Chubby♨️@kimmonismus · 4天前53

Intel shares jumped on the report that Google and Nvidia are testing it as a backup foundry, with Google already ordering more than three million TPUs for 2028. The move is less about an Intel turnaround and more about TSMC's capacity crunch, since its leading-edge and advanced-packaging lines are fully booked in the AI boom. Intel's opening is EMIB packaging plus early 18A trials, which would make it a credible second source if SK Hynix's memory tests work out. Do we see a big intel comeback?

译Intel shares jumped on the report that Google and Nvidia are testing it as a backup foundry, with Google already ordering more than three million TPUs for 2028. 英特尔股价因报道称谷歌和英伟达正在测试其作为备用代工厂而上涨，谷歌已为2028年订购超过300万个TPU。 The move is less about an Intel turnaround and more about TSMC's capacity crunch, since its leading-edge and advanced-packaging lines are fully booked in the AI boom. Intel's opening is EMIB packaging plus early 18A trials, which would make it a credible second source if SK Hynix's memory tests work out. 此举更多是由于台积电产能紧张（其先进制程和先进封装线在AI热潮中已被完全预订），而非英特尔翻身。英特尔的机遇在于EMIB封装加上早期18A试验，如果SK海力士的内存测试成功，它将成为一个可靠的第二供应商。 Do we see a big intel comeback? 我们会看到英特尔大回归吗？

OpenBMB@OpenBMB · 4天前54

🚀 The Build Small Hackathon is officially live! Check it out here: 🔗 http://build-small-hackathon-field-guide.hf.space/partners/openbmb We're proud to partner with @HuggingFace and support builders around the world with OpenBMB models. Here are some inspiring use cases built with our models: 🪐 MiniCPM-V 4.6 (image / OCR / multimodal) → Automated financial document analysis https://x.com/OpenBMB/status/2062889699056984281 ⚡ MiniCPM5-1B (lightweight, on-device) → AI Desktop Pet running entirely on your local device https://x.com/OpenBMB/status/2058903454308069670 🌐 MiniCPM-o 4.5 (omni-modal) → Real-time video analysis application https://x.com/OpenBMB/status/2022341243859337613 🎙️ VoxCPM2 (voice / TTS) → Voice cloning application https://x.com/OpenBMB/status/2041169065020936464 🏆 $10,000 OpenBMB Special Prize is available for outstanding projects built with our models. Get creative, try out our models, and see what you can build in this hackathon! We can't wait to see what you create. ✊ #BuildSmall #MiniCPM #OpenSourceAI #HuggingFace

译面壁智能（OpenBMB）与HuggingFace合作举办的Build Small黑客松正式上线，鼓励开发者使用OpenBMB模型构建应用。官方列举四大用例：MiniCPM-V 4.6用于自动化金融文档分析；MiniCPM5-1B轻量端侧模型驱动AI桌面宠物；MiniCPM-o 4.5全模态实时视频分析；VoxCPM2语音克隆。活动设有$10,000 OpenBMB特别奖。

OpenCode@opencode · 4天前55

OpenCode Go has crossed 100,000 subscribers now doing 1.5 trillion tokens per day

译OpenCode Go 订阅用户突破 10 万 现每日处理 1.5 万亿 tokens

X.PIN@thexpin · 4天前58

DeepSeek just took its first bite out of the US enterprise market, hitting #1 on Ramp’s leaderboard for first-time corporate buyers. But it’s still early days. As of April, DeepSeek’s adoption rate is just 0.1%.

译DeepSeek刚刚首次涉足美国企业市场，在Ramp针对首次企业买家的排行榜上登顶。 但为时尚早。截至4月，DeepSeek的采用率仅为0.1%。

X.PIN@thexpin · 4天前56

The price of humanoids in China just fell off a cliff! Unitree is facing a massive margin squeeze. Their avg. selling price keeps dropping: 🔹2023: $81.6k 🔹2024: $35.8k 🔹2025 (Q1-Q3): $23k Meanwhile, their unit cost only fell from $10k to $8.5k. Bumi, a fierce competitor in the humanoid market, just launched their cosumer-level robot at $1.4k after subsidies. The robot stands at 3′1″, aimed at education and companionship. Nowadays, core components like servo motors and reducers are mainly domestically manufactured. Robots are borrowing heavily from EV and smartphone supply chains, with batteries and vision sensors all repurposed from mature production lines.

