Vincent Jow

Vincent Jow@vincentjow_ · 2026/04/28政策影响

10 BTC miners pivoting to AI infrastructure. The cheapest trades at $3M per MW. The most expensive trades at $10M. We published our memo and live comps with @artemis to standardize the comparison. Comps: https://t.co/iZnTqevg2h Memo: https://t.co/WSu6mSBiTI Demand for AI is accelerating and power is the bottleneck. BTC miner conversion represents both the fastest and the cheapest way to bring power online. But the data is messy. Disclosures differ across operators and prices are volatile with 10-30% short interest. We've created a live dashboard to frame the opportunity and compare operators on the metrics that matter. -- LEGAL DISCLAIMER: This post is for informational purposes only and should not be relied upon as investment advice. This post is not a recommendation for any security or investment. Please see full post on Modular site for all disclaimers, sources, and supporting material

AI：偏政策推进，强调合规落地与规则明晰，对监管主线更敏感。

Vincent Jow@vincentjow_ · 2026/01/15观点输出

https://t.co/tepLyahL4T

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Vincent Jow@vincentjow_ · 2026/01/08市场影响

If a team isn’t excited to buy their tokens at these levels, why should investors be? The point of token buybacks is to signal commitment to tokenholders. If you believe your token is undervalued, you should be excited to buy it. Growth should always come first, but it’s striking how quickly projects abandon buybacks altogether just because it’s “not working”. There are many reasons why your token price is not going up, but buybacks is certainly not one of them. My advice - Make the commitment only when you are ready and able to stick to the commitment. It doesn’t need to be 100% of revenues but it doesn’t need to be 0% either. Set price-based guardrails to prioritize capital allocation elsewhere when needed. And of course never do it if you don't have real, organic sustainable revenue yet.

AI：偏平台经营与生态扩张，强调交易平台和应用入口的长期位置。

Vincent Jow@vincentjow_ · 2025/05/27赛道影响

Geodnet continues to be one of the few - and most exciting - public ways to make a thematic bet on robotics and humanoids. Rather than betting on specific humanoid co (there are 200+) or the complex supply chain (e.g. debating actuator designs, in-house/outsourced, China/US etc.), GEOD is a pure play robotics vision & positioning platform 1. Strong core GPS business (e.g. drones, tractors) doing $4M of ARR (4x y/y), 80% buy/burn & $50M pipeline. One of the few DePIN networks with real, organic demand 2. Has just expanded to the full suite of robotics positioning technologies to include both LIDAR & Computer Vision, which will power humanoids & autonomous vehicles, w/ ROS integration (used by nearly every humanoid co). Already working with Unitree, Xpeng, etc. 3. Building an all-in-one custom positioning chip to fuse GPS, LIDAR, and vision on-device - unlocking low-cost, high-performance positioning at scale

AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。

Vincent Jow@vincentjow_ · 2025/05/21观点输出

RT @jamesjho_: Sharing our @pendle_fi thesis co-authored with @TheSpartanGroup @tumilet Stables are a huge growth area for crypto Pendle…

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Vincent Jow@vincentjow_ · 2025/05/14政策影响

Stablecoin Expansion ↔️

AI：偏政策推进，强调合规落地与规则明晰，对监管主线更敏感。

Vincent Jow@vincentjow_ · 2025/04/21观点输出

RT @jamesjho_: https://t.co/twbjeYtFFX

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Vincent Jow@vincentjow_ · 2025/03/03观点输出

The U.S. government holds: $770B in gold (~5% of $18T market cap) $19B in BTC (~1% of $1.9T) $100M in other crypto (0.1% of $1T excl. BTC/stables) Governments worldwide collectively hold 17% of total gold supply. Assuming a 10-20x inflow multiplier, if governments increase ownership to 5% of BTC, that implies a 40-80% price uplift. Polymarket odds of the U.S. adding BTC to reserves in 2025 jumped from 40% to 60%. 20% probability increase x 40-80% = ~10-15% move

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Vincent Jow@vincentjow_ · 2025/02/25观点输出

RT @shayonsengupta: 1/ we're manufacturing consciousness and embodying it in machines. this is no time to stare at charts and feel dejected…

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Vincent Jow@vincentjow_ · 2025/02/10赛道影响

The next and final stop for Web3 AI is physical AI & humanoids. Web3 x AI attacked some of Web2 AI's biggest challenges: - The GPU shortage fueled the rise of Web3 GPU marketplaces (Render, Akash, Ionet) - Data was the oil to power AI, creating crowdsourced data marketplaces (Grass, Vana) - AI agents couldn't experiment freely on social & financial rails, which created Web3 agents (Freysa, Virtuals, Eliza) Labor is the largest cost in the global economy, accounting for 30% of global GDP (~$30T). As humanoids begin to outnumber humans, this structural shift will play out over decades. We're already at the inflection point, with the biggest companies and venture funds aggressively investing: - February: Figure raised $675M at $2.6B (Amazon, Nvidia) - March: Nvidia unveiled GR00T at 2024 GTC Keynote - May: OpenAI rebooted its robotics research team (previously shut in 2021) - July: Skild AI raised $300M at $1.5B (Lightspeed, Coatue, Softbank, Amazon) - November: Physical Intelligence raised $400M at $2B (OpenAI, Sequoia) This is happening now. Humanoids from Figure, Digit, Apptronix & Walker are already operational in BMW, Amazon, GXO, Mercedes & BYD factories. They will start replacing jobs that are physically intensive, boring/repetitive, dangerous and/or labor constrained before moving further down the stack. Figure x BMW Update: https://t.co/pAhKR2TOGU Mass production is coming. Tesla aims to produce 10K humanoids this year, ramping up to millions per year. Figure is aiming to ship 100K humanoids in the next 4 years. 1x is scaling to millions by 2028. Elon has called out Optimus as a $10T+ revenue opportunity with 10 humanoids to every human. Web3 will play a key role in the rise of humanoids. As we move from specialized to generalized robotics, data collection will be key. Unlike LLMs, which were able to train their models on the entire internet, we will largely need to create this dataset from scratch. We are already seeing Web3 companies aiming to solve the data bottleneck. Unlike the typical Web3 value proposition of cheaper costs, these companies offer unique, irreplaceable datasets, shown by the caliber of their early partnerships: - @reborn_agi is doing advanced motion capture and is working with several of the leading humanoid companies - @MeckaAI is producing data using smartphones and has garnered the interest of @1x_tech - @frodobots has produced a fleet of sidewalk robots and is working with @GoogleDeepMind and leading robotics researchers from UC Berkeley, UT Austin, Carnegie Mellon University, etc. - @PrismaXai just emerged from Stanford Blockchain accelerator - Existing companies like @Hivemapper and @NATIXNetwork are also collecting useful data Another key challenge is real-world navigation for robotics: - @geodnet is powering leading robotics companies like @UnitreeRobotics with GNSS which allows them to seamlessly navigate in the world with <1cm accuracy. GEOD is also collaborating with other partners to support remote management & data collection - @AukiNetwork is building a network perception map for robots to interact with physical spaces, starting with grocery chains in Europe Humanoids are coming and Web3 will ensure they scale.

AI：偏平台经营与生态扩张，强调交易平台和应用入口的长期位置。