (two to Global 1000 companies) across a variety of industries. He is also an adjunct professor of entrepreneurship at @CarnegieMellonUniversity and an adviser to several tech companies and Cybernetix Ventures. He was a senior policy analyst in the @WhiteHouseof Science and Technology Policy, Mark brings deep expertise at the intersection of technology, public policy, and entrepreneurship. Most recently, he is the coauthor of The Real AI: Creating Infrastructure and Cognitive Capital to Real Work to be released in the Fall 0f 2026 by Taylor and Francis Publishing and has a companion Substack on AI: Real AI = Real Work. Mark replied with his LinkedIn account to the LinkedIn post of this from Humanoid so it’s highly likely legit. Also important to mention, they were quiet on their X account for a few months. They had a points program for early supporters that they eventually retired. Those early supporters will be rewarded with ‘community initiatives’ now, they’re using specific legal language on this. My guess -> they’ve been building in stealth for some time & been working with Virtuals in the background on this. --Tokenomics-- The token launched through the anti-sniper protocol from Virtuals which has a decaying buy/sell tax from 99% to 1% over 98 minutes. This makes it so no one can snipe/bundle and price discovery happens organically. All collected fees were used for buybacks, which here was 380K worth of tokens bought over 24h after launch. The circulating supply (the rest is locked for the team) is kinda hard to calculate. It started at 18%, the ACF then gradually adds supply into circulation as price goes higher (until now it added about 9%), then the buyback wallet took almost exactly 5% of the total supply out (the total supply, not 5% of the circulating supply!). This calculation, among the tokens locked for the team on Virtuals brings us to a total of 22% of supply circulating. You can check this on the Virtuals page of HAN. The utility on the token is interesting & there’s a lot of legal protection on it language wise in the whitepaper. Section 10 lists these five, all hedged with "may": 1. Protocol access —> using approved protocol functions. 2. veHAN participation —> lock $HAN to get non-transferable participation power. Longer locks = more weight. But the v0 vault explicitly has no reward payout function, no early unlock, and doesn't burn principal. veHAN can activate reward-eligible submission lanes and gate the review network. 3. Motion Credits —> buy non-transferable capacity credits that raise how many reward-eligible submissions you can make per epoch. They don't buy rewards directly. Purchases route HAN into burn/sink, contributor, compute, anti-abuse, infrastructure, or future staker pools. Notably the paper hedges that the canonical HAN contract "may or may not expose a burn() function," so burn accounting has to support transfers to a dead address as a fallback. 4. Enterprise settlement —> enterprises can pay in fiat or supported protocol assets. Any path from enterprise revenue into protocol rewards needs separate approvals. 5. Future governance —> if approved. --TA-- The token 'graduated' at 4.3M after the anti-sniper organic price discovery model ended. During the anti-sniper protocol there was 3M+ in volume, once the token graduated from the sniper protocol, it generated 5M+ in volume until now (which is like 29H since graduation) Currently it’s ranging between 5M <-> 10M FDV & already had two 50%+ pullbacks, (ps pow rant & tendies rotation). Beside that, holding up very well for a first day after TGE. --TLDR;-- First mover in Robotics on Robinhood, first advisor announced is ex white house, very good supply control, professional communication & updates, strong public support from Virtuals (seems like s-tier project), chart consolidating between 5-10M FDV atm Personally, I’m very bullish on this one. Higher it ps: check out the robot on the website it's really cool