译中国人形机器人价格断崖式下跌。宇树科技（Unitree）均价从2023年$81.6k降至2024年$35.8k，2025年前三季度进一步跌至$23k，而单位成本仅从$10k降至$8.5k，利润空间被急剧压缩。竞争对手Bumi推出消费级人形机器人，补贴后售价仅$1.4k，身高3英尺1英寸，面向教育和陪伴场景。核心部件如伺服电机、减速器已实现国产化，机器人供应链大量借鉴电动汽车和智能手机成熟产线，电池、视觉传感器均从现有产线复用。

Alibaba Cloud@alibaba_cloud · 4天前44

Ready to build next-gen AI agents and win $70,000+ in total prizes? 🚀 Qwen Cloud Global AI Hackathon is live! Harness frontier models, unlock global exposure, and ship your vision. Check the poster below for milestones. 🔗 Register now: https://click.qwencloud.com/m/20000000281/

译准备好打造下一代 AI 智能体并赢取总计超过 70,000 美元的奖品了吗？🚀 Qwen Cloud 全球 AI 黑客马拉松现已启动！利用前沿模型，获得全球曝光，实现你的愿景。查看下方海报了解里程碑。 🔗 立即注册：https://click.qwencloud.com/m/20000000281/

歸藏(guizang.ai)@op7418 · 4天前62

也是没招了，Notion直接点名 Anthropic Opus 4.7 和 Opus 4.8 模型性能下降的问题。 他们也澄清了，说都会存在这种问题，目前性能下降问题已经恢复。 但是 Anthropic 这个模型出问题的概率真是比其他两家高非常多。 我截了一下这三家的状态页面，Anthropic 甚至连 99% 的可用性都保不住。

译Notion AI 状态页面显示，Anthropic Opus 4.7 和 4.8 模型出现性能下降，导致用户选择这些模型时失败率升高。Notion 已禁用模型选择器中的所有 Anthropic 模型，并将请求路由至其他供应商，大多数用户可继续使用 Notion AI。Anthropic 澄清此类问题普遍存在，目前性能已恢复。但推文作者指出，Anthropic 模型出问题概率远高于其他两家，其状态页面甚至无法保证 99% 可用性。

Rohan Paul@rohanpaul_ai · 5天前61

Nvidia just cleared the memory bottleneck significantly for Vera Rubin by qualifying HBM4 from Samsung, SK Hynix, and Micron & moving those parts into full production. per news, Vera Rubin HBM4 share: SK Hynix 60–70%, Samsung 25–30%, Micron rest.

译NVIDIA 正式认证三星、SK 海力士和美光的 HBM4 内存，并投入量产，以解决 Vera Rubin 超算的内存瓶颈。据消息，Vera Rubin 的 HBM4 份额分配为：SK 海力士 60–70%、三星 25–30%、美光占剩余部分。SK 海力士与 NVIDIA 已达成多年合作，将共同开发 Vera Rubin AI 超算等平台的内存，并利用 CUDA-X、PhysicsNeMo 等工具加速芯片设计与半导体仿真。双方强调，先进 DRAM 与 HBM 须提前数年协同设计。

小互@xiaohu · 5天前71

Claude 现在写了 Anthropic 大部分代码 截至 2026 年 5 月，合并进 Anthropic 代码库的代码里，超过 80% 是 Claude 写的。 而在 2025 年 2 月 Claude Code 发布前，这个比例还只有个位数...

译Claude 现在写了 Anthropic 大部分代码 截至 2026 年 5 月，合并进 Anthropic 代码库的代码里，超过 80% 是 Claude 写的。 而在 2025 年 2 月 Claude Code 发布前，这个比例还只有个位数...

Berryxia.AI@berryxia · 5天前65

波士顿机器人直接要替代搬运工了吗？ 以后抗楼战神的活儿要被抢走了😄

译波士顿动力展示了其新Atlas人形机器人，它举起并搬运了一个50磅的冰箱。 机器人利用全身平衡并移动重物，动作自然且类人。 在现实世界中执行之前，Atlas在虚拟仿真中练习了数百万小时。 网友调侃：“以后抗楼战神的活儿要被抢走了😄”

Rohan Paul@rohanpaul_ai · 5天前55

In Seoul, Nvidia CEO Jensen Huang handed out SK Hynix x 7-Eleven HBM Chips snack bags while addressing the crowd. He teased, “No HBM for you since I need all the HBM” https://x.com/StockSavvyShay/status/2063266848540656018/video/1

