Yuchen Jin@Yuchenj_UW · 5月23日68

Wow. A massive 75% discount from DeepSeek. Either they’ve done some serious inference optimizations, or Huawei chips are just that much cheaper? More open-source AI models, better token economy.

译哇。DeepSeek给出了高达75%的折扣。 要么他们做了重大的推理优化，要么华为芯片就是这么便宜？ 更多开源AI模型，更好的token经济。

Claude@claudeai · 5月23日42

Kay Zhu is the co-founder and CTO of @genspark_ai, the all-in-one AI workspace built on Claude. In a market moving this fast, where anyone can build, he thinks the team is what makes the difference:

译Kay Zhu是@genspark_ai的联合创始人兼首席技术官，这是一个基于Claude构建的一体化AI工作空间。 他认为，在一个发展如此迅速、任何人都能参与构建的市场中，团队才是决定成败的关键。

Chubby♨️@kimmonismus · 5月22日54

Ben is building the future of fully autonomous businesses. And he just raised $30M. Even crazier: Solo founder Already at $10M run rate 7,600+ businesses already using the platform The company is growing insanely fast. Will cover the full Polsia story in the newsletter soon. Congrats @Bencera

译AI平台Polsia在创始人Ben主导下完成3000万美元融资，估值达2.5亿美元。该公司采用“创始人+AI”模式，无需雇员即可自主运营商业实体，目前年化营收已近1000万美元，服务超7600家企业。Polsia甚至自主完成了本次融资流程，创始人仅需签署文件。这种完全由AI驱动的商业运营模式展现了快速增长的态势。

宝玉@dotey · 5月22日40

帮转 DeepSeek Harness 团队招人

译DeepSeek旗下Harness团队正面向社会招聘，开放了研发工程师、产品经理及Harness方向研究员的职位。该团队招募全职与实习生，工作地点位于北京。感兴趣的候选人可通过官方招聘链接进行投递申请。

Orange AI@oran_ge · 5月22日65

DeepSeek 那个价格屠夫又回来了... 限时折扣改永久降价 比同等水平的其他模型便宜3倍 世界不能没有鲸鱼！

Rohan Paul@rohanpaul_ai · 5月22日46

Unitree Robotics' G1 humanoid robot playing table tennis against a human during a public exhibition demo at the Global Unicorn Innovation Exhibition in Hangzhou, China.

译宇树机器人公司的G1人形机器人在中国杭州全球独角兽创新展览的公开演示中与人类进行乒乓球对打。

Rohan Paul@rohanpaul_ai · 5月22日32

Edge AI runs on each insect backpack, enabling low-latency coordination, secure data exchange, formation control as a group and task execution. Swarm Biotactics scales by breeding insects and has raised ~€13M.

译边缘AI运行在每只昆虫的背包上，实现低延迟协调、安全数据交换、群体编队控制和任务执行。 Swarm Biotactics通过培育昆虫实现规模化，已融资约1300万欧元。

Rohan Paul@rohanpaul_ai · 5月22日60

Google was not far from launching its own ChatGPT-like product back in 2022. Sundar Pichai talks about how Google had already built a almost equvalent product around LaMDA, but its raw behavior was too toxic for Google standards, and unreliable to ship broadly. Pichai: "I think as a company, you had this search quality bias, and so, you know, we had a higher bar, for what we thought was an acceptable product quality to go out." The LaMDA model of Google was not simply a case of brilliant research sitting on a shelf, because Google had already started turning the model into a product and even exposed a constrained version through AI Test Kitchen. Pichai is also saying hindsight makes the miss look cleaner than it felt at the time, because even OpenAI’s move was not instantly obvious as the winning form factor. --- From @collision podast with @sundarpichai (full video link in comment)

译谷歌CEO桑达尔·皮查伊透露，公司在2022年已基于LaMDA模型开发出接近ChatGPT的产品，但因输出内容存在毒性且不符合内部质量标准而未能广泛发布。皮查伊指出，公司存在“搜索质量偏见”，对产品的可靠性和安全性设有更高门槛。LaMDA模型并非仅停留在研究阶段，谷歌曾通过AI Test Kitchen进行有限测试。他补充道，从后视镜看，OpenAI的成功在当时并非显而易见，这使得谷歌的决策过程更为复杂。

PixVerse@PixVerse_ · 5月22日24

Welcome @OoChihiroOO! We welcome and continue to invite more talented, like-minded creators to join us. 💜 Hope you enjoy PixVerse.

