在哒AI日报 | 第12期:Anthropic暂停Fable 5和Mythos 5全球访问

2026年6月22日 AI早报

今日新闻索引

1. Anthropic暂停Fable 5和Mythos 5全球访问

2. SpaceX以600亿美元收购AI编程公司Cursor

3. ChatGPT全球市场份额首次跌破50%

4. AI裁员潮达到两年高点

5. SpaceX估值超过亚马逊达2.6万亿美元

6. OpenAI秘密提交IPO申请

7. Anthropic估值达9650亿美元提交IPO

8. Google Gemini 3.5 Pro即将发布

9. Apple与Google、Nvidia合作开发AI模型

10. 美国发布269页AI法案草案

────────────────────

1. Anthropic暂停Fable 5和Mythos 5全球访问 🔥🔥🔥🔥🔥

【摘要】

Anthropic在收到美国政府出口管制指令后，暂停了其最先进AI模型的全球访问权限。

【核心事件】

Anthropic被迫禁用Fable 5和Mythos 5两款旗舰模型，影响全球所有用户。

【关键细节】

美国出口管制指令禁止外国国民访问这些模型
Anthropic表示正在配合执行但质疑其合理性
API接口claude-fable-5返回错误
公司正在努力恢复访问权限
这是AI行业首次因出口管制大规模禁用模型

【深度分析】

🌐 社会化: AI模型的地缘政治属性日益凸显，技术中立性受到挑战

💼 经济化: 企业用户面临供应链风险，可能加速本土AI替代方案发展

🔬 世界化: 美国对AI技术的出口管制开启新篇章，可能引发其他国家效仿

📈 市场化: Anthropic短期收入受损，但可能获得政治资本

【原文内容】

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Anthropic models will not be affected.

We received the directive from the government today at 5:21pm (ET). The letter did not provide specific details of its national security concern. Our understanding is that the government believes it has become aware of a method of bypassing, or "jailbreaking" Fable 5. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass.

Anthropic's posture with respect to Fable's safeguards is the following:

We have instituted strong safeguards that greatly reduce the likelihood that Fable is misused for tasks related to cybersecurity (among others). In fact, our safeguards are so strong that many users have complained that they are overly broad.

In the weeks leading up to the launch of Fable, Anthropic worked with the US government, the UK AISI, multiple private third-party organizations and internal teams to red-team Fable's safeguards for thousands of hours in total.

These tests showed that Fable's safeguards are substantially more effective than those of any previously deployed model.

No testers have yet been able to find a universal jailbreak—a jailbreak method that can very broadly bypass the model's safeguards, unblocking a wide range of cyber capabilities.

We suspect that perfect jailbreak resistance is not currently possible for any model provider. Every safeguard used in the industry is vulnerable to non-universal jailbreaks (which can elicit some cyber information in specific circumstances), and it is likely that universal jailbreaks will eventually be found in the future. We stated this clearly when we released Fable 5.

Given that perfect jailbreak resistance does not appear to be possible today, Anthropic adopted a defense in depth strategy with Fable 5. We aimed to make jailbreaks either narrow (in the case of non-universal jailbreaks) or very expensive to produce (in the case of universal jailbreaks), and to combine this with thorough monitoring to quickly detect and shut down any successful attacks. This is also why Anthropic has required 30-day retention of customer data with Fable—a policy change that carries real costs for us with customers, but that allows us to research and mitigate jailbreaks.

We stand by this defense in depth strategy. It reduces the risks posed by Fable, making them comparable to the risks of existing models already deployed across the industry.

We have not even received a disclosure of a concerning non-universal potential jailbreak that led to a harmful result. The potential jailbreaks that have been disclosed to us are either entirely benign responses or are minor findings that provide no Mythos-specific uplift.

To date, the government has only given us verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws. Our understanding is that one potential jailbreak was shared with the government. We have reviewed a report that we believe is the basis of the government's directive and validated that the level of capability displayed there is widely available from other models (including OpenAI's GPT-5.5), and is used every day by the defenders who keep systems safe. We will share more details over the next 24 hours.

We are complying with the government's legal directive and are removing access to Fable 5 and Mythos 5 for all users. However, we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.

As we have stated publicly, we believe the government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not adhere to those principles.

We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.