译在首尔，Nvidia CEO 黄仁勋向人群分发 SK Hynix x 7-Eleven HBM 芯片零食袋。 他调侃道：“没有 HBM 给你，因为我需要所有 HBM”

Rohan Paul@rohanpaul_ai · 5天前61

SK hynix and NVIDIA just formed a multi-year memory partnership to build the chips behind the next wave of AI factories. So SK hynix is being pulled deeper into AI servers, personal AI, and physical AI. SK hynix will codevelop memory for NVIDIA Vera Rubin AI supercomputers, Vera CPUs, RTX Spark-powered PCs and Jetson Thor platforms, and will use NVIDIA CUDA-X, PhysicsNeMo, Omniverse, OpenUSD and cuOpt to accelerate chip design, semiconductor simulations and autonomous fab digital twins. Memory can no longer be designed after the GPU is finished, because advanced DRAM and high-bandwidth memory need years of co-design, fabrication planning, and capital spending. The partnership also moves AI into chipmaking itself, using CUDA-X, PhysicsNeMo, and simulation tools to speed semiconductor physics, lithography, and internal engineering code. SK hynix also plans fab digital twins, which are software copies of factories where robots, tools, wafers, and logistics can be tested before changes hit the real production line. It reinforces SK Hynix's position as a key supplier in the AI hardware supply chai

译SK海力士与英伟达达成多年内存合作，将为英伟达下一代AI超级计算机Vera Rubin、Vera CPU、RTX Spark PC及Jetson Thor平台联合开发内存。合作强调内存设计需提前数年与GPU协同设计、制造规划及资本投入，不能再于GPU完成后才设计。双方还将把AI引入芯片制造，利用NVIDIA CUDA-X、PhysicsNeMo、Omniverse、OpenUSD、cuOpt加速芯片设计和半导体仿真，并计划建立工厂数字孪生，在软件中模拟机器人、晶圆和物流后再投入产线。该合作进一步巩固了SK海力士在AI硬件供应链中的关键地位。

swyx@swyx · 5天前43

releasing tmr - the biggest code eval launch of the year glad to have played a small part in defining the agenda for this very critical next phase in koding

译swyx 宣布明天将发布"今年最大的代码评测基准"。他在引用推文中提到，已有第二家初创公司基于他的播客内容开展销售和评测，并感慨播客直接或间接影响了许多人的职业路线。

gabriel@gabriel1 · 5天前55

i have resigned from openai i left sora early this year to start a team at openai to build something great. but i've always been a founder, and there is one last product i need to build before AGI already miss all my friends and Colleague(s), i believe in you! more soon

译我已从OpenAI辞职。 今年早些时候我离开了Sora，在OpenAI成立了一个团队来打造伟大的东西。但我一直是个创始人，在AGI到来之前，我还有一个最后的产品需要打造。 已经开始想念我所有的朋友和同事们，我相信你们！后续更多。

Rohan Paul@rohanpaul_ai · 5天前42

Robot's locomotion and recovery under unexpected force in real time. The recovery phase was something 😀