译欢迎 @OoChihiroOO！ 我们欢迎并继续邀请更多有才华、志同道合的创作者加入我们。 💜 希望你喜欢 PixVerse。

PixVerse@PixVerse_ · 5月22日25

Awww...Ohhh🙀 Nice work, @going_bait ! It got me fr.

译哇哦...噢🙀 干得漂亮，@going_bait！这效果太真实了。

Alibaba Cloud@alibaba_cloud · 5月22日38

Powering regional tech ecosystems. Meet HappyHorse and Qwen at the Toronto Tech Week.

译多伦多科技周开幕活动上，Beever AI将重点展示三款产品：开源平台Beever Atlas，可将Slack、Teams等团队对话转化为可搜索的“活记忆”；企业级大模型训练平台Votii MAGIC将进行全球商用发布，支持本地化构建领域专属AI，并已用于训练首个完全预训练的粤语大模型；现场还将提供基于阿里云的HappyHorse AI视频生成互动体验。活动汇集了加拿大银行、政府、投资机构及学术界代表，旨在赋能区域技术生态。（173字）

Alibaba Cloud@alibaba_cloud · 5月22日54

Foundation Model Forum | Qwen Conference 2026 Decoding the next leap in core intelligence. Join the session at Sands Expo Singapore to explore pre-training breakthroughs, reasoning logic, and future model roadmaps. 🚀 Secure your seat now: https://click.qwencloud.com/m/20000000190/

译基础模型论坛 | Qwen Conference 2026 解码核心智能的下一次飞跃。加入新加坡金沙会展中心的会议，探索预训练突破、推理逻辑与未来模型路线图。 🚀 立即预约席位：https://click.qwencloud.com/m/20000000190/

Emad@EMostaque · 5月22日64

20x sales for something growing that fast js a great deal for @SpaceX It will be immediately accretive given @SpaceX is likely coming to the market at over 50x sales

译对于增长如此迅速的公司，20倍销售额的收购价对@SpaceX来说非常划算 考虑到@SpaceX上市时估值可能超过50倍销售额，这笔交易将立即增厚其收益

Kling AI@Kling_ai · 5月22日32

Kling AI wraps up a remarkable conference at Marché du Film, Cannes! Cannes was buzzing on May 18✨ Kling AI took the stage at this year’s Marché du Film, bringing together filmmakers and industry talents from across the globe. It was inspiring to see so many creatives embrace the next era of AI-driven storytelling — our vision for the future of filmmaking received great feedback and support. The journey is just beginning. We’re excited to keep pushing the boundaries of creative storytelling with AI!

译Kling AI在戛纳电影节的市场展（Marché du Film）上成功展示其AI驱动电影制作的未来愿景。活动中汇聚了全球影视创作者与行业人才，其关于AI创意叙事的理念获得了广泛反馈与支持。此次展示标志着Kling AI推动AI与创意产业结合的旅程刚刚开始，未来将继续探索拓展AI在叙事创作领域的边界。