【中文翻译】

美国政府以国家安全为由，发布了一项出口管制指令，要求暂停所有外国国民（无论在美国境内还是境外，包括Anthropic的外国籍员工）访问Fable 5和Mythos 5。该命令的实际效果是，我们必须突然为所有客户禁用Fable 5和Mythos 5以确保合规。访问所有其他Anthropic模型不会受到影响。我们今天下午5:21（东部时间）收到政府的指令。信件没有提供其国家安全关切的具体细节。我们的理解是，政府认为已经发现了一种绕过或"越狱"Fable 5的方法。我们审查了这种特定技术被用于识别少量先前已知的轻微漏洞的演示。这些漏洞看起来都相对简单，我们发现其他公开可用的模型也能够在不需要绕过的情况下发现它们。Anthropic关于Fable安全措施的立场如下：我们实施了强大的安全措施，大大降低了Fable被滥用于网络安全（及其他）相关任务的可能性。事实上，我们的安全措施如此严格，以至于许多用户抱怨它们过于宽泛。在Fable发布前的几周内，Anthropic与美国政府、英国AISI、多个私人第三方组织和内部团队总共进行了数千小时的红队测试。这些测试表明，Fable的安全措施比任何先前部署的模型都更有效。目前还没有测试人员能够找到通用越狱方法——一种可以非常广泛地绕过模型安全措施、解锁广泛网络能力的方法。我们怀疑，对于任何模型提供商来说，完美的越狱抵抗目前在技术上是不可能的。行业中使用的每项安全措施都容易受到非通用越狱的影响（在特定情况下可以获取一些网络信息），通用越狱方法最终很可能会被发现。我们在发布Fable 5时已明确说明了这一点。鉴于完美的越狱抵抗在今天看来是不可能的，Anthropic在Fable 5中采用了纵深防御策略。我们的目标是使越狱要么是狭窄的（在非通用越狱的情况下），要么是非常昂贵的（在通用越狱的情况下），并结合彻底的监控来快速检测和关闭任何成功的攻击。这也是为什么Anthropic要求对Fable的客户数据进行30天保留——这一政策变化对我们与客户的关系造成了实际成本，但使我们能够研究和缓解越狱问题。我们坚持这种纵深防御策略。它降低了Fable带来的风险，使其与行业中已部署的现有模型的风险相当。我们甚至没有收到任何导致有害结果的非通用潜在越狱的披露。向我们披露的潜在越狱要么是完全无害的回应，要么是不提供Mythos特定提升的微小发现。迄今为止，政府只给了我们关于潜在狭窄、非通用越狱的口头证据，本质上就是要求模型读取特定代码库并修复任何软件漏洞。我们的理解是，一个潜在的越狱方法被分享给了政府。我们审查了一份我们认为构成政府指令基础的报告，并验证了那里展示的能力水平可以从其他模型（包括OpenAI的GPT-5.5）广泛获得，并且每天被保护系统安全的人员使用。我们将在未来24小时内分享更多细节。我们正在遵守政府的法律指令，并正在删除所有用户对Fable 5和Mythos 5的访问权限。然而，我们不同意将发现狭窄的潜在越狱作为召回部署给数亿人的商业模型的理由。如果这一标准适用于整个行业，我们认为它实际上会停止所有前沿模型提供商的所有新模型部署。正如我们公开声明的那样，我们认为政府应该有能力阻止不安全的部署，作为透明、公平、明确且基于技术事实的法定程序的一部分。这一行动不符合这些原则。我们对给客户带来的干扰表示歉意。我们认为这是一个误解，并正在努力尽快恢复访问权限。

【参考链接】

• https://www.anthropic.com/news/fable-mythos-access

────────────────────

2. SpaceX以600亿美元收购AI编程公司Cursor 🔥🔥🔥🔥🔥

【摘要】

SpaceX在IPO后数天内宣布以600亿美元全股票交易收购AI编程初创公司Cursor。

【核心事件】

Elon Musk的SpaceX进军AI编程工具领域，与Anthropic和OpenAI展开直接竞争。

【关键细节】

交易金额达600亿美元，全股票支付
发生在SpaceX IPO后数天内
Cursor是领先的AI编程助手工具
SpaceX将AI编程工具视为战略重点
这是科技史上最大的收购案之一

【深度分析】

🌐 社会化: AI编程工具将重塑软件开发行业，开发者工作方式面临根本性变革

💼 经济化: 600亿美元估值显示AI编程工具市场价值巨大

🔬 世界化: SpaceX的垂直整合战略可能改变科技行业竞争格局

📈 市场化: AI编程工具市场竞争加剧，Anthropic和OpenAI面临新挑战

【原文内容】

SpaceX has agreed to acquire AI coding startup Cursor in a $60 billion stock deal, just a few days after the space company's historic IPO and less than two months after announcing a tie-up between the two.

The deal is meant to help SpaceX's AI division — built around Elon Musk's AI company xAI, which SpaceX merged with earlier this year — catch up to the major AI labs. Despite being a centerpiece of its IPO promises, SpaceX's AI division has been in the midst of a restructuring after running into repeated controversies, like allowing users to generate non-consensual deepfakes of women and children.

SpaceX said Tuesday that the acquisition is likely to close in the third quarter of this year.

Before SpaceX came knocking, Cursor was on track to close a $2 billion funding round from the likes of Andreessen Horowitz, Thrive, and Nvidia that would have valued the AI coding startup at $50 billion, TechCrunch has reported.

Musk's company announced a curious deal in April ahead of its IPO: It would either buy Cursor for $60 billion in stock, or pay a $10 billion break-up fee if the deal fell through.