译机器人在实时突发外力下的移动与恢复。 恢复阶段有点意思 😀

宝玉@dotey · 5天前81

ChatGPT 要变 AgentGPT 了 当然 ChatGPT 应该不会改名字，但 ChatGPT 应该不再是一个单纯的 Chat 工具了。 OpenAI 内部一位高管对《金融时报》说："Chat is dead."（聊天已死。） OpenAI 正在准备 ChatGPT 自 2022 年上线以来最大规模的改版。改版的方向很明确：不再只做聊天机器人，要做超级应用。编程工具 Codex、能自主执行多步骤任务的 AI agent、图像生成、第三方合作伙伴的应用（目前已知包括 Canva 和 Booking），全部塞进一个统一界面里。 首席产品官 Tibo（Thibault Sottiaux） 对《金融时报》说：目标是打造一个跨越所有平台的个人 AI 助手，覆盖工作和生活的方方面面。更远期的计划是，连用户输入 prompt 这一步都省掉，让模型直接理解用户意图。 【注：Tibo 目前统管 ChatGPT、Codex 和平台产品团队，OpenAI 最近刚把这几个原本独立的团队合并到了他手下。】 改版预计未来几周内开始上线，先从网页端和手机端推出。新界面会更积极地引导用户去用编程、图像生成和第三方应用功能，而不是停留在对话框里聊天。 这次改版的驱动力不只是产品理念，还有来自商业上的压力。 ChatGPT 现在有 9 亿周活跃用户，超过 5000 万付费个人用户，月收入 20 亿美元。但大部分用户用的是免费版，公司至今没有盈利。企业客户目前贡献了约 40% 的收入（来自大约 200 万家企业客户），OpenAI 希望年底前把这个比例提到 50%。Codex 是增长最快的产品线，桌面版上线后周活已经超过 500 万。 把用户从低利润的免费聊天引导到高利润的企业工具和编程产品上，是 IPO 前必须讲出来的故事。 说到 IPO，OpenAI 今年 3 月刚完成了 1220 亿美元的融资轮（Amazon 出了 500 亿，Nvidia 和软银各 300 亿），估值 8520 亿美元。高盛和摩根士丹利正在为 IPO 做准备，目标估值可能超过 1 万亿美元。 但竞争对手 Anthropic 也没闲着。6 月 1 日，Anthropic 向 SEC 提交了保密 S-1 文件，最新一轮融资估值已经达到 9650 亿美元，5 月年化收入 470 亿美元。也就是说，按估值算，Anthropic 已经超过了 OpenAI。在企业市场，Claude Code 的增长也在直接蚕食 Codex 的地盘。Google 那边，Gemini 上个月在 I/O 大会上宣布月活达到 9 亿，逼近 ChatGPT。 OpenAI 现在面对的局面是：消费端被 Google 追，企业端被 Anthropic 抢，自己又还没盈利，IPO 窗口就在眼前。把 ChatGPT 从聊天工具变成平台，既是产品升级，也是给投资人讲的新故事。 对普通用户来说，最直接的变化是：打开 ChatGPT 的界面会跟现在不一样，不再是一个简单的对话框，而是一个带有明确功能入口的应用。对开发者来说，Codex 和 agent 能力的整合意味着更多工作可以在 ChatGPT 内完成，不用跳转到其他工具。 至于 OpenAI 能不能靠这次改版让投资人相信它不只是一家聊天机器人公司，这可能是 Sam Altman 今年最重要的一次产品赌注。

译OpenAI 正筹备 ChatGPT 自 2022 年上线以来最大规模改版，从聊天机器人转向超级应用/Agent 平台，整合编程工具 Codex、图像生成及第三方应用（Canva、Booking）。高管称“聊天已死”，目标成为跨平台个人 AI 助手，未来甚至省去用户输入提示词。改版预计未来几周在网页和手机端上线。商业压力：ChatGPT 有 9 亿周活用户、5000 万付费用户、月收入 20 亿美元但未盈利；企业客户贡献约 40% 收入，目标年底达 50%；Codex 桌面版周活超 500 万。竞争对手 Anthropic 估值已达 9650 亿美元。OpenAI 刚完成 1220 亿美元融资，估值 8520 亿美元，IPO 目标估值可能超 1 万亿美元。

Rohan Paul@rohanpaul_ai · 5天前76

FT: The proposal suddenly of a sovereign-wealth-style fund got more attention inside the White House after Sam Altman visited Capitol Hill this week. The likely mechanism is that AI firms donate a small slice of equity into a public wealth fund, and that fund passes gains to Americans through accounts or dividends rather than the government directly running the companies. This would differ from the $9B stake the Trump administration took in Intel last year, because people would hold shares individually instead of the US govt owning equity directly. --- ft .com/content/8559a3f9-86de-4a1c-8a75-6623e83e6a00

译据FT报道，特朗普政府正与OpenAI探讨通过公共财富基金机制让政府入股AI初创公司。方案是AI企业捐赠小部分股权至该基金，基金通过账户或分红将收益返还美国公民，而非政府直接运营公司。这不同于特朗普去年对Intel的90亿美元直接持股。OpenAI此前已提出公共财富基金概念，持有长期资产，让公民分享AI增长红利。政治背景方面，选民担忧失业、数据中心成本与企业控制，而AI公司需要华盛顿在基建、采购和监管上的支持。

Satya Nadella@satyanadella · 5天前64

Great to see NHS England scaling Microsoft 365 Copilot to more than 500,000 staff. In early trials, staff saved an average of 43 minutes per day, helping put more time back into what matters most, caring for patients. https://ukstories.microsoft.com/features/nhs-england-accelerates-ai-adoption-with-microsoft-365-copilot-to-improve-service-delivery-reduce-costs-and-create-more-time-for-care/