Rohan Paul@rohanpaul_ai · 5月22日74

Bloomberg: Cursor has reached a $3B annual sales rate as SpaceX prepares a possible $60B acquisition that would tie AI coding software directly to one of the world’s largest engineering machines. Cursor reportedly crossed $2B in annualized revenue in February and $3B by late April, with more than 3,000 customers paying at least $100K a year. SpaceX’s interest is not just financial, because a company building rockets, satellites, factories, chips, robotics, and flight software can gain real leverage if AI tools shorten engineering cycles. The deal structure is unusual: SpaceX can buy Cursor for $60B, or pay a $10B fee tied to their work together, including $1.5B in cash and $8.5B as a deferred services fee. Composer 2.5 using SpaceX data-center compute is another huge levereage, because top coding models are now limited not only by algorithms, but by access to huge training and inference capacity. Cursor is also huge for the SpaceX IPO because it turns SpaceX’s AI story from “we are spending billions on compute” into “we may own one of the fastest-growing AI software products selling into real companies.” SpaceX’s filing is so crucial here because public investors are being asked to value the company around a massive future AI thesis. Cursor helps answer the investor question: who will actually use all this compute, and what software layer will turn those GPUs into revenue? SpaceX IPO buyers are not only buying SpaceX’s current business, they are buying the idea that SpaceX can become an AI infrastructure and software platform. Cursor also makes SpaceX shares more useful as acquisition currency, because a public listing gives SpaceX a market-priced stock it can use for a $60B deal instead of spending only cash. The $10B walk-away exposure signals seriousness, because Cursor would receive $1.5B as a termination fee plus $8.5B as a deferred services fee if the deal fails under the stated terms. The news is soo great for the SpaceX IPO in so many ways, as it gives investors a cleaner story: SpaceX is not only building AI capacity, it is trying to own the coding product that consumes that capacity and sits inside daily developer work in every enterprises. --- bloomberg .com/news/articles/2026-05-21/cursor-hits-3-billion-annual-sales-rate-ahead-of-spacex-deal

译据彭博社报道，AI编程软件Cursor年销售额已突破30亿美元，并正与SpaceX洽谈一项约600亿美元的收购交易。这笔交易的战略意义远超财务投资：对于正准备启动史上最大规模IPO之一的SpaceX而言，收购Cursor能将其对AI算力的巨额投资故事变得完整。它向投资者证明，SpaceX不仅能建设算力，更能将其转化为面向企业开发者的软件产品和收入流，这与其从航天承包商转型为“平台公司”的IPO叙事高度契合。交易结构特殊，包含高达10亿美元的潜在分手费，显示了双方的严肃态度。

Rohan Paul@rohanpaul_ai · 5月22日35

"Existing high-power chips in China are still a fraction of the performance of NVIDIA's leading chips. And it's actually really, really hard to build up a scaled and complex semiconductor supply chain" Paulina McPadden, Inv Manager at Baillie Gifford

译中国现有的高性能芯片性能仍只是英伟达领先芯片的一小部分。 而且，建立规模化且复杂的半导体供应链实际上非常、非常困难。 ——Paulina McPadden，Baillie Gifford 投资经理

Alibaba Cloud@alibaba_cloud · 5月22日15

Agent-Native Infra Forum | Qwen Conference 2026 Re-architecting the cloud stack for autonomous intelligence. Join the session at Sands Expo Singapore to unlock the hardware, memory, and scaling infrastructure built for agents. 👉 Secure your seat now: https://click.qwencloud.com/m/20000000190/

译Agent-Native Infra Forum | Qwen Conference 2026 为自主智能重构云技术栈。加入在新加坡金沙会展中心举办的会议，探索专为智能体打造的硬件、内存与扩展基础设施。 👉 立即预约席位：https://click.qwencloud.com/m/20000000190/

SiliconFlow@SiliconFlowAI · 5月22日63

@deepseek_ai V4 Flash just hit #1 on @OpenRouter — 3.02T tokens, up 109% this week. If you haven't tried it yet, now's a good time. More Info⬇️

译@deepseek_ai V4 Flash 刚刚登顶 @OpenRouter 榜首——3.02T tokens，本周增长109%。 如果你还没试过，现在是个好时机。 更多信息⬇️

