Jason Guo @StarSpaceVC

关键是今天就事先说观点，而不是事后解读 K 线图。

《 AI 有泡沫吗》 ◆ 定价权博弈和撕扯 ◆ 华尔街老登资金（达利欧逻辑）vs 科技新钱（木头姐逻辑） ▷ 达利欧看：债务/利率/周期/历史数据， ▷ 木头姐看：生产效率/Token 需求/营收增速 ◆ 结论观点： ▷ 新旧时代博弈，老钱想仍用旧逻辑收割新筹码 ▷ AI 下半场才刚刚开始 ◆ 《StarSpace 投资逻辑》： ▷ 是否好生意 ▷ 市场竞争力 ▷ 客户迁移成本 ▷ FCF 现金流 ▷ 股东回报 ▷ 趋势周期节拍

keyboard_arrow_left Previous keyboard_arrow_right Next

Gavin Baker on the other side of Ben's bandwidth story: at the peak of the 2000 bubble, 97% of America's fiber sat dark, and today there are no dark GPUs. "I had, depending on how you look at it, the privilege and the misfortune of being a tech investor during the year 2000 bubble, which was really a telecom bubble." "The year 2000 bubble was defined by something called 'dark fiber.' Dark fiber was literally fiber that was laid down in the ground, and not lit up... At the peak of the bubble, 97% of the fiber that had been laid in America was dark." "Contrast that with today. There are no dark GPUs. All you have to do is read any technical paper, and one of the biggest problems in a training run is that GPUs are melting." @GavinSBaker

a16z @a16z

2 days ago

Ben Horowitz on the lesson LoudCloud, his early cloud company, taught him about today's AI neoclouds, and why they now face the opposite problem: "It can end up being like a very dangerous arbitrage business, where you can get upside-down easily. And so if you look at LoudCloud,

32 22 258 183K 69

特斯拉新 Roadster 发布：靠压缩空气喷气在空中翻正落地，马斯克说 2028 年投产，售价未公布 10 月 15 日晚，特斯拉在得州 SpaceX 的 McGregor 试验场直播发布了第二代 Roadster 跑车。这辆车 2017 年就亮过相，当时说 2020 年交付，之后一推再推，拖了将近九年。 演示里，车开上一段弯曲的大坡道，倒挂着往上跑，然后从坡道尽头冲出去，在空中把车身转回正面，四个轮子落地。之后它又从一排特斯拉车（包括 Semi 电动卡车）上方翻跳过去。马斯克在现场说：“飞行汽车是真的。” 它靠的是冷气推进器。原理跟吹满的气球一松手就乱窜差不多：碳纤维高压罐里存着压缩空气，从喷口猛地放出去就产生推力，不烧燃料，也没有火焰。特斯拉说推进器的推力是轮胎抓地力上限的四五倍，所以它起步能比只靠轮子快得多。倒挂不掉下来靠的是另一套系统：风扇把车底的空气往外抽，把车吸在坡面上。 所以这里的“飞”只是几秒钟的喷气跳跃，气放完就没了，离能载人飞一段路的飞行汽车还远。据科技媒体 Glitchwire 报道，演示车把后排座位换成了一个装 260 公斤压缩空气的高压罐，车身四周有 10 个喷口；The Information 此前报道，因为推力和噪音太大，演示车是远程遥控的。这几点特斯拉官方没有确认。 翻跟头的是最激进的 Rocket 版：单座，只接受邀请购买，不能上公路。零到 60 英里/小时（约 97 公里/小时）加速不到 0.59 秒，极速超过 370 公里/小时。 普通人能订的是另外两个双座版本。Performance 版没有推进器，靠主动空气动力学和风扇吸地增加抓地力，加速不到 1.6 秒，订金 5 万美元。Thrust 版在车尾加了推进器，加速不到 0.9 秒，限量发售，订金 25 万美元。三个版本续航都超过 480 公里。 特斯拉没公布售价，订金不等于车价，2017 年说的起步价是 20 万美元。Performance 版的 5 万美元订金分两次付：先付 5000 美元（特斯拉说可全额退），10 天内再补 4.5 万美元，否则订单取消。 投产时间是马斯克在现场说的 2028 年，特斯拉官方帖子里没有写日期。这辆车的交付时间此前已从 2020 年先后推到 2021、2022、2025 和 2027 年。 x.com/Tesla/status/2…