译很高兴看到 NHS England 将 Microsoft 365 Copilot 推广给超过 50 万名员工。 早期试验中，员工平均每天节省 43 分钟，帮助将更多时间投入到最重要的事情——患者护理上。https://ukstories.microsoft.com/features/nhs-england-accelerates-ai-adoption-with-microsoft-365-copilot-to-improve-service-delivery-reduce-costs-and-create-more-time-for-care/

🚨 AI News | TestingCatalog@testingcatalog · 5天前63

OPENAI 🔥: ChatGPT app crossed 600 million monthly active users for the first time, according to Similarweb. Growing 👀

译OPENAI 🔥：据Similarweb数据，ChatGPT应用月活跃用户首次突破6亿。增长中👀

AYi@AYi_AInotes · 5天前71

我觉得Anthropic 在 OpenAI 挖人这件事，一直被低估了。 这次走的是 Clive Chan， OpenAI 自研芯片团队2号人物， 那个在 Broadcom 合作项目里主导芯片设计细节的， 2.4 年亲历了整个项目从零到快要部署的全过程的工程师。 人才战争的套路其实也不难看懂， 倒不是说谁钱给得多， 更主要是谁把下一场战争的地图， 比对方更早看清楚。 OpenAI 的战场在 IPO、规模化、外部合作落地等方面， Anthropic 的战场在基础科学、长期押注、把最挑剔的硬件人才堆到同一个屋檐下。 这次挖 Clive 应该不是一次孤立的招聘动作， 回顾Anthropic 在过去一年从 OpenAI 拿走的人里， 有做基础研究的，有做安全的，有现在专门盯着 AI 硬件的， 拼的都是同一条逻辑： 你在对方团队里最核心的人，感受到的是窗口期正在关闭， 还是我这边的山，刚刚开始好爬。 Clive 的原话是"无法抵抗爬一座新山的冲动"， 这不是啥礼貌的离职套话， 更像是一个 2.4 年都在同一个项目里磨的人，在告诉大家他心里真正想要表达的东西是什么。 一直以来人才战争里最难防的都不是给更高的钱， 而是如果某一天你的核心员工感受到你给的山已经爬到了顶，而旁边那座还没有人去过。

译Anthropic 挖走 OpenAI 自研芯片团队第二位硬件员工 Clive Chan，他主导与 Broadcom 合作的芯片设计，2.4 年亲历项目从零到即将部署。Clive 称无法抵抗“从山脚爬一座新山”的冲动，本周加入 Anthropic。主推文分析挖角逻辑：并非靠薪资，而是让核心员工感到 OpenAI 窗口期关闭，Anthropic 在基础科学、长期押注和硬件人才密度上提供更大攀登空间。过去一年已陆续挖走基础研究、安全及 AI 硬件等核心人才。

Emad@EMostaque · 5天前27

Only one solution

译据《纽约邮报》报道，一台机器人足球运动员踢球力量大到将墙壁击穿，可能会使人类变得过时。 Emad Mostaque 评论道：“唯一的解决方案。”

Rohan Paul@rohanpaul_ai · 5天前59

An Anthropic model that has not been released yet surfaced on Chinese API proxies before its launch. per the leak, Oceanus is priced at an insane $16/M input and $80/M output tokens—nearly 3x more expensive than Claude Opus. The model is codenamed Oceanus and belongs to the upcoming Mythos release. It was leaked, repackaged, and sold before Anthropic completed safety testing. --- reddit .com/r/Anthropic/comments/1txfdbw/forget_claude_mythos_the_leaked_oceanus_code/

译一个尚未发布的 Anthropic 模型在发布前就出现在中文 API 代理上。 据泄露信息，Oceanus 定价为惊人的 $16/M 输入 token 和 $80/M 输出 token——几乎是 Claude Opus 的 3 倍。 该模型代号为 Oceanus，属于即将发布的 Mythos 版本。 它在 Anthropic 完成安全测试之前就被泄露、重新打包并出售。

Chubby♨️@kimmonismus · 5天前50

Anthropic is reportedly caught in a bizarre contradiction: the Pentagon has blacklisted it as a supply-chain risk, while the NSA is allegedly using Claude Mythos for offensive cyber operations. Although Anthropic was classified as a "supply-chain risk" after the dispute with DoW, it is too good and too important not to be used.