OpenRouter@OpenRouter · 5月22日65

DeepSeek V4 Flash has topped the weekly leaderboard

译DeepSeek V4 Flash已登顶周排行榜

歸藏(guizang.ai)@op7418 · 5月22日49

时间到了 Claude 又炸了

宝玉@dotey · 5月22日57

微软开始大规模收回内部员工的 Claude Code 许可证，要求开发者转向自家的 GitHub Copilot CLI。 据 The Verge 报道，微软去年 12 月开始在内部推广 Claude Code，鼓励项目经理、设计师等非技术岗位尝试用 AI 写代码。半年下来，Claude Code 在微软内部非常受欢迎，但这个受欢迎反而成了问题：它让微软自家刚推出的 GitHub Copilot CLI 显得尴尬了。 Copilot CLI 是 GitHub Copilot 的命令行版本，跟 Claude Code 的定位高度重叠。微软负责 Windows、Microsoft 365、Outlook、Teams 和 Surface 的 Experiences + Devices 团队，已经要求工程师在 6 月底前完成迁移，全面切换到 Copilot CLI。 微软对内的说法是统一工具链，但知情人士透露，背后也有成本考量。毕竟每个 Claude Code 的 license 费用都是交给竞争对手 Anthropic 的。 据悉，被要求迁移的工程师对此并不情愿。

译微软已开始大规模收回内部员工对Anthropic公司Claude Code工具的访问权限，要求开发者转向自家的GitHub Copilot CLI。尽管Claude Code在微软内部广受欢迎，但其成功反而凸显了自家Copilot CLI的定位尴尬。微软以统一工具链为由推动此次迁移，但核心考量也包括将支付给竞争对手Anthropic的成本收回。此举已引发被要求迁移工程师的普遍抵触。

Orange AI@oran_ge · 5月22日70

当你创造的价值超过你的薪资，你的薪资就成了诅咒 builder 因为 AI 生产力变 10x，价值持续升高 seller 做的一直是在放大收入，也比较安全 而这里说的第三类人，就成了重灾区

译Cloudflare近期裁员约1100人，CEO Matthew Prince在专栏中将员工分为“建造者”、“销售者”和“度量者”三类。AI生产力提升直接利好工程师等“建造者”，销售因需人际互动也较安全。最大风险落在负责财务、审计、中层管理等可重复性工作的“度量者”身上，因其任务正是AI所擅长。公司随后招聘大量AI原生代实习生实施“腾笼换鸟”，但此举引发股价下跌。核心观点是，AI替代风险与工作的结构性和可重复性高度相关。

宝玉@dotey · 5月22日77

Cloudflare CEO Matthew Prince 在《华尔街日报》发了一篇专栏，标题是：《我是怎么选择用 AI 替换哪些员工的》 Cloudflare 刚裁掉约 1100 人，占全员五分之一，是这家公司 16 年来第一次大规模裁员。然后Cloudflare 今年招了 1111 名实习生，基本上和裁员的人数相当，等于是腾笼换鸟了。 而且夸张的是， Cloudflare 今年收到了将近 100 万份实习申请，录取率只有千分之一，就业环境可见一斑，也难怪毕业典礼上 CEO 们吹 AI 下面嘘声一片！ 至于裁员的理由，当然是归结为 AI 了。为了说清楚裁员的理由，Prince 搬出了管理学家彼得·德鲁克 1954 年的《管理的实践》，把公司里的人分成三类： 1. 建造者（builder） 2. 销售者（seller） 3. 以及“度量者”（measurer），负责其余一切，包括财务、审计、法务、合规、中层管理、运营、市场。 AI 不动前两类。工程师效率翻十倍，他说有多少招多少；销售也安全，因为掏钱的是人，人愿意跟懂自己需求的人打交道。 会被 AI 顶掉的是第三类“度量者”，因为他们做的统计业绩、出报表、跑审计，正是结构化、可重复、AI 最擅长的活。这次裁的，绝大多数就是这批人。 他举了几个具体例子：Cloudflare 的内部审计以前每个季度只能抽查几个业务风险领域，现在转向全业务持续审计；财务关账更快了，错误更少了；中层管理者被大幅裁减，因为 AI 让每个经理可以直接管更多人。 而且用来替代这些人的实习生，Matthew Prince 的话来说：是天生的 AI 原生代。他们无一例外，全都是“建造者”或者“销售者”。 财报和裁员一起公布后，Cloudflare 股价一度跌掉二十多个百分点；公司这一季其实还亏了 6200 万美元，光遣散和重组就要花 1.4 到 1.5 亿美元。一边说不是省钱，一边背着上亿重组开支，投资人显然没全买账。

