Over the past month, we have continued pushing forward across growth, data production, commercialization, and ecosystem expansion. On the growth side, Axis Hub has now surpassed 133K total registered users, up from 80K+ about a month ago. Within this user base, more than 40,000 contributors have passed Axis’s quality-contributor filters. Among this qualified-contributor cohort, we currently have 13,480 MAU and 9,409 WAU, representing a 69.8% WAU/MAU ratio, which demonstrates strong recurring participation. Our weekly active users are contributing at a very high frequency, averaging around 9–10 valid data trajectories per user per day. We are also seeing a group of emerging “human experts” within the contributor network who are able to use our post-training features to correct model failure states and help models continuously improve and iterate. On the data side, Axis has now generated more than 3.4Mn valid simulation data trajectories, up from around 2Mn one month ago, representing approximately 9,500 hours of precise trajectory data. In addition, we have captured more than 150,000 cumulative hours of ego-centric real-world data, and 100 hours of loco-manipulation data.

#Unitree# lists on the STAR Market tomorrow — and just ahead of it, unveiled a humanoid robot called "Superman." Per Unitree's Aug 17 demo, the robot cleared a 2-meter standing jump and hit 12.66 m/s on a 0.85m leg — figures the company says exceed the human records for standing jump and running speed. The first humanoid-robotics listing is a moment worth zooming out for. Physical AI isn't one company — it's a whole supply chain, from actuators and sensors to the vision and control models behind them. That's the map worth having. NFA. For informational purposes only.

StableStock · July build in public Market got hit — semis + KOSPI crashed, and our spot volume went with it: Jun $120M → Jul $82M But the part that matters held: ✅ Revenue~$300k ✅ 2nd month in a row at breakeven Slow month on the screen. Heads-down building behind it. Shipping next: 🇰🇷 Korean stocks — August. Stablecoin, straight into Korea. 📱 App + funding-rate vault — next 60 days more markets loading Reach the world's real markets with stablecoins. That's the whole game.

SK Hynix reported Q2 on July 29: operating profit ₩60.5T, up 557% YoY, a record 76% margin — yet ~6% below consensus. The stock fell 9.6% that day. Within 48 hours, three major Korean brokers looked at the same report and gave three completely different answers: ₩4.7M, ₩2.8M, ₩1.48M. That's a 3.2x spread on one stock. How? KIS (₩4.7M, Buy): the miss was a timing issue — high-value shipments slipped into H2. Q3 DRAM prices now seen up 20% QoQ, HBM4 ramping, five-year LTAs locking in ASPs. Raised earnings estimates 10-11%. Mirae Asset (₩2.8M): didn't touch earnings estimates at all. Cut the multiple on one thing: China's DUV lithography localization clouding the long-term supply picture. Even says the selloff is overdone. BNK (₩1.48M, Hold): the miss was the start of demand slowing. NAND spot falling since April, hyperscalers pivoting capex to cost-efficiency, debt-funded expansion into softening demand. Oversold, but rebound capped. KIS bets on price. Mirae worries about structure. BNK questions demand and funding. The entire spread sits on four events nobody knows yet: Q3 DRAM +10% or +20%? When does NAND spot bottom? Does hyperscaler capex actually pivot? When does China DUV hit real supply? I'm with KIS on this one. Two things the bears are getting wrong. On LTAs — yes, locking 60-70% of capacity into long-term contracts means giving up some margin versus today's spot DRAM prices. That's real, and it's also the point. You trade peak-cycle margin for years of revenue visibility. Short-term pain, structural repricing. The market is still valuing Hynix like a cyclical while the company is busy becoming a growth stock — and the HBM roadmap is the core of that transition. A memory maker running flat-out on HBM4/4E with pre-sold capacity isn't the same asset class as the Hynix of 2019. On China DUV — the market is way too generous with the timeline. Shipping a domestic lithography tool is not the same as producing competitive memory at scale. Yield ramp on advanced nodes takes years, not quarters, and the gap compounds at every generation. Real supply impact on Hynix's earnings inside this cycle: minimal. Cutting a target multiple today for supply that doesn't exist yet is pricing a 2029 risk into a 2026 stock.

Same SK hynix, but its US ADR trades up to 37% above the Seoul shares — and no one can arb it. ADRs convert into local shares, but local shares can't convert back without company consent. Only new ADR issuance closes the gap. TSMC's ADR premium has held ~19% for years.

Leverage without the interest bleed Open and close within the day — pay 0 interest Only overnight positions accrue 700+ global names, up to ~10x →

stablecoin × stock — composability, realized. Short stock perps on Binance / Hyperliquid, go long the spot leg on StableStock. We've nailed the hardest part — the spot leg: Zhipu / MiniMax / 2x Hynix, Samsung + hundreds of US equities. One-click stablecoin allocation, low-rate leverage, zero spread. Institutions/whales — DM us on more margin size, higher leveraged, T+0, lower fees.

1/ RWAs are the logical conclusion of stablecoins. I've written a new piece that: - Proposes a taxonomy for the different types of RWAs - Discusses the adoption sequence of RWAs - Outlines the issues that are preventing the continued adoption of RWAs 👇

Cumulative trading volume achieved $100m, while AUM is around $19.3m! If includes our ipo subscription volume, total volume achieves around $160m today. It spent 5 months in the first $50m trading volume, but only cost 40 days in the second $50m trading volume. It spent 5 months in the first $10m AUM, but only cost 30 days in the second $10m AUM.

Stablestock is going through a huge momentum growth. We will launch more interesting and new facilities in the few 2 months about stock trading. Apr 6x than March, May will 2x than April.