译Anthropic 据称陷入一个奇怪的矛盾：五角大楼将其列为供应链风险，而 NSA 据称正在使用 Claude Mythos 进行进攻性网络操作。 尽管在与 DoW 的争端后，Anthropic 被归类为“供应链风险”，但它太出色、太重要了，以至于不得不被使用。

凡人小北@frxiaobei · 5天前67

AI 时代的又一个受益者， Supabase 宣布完成 5 亿美元的 F 轮融资。 公司的估值已达 100 亿美元。而在一年前它只值 20 亿。

译AI 时代的又一受益者，开源后端即服务（BaaS）平台 Supabase 宣布完成 5 亿美元 F 轮融资，公司估值达到 100 亿美元，而一年前估值仅为 20 亿美元。

凡人小北@frxiaobei · 5天前61

腾讯也回过神了。 内部宣布调整 AI token 额度。全员统一额度改为按工作任务动态调配。 通知明确表示，总投入只增不减，对能用 AI 带来显著提效和价值产出的同学，保障 token 额度，不搞 token 消耗量排名，不贩卖焦虑。

译腾讯内部宣布调整AI token额度，全员统一改为按工作任务动态调配。通知明确总投入只增不减，对能用AI带来显著提效和价值产出的员工保障token额度，不搞token消耗量排名，不贩卖焦虑。

ginobefun@hongming731 · 6天前23

#BestBlogs 早报 06-07 本期聚焦智能体时代的「工程底层」：Agent 系统的可靠性到底靠什么支撑？

译#BestBlogs 早报 06-07 本期聚焦智能体时代的「工程底层」：Agent 系统的可靠性到底靠什么支撑？

Rohan Paul@rohanpaul_ai · 6天前71

Ron Baron on SpaceX business case. "You can't own enough of the greatest idea you've ever had. That's the way I feel about SpaceX," ~ Ron Baron is billionaire investor and the founder of Baron Capital, with over $49B + in assets under management.

译亿万富翁投资者Ron Baron表示，SpaceX是“你永远无法拥有足够多的最伟大想法”。据Reuters，SpaceX 750亿美元IPO已获约1500亿美元需求（2倍超额认购）。路演中，SpaceX称其AI业务可追逐23万亿美元市场，通过降低太空访问成本，目标连接超过30亿未联网人口。SpaceX还认为，美国发电量与计算机产能增长已落后中国，可通过在太空部署数据中心等基础设施来弥补。

MiniMax (official)@MiniMax_AI · 6天前33

We’re heading to the AWS Builder Loft in SF on June 9 for an evening on open-weight foundation models on Amazon Bedrock. We’ll be showcasing MiniMax M3, including MiniMax Sparse Attention, the architecture that helps enable M3’s 1M-token context window, strong coding/agentic performance, and native multimodality in one open-weight model. Joined us are @ArtificialAnlys, @coderabbitai , @nvidia , and Daily. 📍 AWS Builder Loft, SF 📅 June 9 · 5:00–8:30 PM 🪪 Gov-issued ID required Come say hi at the MiniMax table. Register below 👇. #OpenWeights #AmazonBedrock #MiniMax

译MiniMax 将于6月9日在旧金山 AWS Builder Loft 举办开源权重模型之夜，重点展示其开源权重模型 MiniMax M3，及其关键架构 MiniMax Sparse Attention。该架构支撑了 M3 的 1M-token 上下文窗口、强大的编码与智能体性能以及原生多模态能力。合作伙伴包括 @ArtificialAnlys、@coderabbitai、NVIDIA 和 Daily。活动时间为 5:00–8:30 PM，入场需出示政府颁发的身份证件。

Rohan Paul@rohanpaul_ai · 6天前70

Rreuters: SpaceX’s $75B IPO is already seeing about $150B in demand (2x oversubscribed) SpaceX has said its AI business could chase a $23 trillion market, that it can use space to build AI compute capacity beyond the limits of Earth-based companies. "By dramatically reducing the cost of access to space, we have been able to expand our mission to address some of the Earth's most pressing challenges, including bridging the digital divide ​by aiming to connect over three ‌billion unconnected people to the internet and humanity's collective knowledge," SpaceX said in the roadshow ​presentation. SpaceX said U.S. electricity generation and computer-capacity growth have fallen behind China, partly due to barriers around major U.S. projects. SpaceX thinks that shortfall can ‌be remedied by putting data centers and ​other infrastructure in space using ​SpaceX launches. --- reuters. com/business/finance/spacex-ipo-running-two-times-oversubscribed-sources-say-2026-06-05