译Cloudflare裁员约1100人，同时招聘近似数量的实习生。CEO在《华尔街日报》撰文，引用德鲁克理论，指出AI主要替代从事审计、财务、中层管理等结构化工作的“度量者”，而工程师与销售人员影响较小。公司已将审计转向AI驱动，财务流程得以优化。大量实习岗位被视为面向“AI原生代”。财报显示公司当季亏损，且因裁员产生高额重组费用，消息公布后股价一度大跌超20%。

Huawei Cloud@HuaweiCloud1 · 5月22日44

At #HiFS2026, Huawei introduced the #HuaweiCloudStack financial solution — delivering a new resilient, open, and intelligent architecture to help financial institutions shift from closed systems to cloud-based, distributed, and intelligent infrastructure. ✅Resilient cloud infrastructure for VM replacement ✅ Seamless cloud migration of core applications ✅ Intelligent data lake for marketing, risk control, and operations

译在 #HiFS2026 上，华为推出了 #HuaweiCloudStack 金融解决方案——提供一种新的弹性、开放、智能的架构，帮助金融机构从封闭系统转向基于云的分布式智能基础设施。 ✅ 弹性云基础设施，用于虚拟机替换 ✅ 核心应用无缝云迁移 ✅ 智能数据湖，用于营销、风控和运营

Huawei Cloud@HuaweiCloud1 · 5月22日50

The Huawei Intelligent Finance Summit 2026 (HiFS 2026) was held at Shanghai, gathering global financial industry leaders and experts for in-depth dialogue on AI-driven digital transformation in finance. As we move into the intelligent era, Huawei provides fully cloud-based compute, storage, and networking, amplified by system-level software-hardware synergy, to form a resilient financial foundation. On this foundation, we are building a financial-grade AI agent enablement platform that supports a broad range of models and use cases, powering the leap to intelligent finance. Learn more: https://tinyurl.com/wtkm3n5c #HuaweiCloud

译华为于上海举办智能金融峰会2026（HiFS 2026），与全球金融业领袖探讨AI驱动的数字化转型。华为基于全栈云化的计算、存储与网络能力，通过软硬件系统级协同，构建了坚实的金融基础设施。在此之上，华为正打造一个支持多种模型与场景的金融级AI Agent赋能平台，旨在推动金融业向智能化阶段跃迁。

Berryxia.AI@berryxia · 5月22日46

YoLo这个传统模型在AI这一波来了之后，迭代速度和优化支持越来愈好。 支持交通信号灯的智能辅助系统，通过识别红绿灯和等待时间来辅助整个交通系统。

ginobefun@hongming731 · 5月22日45

#BestBlogs 早报 05-22 今日主题：Agent 记忆原语、Qwen3.7-Max、自动化与人类专家

MiniMax (official)@MiniMax_AI · 5月22日17

MiniMax just wrapped up an eventful week in the US 🥳 Quick recap 🧵

译MiniMax刚刚结束了在美国充实的一周 🥳 快速回顾 🧵

Greg Brockman@gdb · 5月22日65

OpenAI for YC companies:

译OpenAI为YC公司提供支持： [引用 @ycombinator]：OpenAI将为春季和夏季批次的每家YC公司提供价值200万美元的代币额度。 我们将夏季批次的申请截止日期延长至5月25日，以便更多创始人能够参与。 http://ycombinator.com/apply

Chubby♨️@kimmonismus · 5月22日64

Karpathy will help launch a new team focused on using Claude itself to accelerate pretraining research. Its team is focused recursive self improvement.