今天早上我突然想明白了我一直悬在心里的一个疑问： 当年， Meta 小扎在 Alexandr Wang vs 杨立昆之间，到底选谁？为什么？结果会怎样？ 一位是 Alexandr Wang ，是美国AI独角兽公司 Scale AI 的创始人， 一位是杨立昆，图灵奖得主，Meta首席科学家， 应该说两位都是当今全球顶级 AI 大神，都是大名鼎鼎。 但对于 Meta 来说，当时的 Meta 正面临公司整体战略的 “纠错+转型” 的关键时刻，在这十字路口，千亿美元的赌注到底应该押向谁？ 现在大家都看到了结果。 昨天，看了小扎和Alex Wang的一系列访谈内容，及杨立昆的创业融资故事，结合 Muse 的异军突起， 今早我突然间明白了其背后的逻辑，悟道了： ▷ 应该如何投资创始人 ▷ 扛大旗的领军人最宝贵的是什么 ▷ 以及AI时代，PK的是什么 ▷ 怎样的人会笑到最后 🙏🙏🙏

keyboard_arrow_left Previous keyboard_arrow_right Next

📣 GoodVision AI is featured in @TechCrunch. “The Hidden Decisions Behind Every AI Request” explores how intelligent routing and compute infrastructure are shaping the future of AI inference. Read the full article 👇 techcrunch.com/sponsor/goodvi… #AI #AIInference #AIAgents

.@flexport CEO Ryan Petersen on the Charlie Munger advice he passes on to AI founders: "The key to success is dumb competition." "One of my favorite things that Charlie told me at one of these dinners at his house was when I told him about Flexport, he was like, 'Oh, you have a great business because the key to success is dumb competition.'" "I always say that to people who are building AI companies. What a nightmare. Your competitors are literally AI geniuses." "You should find a way to apply AI in some other domain so you're not competing with the smartest people on Earth." "It's like trying to be a boxer. You might beat all the people in your local gym, but at some point you're gonna run into somebody who's gonna kick your ass." @typesfast w/ @shaneparrish