译据路透报道，SpaceX 750亿美元IPO已获约1500亿美元需求（2倍超额认购）。SpaceX在路演中表示，其AI业务可瞄准23万亿美元市场，通过太空建设AI算力突破地球限制。公司称美国电力与算力增长落后于中国，计划用SpaceX发射将数据中心等送入太空弥补差距。此前SpaceX披露与Google的云服务协议，Google每月支付9.2亿美元（约110亿美元/年）用于xAI数据中心算力。

Rohan Paul@rohanpaul_ai · 6天前69

FT: Trump administration, OpenAI discussing possible government stake in the AI startup. OpenAI has already floated a Public Wealth Fund, which would hold long-term assets and potentially pass AI-related gains back to citizens. It will be a government-backed vehicle outside the company. OpenAI said the fund could “invest in diversified, long-term assets” and would enable citizens to participate in the “upside” of AI growth, possibly by receiving the fund’s returns directly Theres a huge political angle: voters fear job loss, data-center costs, and corporate control, while AI companies need Washington’s support for infrastructure, procurement, and regulation. For some context, The Trump administration has already taken stakes in Intel, IBM and other quantum and critical mineral companies during the president’s second term. --- ft .com/content/b1ab6106-77e6-4218-9eb4-e44bd56ca400?syn-25a6b1a6=1

译据英国《金融时报》报道，特朗普政府正与OpenAI就政府入股该AI初创公司进行讨论。OpenAI提出设立一个公共财富基金，持有长期资产并可能将AI增长收益返还给公民。此举具有政治考量：选民担忧失业、数据中心成本和企业控制，而AI公司需要华盛顿在基础设施和监管等方面的支持。特朗普在其第二任期已入股Intel、IBM等量子及关键矿产公司。引用推文指出，美国公众可能成为公司的合作伙伴。

Rohan Paul@rohanpaul_ai · 6天前56

Trump Administration is in talks for U.S. Got stake in top AI labs. "there is so much money & it's so big that there are concepts where pieces could be given to the American public, where the American public becomes a partner with the companies."

译特朗普政府正在谈判让美国获得顶级AI实验室的股权。“这笔钱如此之多、规模如此之大，以至于有一些概念可以让部分股权分给美国公众，让美国公众成为这些公司的合作伙伴。”

Elon Musk@elonmusk · 6天前46

Join our AI recommendations & advertising engineering team!

译若你曾从事广告相关工作（任何环节），希望获得10倍影响力，请通过链接申请。当前团队仅18名工程师。加入xAI的AI推荐与广告工程团队！

Rohan Paul@rohanpaul_ai · 7天前52

Several Unitree G1 humanoid robots mirrored a lead dancer's choreography in real time via motion capture at a Shanghai event. The demonstration was part of a record 100-person simultaneous motion tracking challenge.

译几台宇树G1人形机器人在上海的一场活动中，通过动捕技术实时镜像了一位领舞者的编舞。这场演示是创纪录的百人同时运动追踪挑战的一部分。

小互@xiaohu · 7天前69

根据 SpaceX 公布的一份文件 Google与其签订了一份协议，从 2026 年10月至 2029 年6月止，每月支付9.2 亿美元购买算力，具体包括大约 110,000 个 NVIDIA GPU、CPU、内存以及其他组件。 此前 Anthropic也向SpaceX 每月支付 12.5 亿美元购买算力... 算下来每年有 260 亿美元的收入，躺赚啊... 日了狗了，赚钱还是得看老马...

译SpaceX公布文件显示，Google与其签订协议，从2026年10月至2029年6月，每月支付9.2亿美元购买算力，具体包括约110,000个NVIDIA GPU、CPU、内存及其他组件。此前Anthropic也已向SpaceX每月支付12.5亿美元购买算力。两项协议合计每年带来约260亿美元收入。

Deedy@deedydas · 7天前51

Every single AI startup with $10B+ valuation and $100M+ revenue run rate: Crusoe - $10B Mercor - $10B ElevenLabs - $11B Baseten - $11B* Harvey - $11B Lovable - $12B* OpenEvidence - $12B Mistral - $14B Nscale - $14.6B Fireworks - $15B* Sierra - $15.8B Moonshot - $20B Perplexity - $22.6B Cognition - $26B Scale - $29B** DeepSeek - $52-59B* Cursor - $60B* Waymo - $126B** xAI - $250B** OpenAI - $852B Anthropic - $965B 21 total companies. *rumored / unannounced **not fully independent