译Karpathy将协助启动一个新团队，专注于利用Claude本身来加速预训练研究。 该团队专注于递归自我改进。

Deedy@deedydas · 5月22日64

When I first joined Menlo, Modal was the #1 early stage company I wanted to invest in. Back then, they were just a ~$100M company. Finally, 2 years later, it’s an honor to finally partner with Erik, Akshat and team. As an engineer myself, I’ve never found an easier ways to scalably run all AI infra workloads (training, inference, sandbox, etc) across some of the best AI companies in the world: Anthropic, Meta, Suno, Doordash, Applied Compute, Cognition. Fun fact: Modal is one of the only companies where both founders are IOI gold medalists (2003 Sweden and 2014 India) and one of the most efficient startups in terms of revenue / employee

译AI基础设施平台Modal宣布完成C轮融资，以46.5亿美元估值融资3.55亿美元，由General Catalyst和Redpoint领投。该平台为Anthropic、Meta、Suno等众多AI公司提供可扩展的AI工作负载（训练、推理、沙箱等）运行服务，以高效著称。Modal的两位创始人均为国际信息学奥林匹克金牌得主，公司展现出高人均营收的运营效率。

Josh Woodward@joshwoodward · 5月22日18

What a ride for @GoogleLabs at IO! Hope you're enjoying lots of the new stuff!

译@GoogleLabs在IO大会上真是大放异彩！希望你们喜欢这些新功能！

Greg Brockman@gdb · 5月22日34

Daybreak for helping cybersecurity at major banks:

译OpenAI的Daybreak项目为大型银行的网络安全防御带来曙光： [引用 @TheRealAdamG]：https://www.afr.com/companies/financial-services/major-banks-use-openai-s-daybreak-for-cybersecurity-defence-20260519-p5zyn9

Chubby♨️@kimmonismus · 5月22日63

OpenAI made $5.7B in Q1. Anthropic made ~$4.7B. But Anthropic's annualized revenue recently hit $45B. OpenAI's sits at $25B. The difference: annualized revenue extrapolates from the most recent month, and Anthropic's monthly revenue appears to have more than doubled between Q1 and now. That means Anthropic's growth rate flipped the entire ranking sometime in Q2 - while also projecting its first operating profit (~$600M). Meanwhile OpenAI is losing $1.22 for every dollar it earns, ChatGPT user growth has stalled below its 1B target, and it just raised $122B in new funding. One company is getting profitable. The other is raising capital faster than it's growing users. The AI race isn't being won by whoever ships models first. It's being won by whoever figured out unit economics.

译Anthropic在2026年Q1营收约47亿美元，虽低于OpenAI的57亿美元，但其年化收入近期已跃升至450亿美元，远超OpenAI的250亿美元。这得益于Anthropic在Q2的月度收入翻倍以上，并预计实现约6亿美元的首次运营利润。相比之下，OpenAI处于亏损状态，ChatGPT用户增长未达预期，同时仍在大规模融资。这表明AI竞争的关键已从模型发布速度转向谁能建立可持续的商业模式。

Rohan Paul@rohanpaul_ai · 5月22日54

Bloomberg: The Pentagon is testing rival AI models after moving to replace Anthropic’s Claude inside military workflows. 25 power users are comparing models that could handle the department’s most demanding AI tasks. --- bloomberg. com/news/articles/2026-05-21/pentagon-tests-rival-ai-models-in-race-to-replace-anthropic

译彭博社：五角大楼在军事工作流程中替换Anthropic的Claude后，正在测试竞品AI模型。 25名核心用户正在比较可能处理该部门最苛刻AI任务的模型。

Epoch AI@EpochAIResearch · 5月22日63

For frontier AI chips, memory is the largest and fastest-growing component cost. High-bandwidth memory (HBM) has grown from 52% to 63% of total AI chip component spending between Q1 2024 and Q4 2025.