实战认知👍👍👍

《一级市场投资看趋势，二级市场投资看博弈，那么未来投资看什么》 1、 前言： 一级市场投资看趋势，二级市场投资看博弈，那么未来投资看什么？特别是未来 AI主旋时代，想获得超额投资回报，到底投资主题和逻辑是什么？ 2026年以来，全球市场极不稳定，无论是地缘政治，大宗商品、还是AI技术迭代，或是全球资本市场（特别是股票市场），不确定性颇高且频发。 这段时间，我静下心来，复盘本人这15+ 年来全球硬科技、一级二级市场的投资实战经历，思考了一个有意思的问题： 一级市场投资是看项目趋势，预判 0→1、从沙漠起高楼的可能性； 二级市场投资是看博弈，寻找被错杀、被低估的机会，或是预判泡沫逃顶的能力。 二级市场要知道对手是谁，分析他们的逻辑，盯住资金流向； 而一级市场，有时甚至不一定知道对手是谁，关键要看懂趋势，敬畏市场。 那么未来投资看什么？特别是未来 AI主旋时代，是看从“比特”到“原子”跨越机会吗？ 是看能源电力、算力数据、硅光模块通信、Physical AI、AI具身机器人、AI生命科技、土地、太空吗？ 我预判都不是。 想获得超额投资回报，到底未来投资主题和逻辑是什么？ 2、一级市场投资看趋势 就我本人而言，前10年基本是在一级市场做前沿科技的天使投资，后面大概5年投二级市场，主要投美港A股市场。 一级市场的投资逻辑主要是看趋势。 无论从周期、趋势、节奏，还是项目团队、基本面来看，核心是看趋势、看人，做价值投资。 是要有前瞻性，预判这个团队在沙漠上起高楼的可能性，可以建多高，风险在哪里，以及如何规避。 想起来，我前10年一级市场投资时间里，分别天使投了：互联网社交、服务类机器人、区块链及Web3、数字孪生，医学设备、太空软件等早期项目，应该说体感还是比较有代表性的。 总体来说，一级市场早期投资确实不容易，一级市场的 “看见的能力”，都是买单买出来的。 现在，发现不少以前做 PE的、投行券商、律师会计师、技术专家、政府机构等，都延伸做 VC，只是看到回报，没看见风险。 3、二级市场投资看博弈 因为一级市场很多众所周知的原因，我们后5年关注二级市场多些，更多是关注美股。 2023 年我们出手买了英伟达 NVDA、特斯拉 TSLA，以及 OKLA 小型核反应项目；之后半导体、算力、光模块、能源、脑机，到现在的太空数据中心、电力基础设施、Token 推理模型等等。一句话，All in AI，但整个板块轮动，还是蛮惊心动魄的。 我感觉越是波动大，越看你的理解和洞察能力。 二级市场 AI板块，我还是蛮看好的。 但我有一个感觉：二级市场 vs 一级市场，投资逻辑和策略，完全不一样。 二级市场是实打实地看博弈，有做空 vs 做多，有量化，有对冲，有期权，有庄家 vs 散户，有时还牵涉到多国市场联动、汇率利率、股债期货、石油黄金。。。 你都要在博弈之间，去找一种平衡，找一种错杀，找一种抄底，找一种逃顶。 其实，你赚的钱就是博弈的钱。 再从现实社会来看，两党也好，两院也罢，地缘政治也好，森林法则也罢，本质上都是一种博弈的过程，都是利用人性的贪婪和恐惧的原罪。 4、未来AI时代想获得超额投资收益，投资主题和逻辑到底是什么？ 现在主流媒体、大咖、投资人的共识看法，主要趋势有几种观点： 能源电力 算力芯片 光模块传输 Physical AI /世界模型 AI 具身机器人 比特 vs 原子 AI 模型/网络安全 太空 vs 地球 AI vs 人类 AI 永生/生命科学 媒体关注的大多是这些方面，但我个人认为以上都不是，那究竟是什么呢？ 在我看来，这些都很重要，都是瓶颈， 但，这都只是一个过程，或某阶段遇到的瓶颈。 一旦跨过这些瓶颈，它们只是一个过程，不能成为长期投资逻辑。 例如：Physical AI，物理/世界模型、数据、具身机器人，我觉得数据、功能稳定、量产、安全、法律。。。等解决，至少要 10+年以上； 例如：AI 芯片内存通信，时间会解决的； 例如：AI永生，更是一个伪命题，整个社会、环境、资源、法律，整个地球都没有准备好，甚至说是种灾难； 再例如：网络安全，我觉得这是一个非常严肃和严重的话题。AI安全 vs 传统互联网安全，完全不是一回事。传统网络安全，是防止病毒进入，而AI安全，要防止的是递归进化，内生的安全问题。手段、逻辑、软硬件等都完全不一样。 但 AI 网络安全问题，迟早会解决。 那么，最终投资主题和逻辑是什么呢？ 1、数据主权和控制力 or 2、人类 vs AI 机器人大战 不存在中间态。 我个人预判： “数据主权及控制力”， 会成为AI发展的长期投资机会。 Jason Guo 2026-10-12，于深圳

@vistword 认同👍

I am the proudest mother in the world. @elonmusk received the National Medal of Science from @POTUS Other awards were for @JensenHuang @MichaelDell @satyanadella @LisaSu and Serge Brin 🏅🏅🏅 👏👏👏

keyboard_arrow_left Previous keyboard_arrow_right Next

Peter Thiel on how you should think about your future: "What you end up doing when you think of the future as indefinite is you pick a job that you think will look good on your resume, because it will lead to a different job later on. There's even a credentialing version where you go to grad school, you get a consulting job, and you think of it as this branching tree diagram where you never really do something that you want to do. You can think of the future on three horizons. Very short-term is: are you going to be happy, and will you be learning a lot? Medium-term is: how good is it on your resume? And very long-term is: are you working on something important or meaningful? We're in a skewed world where we overweight the medium-term and underweight some combination of the short-term and the long-term. I would recalibrate to focus on the short-term and the long-term, and forget about the medium-term."