译估值超100亿美元、年收入超1亿美元的每一家AI初创公司： Crusoe - $10B Mercor - $10B ElevenLabs - $11B Baseten - $11B* Harvey - $11B Lovable - $12B* OpenEvidence - $12B Mistral - $14B Nscale - $14.6B Fireworks - $15B* Sierra - $15.8B Moonshot - $20B Perplexity - $22.6B Cognition - $26B Scale - $29B** DeepSeek - $52-59B* Cursor - $60B* Waymo - $126B** xAI - $250B** OpenAI - $852B Anthropic - $965B 共21家公司。 *传闻/未公布 **并非完全独立

Rohan Paul@rohanpaul_ai · 7天前48

Today’s edition of my newsletter just went out. 🔗 https://www.rohan-paul.com/p/anthropic-just-disclosed-that-claude 🗞️ Anthropic says 80% of its new production code is now authored by Claude 🗞️ New Google paper shows general LLMs can solve formal math by planning proofs and checking each step. Raised general LLM performance from under 10% to 70% 🗞️ Google’s new open source Gemma 4 12B can analyze audio and video while running fully locally on a consumer 16GB GPU 🗞️ Alibaba’s Qwen3.7-Plus supports text, video, and image inputs at a low price of $0.4/$1.6 per 1M tokens, though it remains proprietary. 🗞️ Anthropic’s new chemistry report has a genuinely wild result.

译Anthropic 称其 80% 的新生产代码由 Claude 编写。Google 新论文显示，通用 LLM 通过规划证明与逐步验证，将形式数学求解性能从低于 10% 提升至 70%。Google 开源 Gemma 4 12B，可在消费级 16GB GPU 上本地运行，支持音频和视频分析。通义千问发布 Qwen3.7-Plus，支持文本、视频、图像输入，价格 $0.4/$1.6 每百万 token，闭源。Anthropic 新化学报告有惊人结果。

jason@jxnlco · 7天前36

i joined openai 109 days ago no days off...

译我109天前加入了OpenAI 没有休息日……

Microsoft Research@MSFTResearch · 7天前49

A BTS look a first timer's experience at the MSR lab at Microsoft Build 2026—featuring the demos, the builders, and the conversations that made it worth the trip.

译幕后花絮——一位新手在微软Build 2026的MSR实验室的首次体验，展示了那些让此行值得的演示、构建者与对话。

Rohan Paul@rohanpaul_ai · 7天前78

Anthropic previously committed to paying SpaceX $1.25B per month for GPU compute. With the new Google cloud deal that was disclosed today, adding $920mil monthly, the two AI labs (Google + Anthropic) together are now collectively paying SpaceX $2.17B per month, a huge $26 billion annualized revenue run rate. To note, Alphabet also has made a huge gain from backing SpaceX. Google invested about $900M in SpaceX in Jan-2015, for roughly 7%, when SpaceX’s valuation was around $12 B. And SpaceX now targets a $1.75T IPO valuation. A reported 6.11% Google stake at 12-25 would be worth about $107B at $1.75T, while a diluted 5% stake would be worth about $87.5B . Against a roughly $900M entry cost, that implies around 97x to 119x on paper, before taxes, lockups, dilution, or any discount investors apply after trading starts. The business shift is also important: Alphabet first backed SpaceX in 2015, partly for satellite internet, but the upside now includes Starlink, launch dominance, and newly filed AI compute contracts.

译Anthropic此前承诺每月向SpaceX支付12.5亿美元GPU算力。Google新披露的云服务协议每月再付9.2亿美元（年化约110亿美元），两家AI实验室合计月付21.7亿美元，年化营收达260亿美元。Alphabet于2015年以约9亿美元入股SpaceX约7%，当前SpaceX目标IPO估值1.75万亿美元，其6.11%股份对应约1070亿美元，投资回报约97-119倍。业务转向显示AI算力正成为战略商品。

Chubby♨️@kimmonismus · 7天前56

Reports claim Claude’s API may have returned another user’s inference output during today’s outage. Anthropic’s status page confirms elevated errors affecting Claude API, Claude Code, Claude. ai and Claude Cowork but Anthropic has not confirmed a customer data leak yet. That would be a cross-tenant isolation failure and would be a worst-case scenario.

译有报告称，Claude 的 API 在今天的中断期间可能返回了其他用户的推理输出。 Anthropic 的状态页面确认，Claude API、Claude Code、Claude.ai 及 Claude Cowork 均出现高度错误，但 Anthropic 尚未确认出现客户数据泄露。 这将是一次跨租户隔离故障，属于最坏情况。