译对于前沿AI芯片而言，内存是最大且增长最快的组件成本。 高带宽内存（HBM）在AI芯片组件总支出中的占比，已从2024年第一季度的52%增长至2025年第四季度的63%。

Rohan Paul@rohanpaul_ai · 5月22日65

The Information: Anthropic is currently in early-stage talks to lease and deploy Microsoft's custom AI chips for inference workloads. Microsoft is pitching Maia 200 as a cheaper way to run some AI inference, and claims maia 200 is more cost-effective than nvidia chips for certain inference jobs. Maia 200 is Microsoft’s second-generation AI accelerator, built on TSMC 3nm, with FP8/FP4 math, 216GB HBM3e, 7TB/s bandwidth, and 272MB SRAM, which makes it aimed at feeding large models fast rather than teaching them from scratch. Anthropic already committed $30B to Azure, Microsoft may invest up to $5B in Anthropic, and Claude is already tied into Microsoft’s Copilot stack, so the chip talks are also a customer-supplier feedback loop. IMO, Maia does not need to beat Nvidia everywhere to matter, because a cheaper chip for narrow, high-volume inference jobs can still shift billions of tokens away from GPUs. --- theinformation .com/articles/anthropic-talks-use-microsofts-ai-chips

译据The Information报道，微软正向AI公司Anthropic推销其第二代AI芯片Maia 200，强调该芯片在特定推理任务中比NVIDIA芯片更具成本效益。Maia 200专注于高速推理而非训练，双方已有深度合作基础：Anthropic已在Azure承诺300亿美元支出，且Claude已整合进微软Copilot。此次芯片合作旨在深化协同。分析认为，Maia 200无需全面超越NVIDIA，只要能在高量推理中提供更低成本选项，便可能将部分计算需求从GPU转移。

Rohan Paul@rohanpaul_ai · 5月22日68

WSJ: Anthropic is now projected to hit its first operating profit in Q2, with revenue jumping 130% to $10.9B after expecting profitability only around 2028. The shock is not only the profit, but the speed at which enterprise AI spending is converting into real revenue. Anthropic made $4.8B in Q1, then told investors it expects $559M in operating profit in the June quarter. The driver is agentic coding, where Claude does longer software tasks instead of only answering one prompt at a time. That changes the product from a chatbot into rented digital labor, so customers can justify much bigger bills. The cost story also changed, because Anthropic spent 71 cents of every revenue dollar on compute in Q1 but expects 56 cents in Q2. Compute is still the main tax on AI, but lower compute per dollar means scale is starting to work in Anthropic’s favor. Anthropic also benefits from using more Google and Amazon chips, a smaller free consumer base, and fewer giant data-center promises than OpenAI. --- wsj. com/tech/ai/mind-blowing-growth-is-about-to-propel-anthropic-into-its-first-profitable-quarter-7edbf2f4

译据《华尔街日报》报道，Anthropic预计将大幅提前盈利，计划于2026年第二季度实现首次运营利润，而非此前预计的2028年。该公司收入预计同比增长130%，达到109亿美元，二季度运营利润可达5.59亿美元。核心驱动力是“代理式编程”，使Claude能处理更长的软件任务，提供可租赁的数字劳动力，从而显著提升客户价值。同时，成本结构持续改善，计算成本占收入比预计从一季度的71%降至56%，规模效应开始显现。此外，更广泛地使用谷歌和亚马逊芯片、较小的免费用户群体以及更少的超大型数据中心投资承诺，也为其带来了竞争优势。

Krea@krea_ai · 5月22日26

this happening in 1h! join us to chat about LoRAs and new updates coming to Krea 2.

译1小时后开始！ 加入我们，一起聊聊LoRAs和Krea 2即将到来的新更新。 [引用 @krea_ai]：我们明天将举办首次社区空间活动！ 通过X Spaces加入我们，讨论Krea 2的更新、分享反馈或展示你的作品。 链接在下方 👇

宝玉@dotey · 5月22日20

之前 ClawHub 名称被占用的问题，官方已经解决了，非常感谢 @chenjingtalk @pat_erichsen @steipete 的支持🙏