英伟达上周又历史新高了 我看B站除了好多恶搞视频 不得不说，现在Ai做视频，基本看不出来痕迹 太夸张了

Robots are 3D printing full-size ship hulls. ㅤ CEAD’s Faber Navalis system uses robotic arms to print complete marine hulls up to 12 m long and 4 m wide in a single automated run—no molds, no segmentation, and no heavy manual assembly. ㅤ A 12-meter RIB-style prototype for the Royal Dutch Navy was printed in 8 days, tested during NATO’s BOMA exercise, and sailed for four days with the Navy in Den Helder. Traditional delivery timelines for similar vessels often run into years. ㅤ The hulls use glass-fiber-reinforced HDPro, roughly 10% lighter than comparable fiberglass while offering higher impact resistance than standard HDPE, plus UV stability and recyclability. Primary targets are USVs, RIBs, workboats, and patrol craft. —— Weekly robotics and AI insights? Subscribe free: 22astronauts.com

《The Truth Behind Dead Fly Resurrection and Some Reflections》 ★ The recent viral episodes on social media — “a fruit-fly brain playing games / controlling a robot” and “resurrecting a dead fly” — have blown up. ☆ My tentative conclusions: 1. It is largely a commercial stunt; 2. There is a real scientific advance, and it is worth thinking about more carefully. ★ I have said repeatedly over the past six years (screenshots available): to understand AI, someone first has to have a human brain neural map (connectome). ☆ At the time, however, that claim was limited to what optical-microscope sectioning could achieve. ☆ Today, scientists (Google Research and others), using electron-microscope images and AI, have fully reconstructed the central nervous system of an adult male fruit fly (MaleCNS): 166,700 neurons and roughly 125 million synaptic connections. ★ My inference: AI did not only help rebuild the biological brain map; it also helped the developers write the code. After decoding, that AI brain map was made to run, and via WiFi it plays games and controls a physical robot dog. 【StarSpace observations and reflections】 1. The principle by which the brain forms memory is that discharges across neuronal synapses generate signals; repeated enough times, those connections become memory. 2. Biological brain maps should therefore differ from one another — a thousand brains, a thousand wiring diagrams. 3. What this experiment most likely shows is that human developers defined how external information is fed into the brain map, read the feedback signals, and then translated those signals into specific movements of the robot dog. 4. That raises two key questions: 4.1、Is the AI-reconstructed brain map (the biological wiring diagram) one-way or two-way? Does that directly show whether “helping sleep” is a false claim? 4.2、Given the same input signal, do different fruit-fly brain maps produce the same output? For example, if the same “turn left” signal is fed in, does the robot dog always turn left? 5. As human understanding of AI deepens, and once a human brain neural map exists, reading AGI / ASI becomes only a matter of time.

keyboard_arrow_left Previous keyboard_arrow_right Next

This was inevitable. That's why @world_id for @Zoom exists. world.org/solutions/worl…

Tavus @tavus

3 weeks ago

Introducing Griffin, the first model to pass the video Turing test. 48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video. It’s the first Human Interaction Model (HIM).

3K 4K 39K 21.1M 27K

x.com/i/article/2100…

The future is not set. There’s no fate but what we make for ourselves. @Figure_robot just dropped a post showing its F.02 humanoid robots jumping into a 75-ton electric arc furnace in Imatra, Finland. The U.S. facilities turned them down over battery fire risk, so Finland got the honor. The idea came from Arnold Schwarzenegger after a poll on social media. The recovered metal is being sold as limited-edition artifacts on Figure’s Shopify store. But the post that actually matters came two weeks earlier, on September 17. Helix 2.5. A single foundation model, pretrained on Figure’s Index dataset of human behavior videos, walked into 30 Bay Area homes it had never seen and started working. No data collected in any of them, no fine-tuning or adaptation. Three tasks: making beds, folding towels, tidying living rooms. 420 trials, all-or-nothing scoring, no partial credit. 237 completed. 56 percent. The control group tells the story. An identical model trained from scratch on the same task data managed 9 percent. Index pretraining was the only variable, and it moved the needle more than six times. Bed making hit 67 percent. Towel folding 62. Toy tidying, the messiest task, 40. It also matched a Helix 02 policy that had been trained directly in the evaluation environment, while using half as much task-specific data. Index now generates about 35 minutes of new human experience every second. Wanna talk numbers? Figure has committed $3.5 billion of compute to Helix. The F.02s were the old generation. Helix 2.5 is the new one. 👉 I write about robotics & AI. If it’s interesting in robotics you’ll hear it here first —— Weekly robotics and AI insights. Subscribe free: 22astronauts.com

This is the only thing that survived