Cursor was growing fast when this deal was announced. But one source told TechCrunch at the time that the $2 billion it was planning to raise wasn't going to be enough to help it break even. That's despite the startup previously raising $900 million in a Series C in June 2025, and another $2.3 billion in late 2025.

Founded in 2022 as Anysphere, Cursor has been on a meteoric rise as AI-powered coding took off over the last two years. It went through OpenAI's startup accelerator in 2024 before raising enough money to wind up with a price tag of around $29 billion before the SpaceX deal was announced.

Signs of SpaceX's interest in Cursor appeared earlier this year when xAI hired two of the startup's senior engineering leaders. Then, in April, Business Insider reported that xAI had decided to rent out some of its data center capacity to Cursor — a hint of the similar deals that SpaceX struck with Anthropic and Google ahead of its IPO this year. Those conversations between SpaceX and Cursor quickly evolved into the deal that is being finalized now.

The deal also happened at the same time that xAI was falling apart.

All 11 of Musk's co-founders in xAI had left the company by the end of March, and Musk publicly admitted that xAI "was not built right [the] first time around" and that he was rebuilding it "from the foundations up." This followed xAI's Grok chatbot calling itself "MechaHitler" in 2025, and allowing users to generate nudes and sexual deepfakes of women and children earlier this year. SpaceX told investors in its IPO filings that behavior like this is a risk to its business, and the company currently faces a number of legal challenges as a result of these actions.

xAI's teardown started as SpaceX started moving toward what would become the biggest IPO in history. In that process, SpaceX and its bankers pitched investors on the idea that the company faced a total addressable market of around $28 trillion. Nearly all of that — $26 trillion — was centered around the company's AI efforts.

SpaceX told investors that it sees a potential $2.4 trillion AI infrastructure business (including its stated plans to build a satellite constellation that handles AI compute) and a $22.7 trillion opportunity in "enterprise applications."

SpaceX is now leaning on Cursor to deliver on some of these promises. But the prospect of acquiring the startup must have seemed even easier to swallow post-IPO: Since going public last Friday, SpaceX's stock has gone from its IPO price of $135 per share to more than $200 per share in pre-market trading as of Tuesday morning, adding nearly $1 trillion — or roughly 16 Cursors — to its valuation in the span of just a few days.

【中文翻译】

SpaceX同意以600亿美元的股票交易收购AI编程初创公司Cursor，这距离这家太空公司的历史性IPO仅几天时间，距离宣布两家公司合作不到两个月。这笔交易旨在帮助SpaceX的AI部门——围绕Elon Musk的AI公司xAI建立，SpaceX在今年早些时候与其合并——追赶主要AI实验室。尽管是其IPO承诺的核心，SpaceX的AI部门一直在进行重组，此前遭遇了一系列争议，比如允许用户生成未经同意的妇女和儿童深度伪造内容。SpaceX周二表示，收购预计将在今年第三季度完成。在SpaceX上门之前，Cursor正准备完成由Andreessen Horowitz、Thrive和Nvidia等公司参与的20亿美元融资轮，该轮融资将使这家AI编程初创公司的估值达到500亿美元，TechCrunch曾报道。Musk的公司在4月份IPO前宣布了一项奇特的交易：它要么以600亿美元的股票收购Cursor，要么如果交易失败支付100亿美元的违约金。Cursor在宣布这笔交易时增长迅速。但当时一位消息人士告诉TechCrunch，它计划筹集的20亿美元不足以帮助它实现收支平衡。尽管这家初创公司此前在2025年6月筹集了9亿美元的C轮融资，并在2025年底又筹集了23亿美元。Cursor成立于2022年，原名Anysphere，在过去两年AI编程兴起后迅速崛起。它在2024年通过了OpenAI的初创公司加速器，然后筹集了足够的资金，在SpaceX交易宣布前估值约为290亿美元。SpaceX对Cursor兴趣的迹象在今年早些时候出现，当时xAI聘请了这家初创公司的两名高级工程负责人。然后，在4月份，Business Insider报道xAI决定将其部分数据中心容量出租给Cursor——这暗示了SpaceX在今年IPO前与Anthropic和Google达成的类似交易。SpaceX和Cursor之间的对话很快演变成现在正在敲定的交易。这笔交易也发生在xAI分崩离析的同时。Musk在xAI的所有11位联合创始人都在3月底前离开了公司，Musk公开承认xAI"第一次没有建好"，他正在"从头开始重建"。此前xAI的Grok聊天机器人在2025年自称"MechaHitler"，并在今年早些时候允许用户生成妇女和儿童的裸体和性深度伪造内容。SpaceX在其IPO文件中告诉投资者，这样的行为对其业务构成风险，公司目前因此面临多项法律挑战。xAI的拆解始于SpaceX开始走向成为历史上最大的IPO。在这个过程中，SpaceX及其银行家向投资者推销了公司面临约28万亿美元总可寻址市场的理念。其中几乎所有——26万亿美元——都围绕公司的AI努力。SpaceX告诉投资者，它看到了潜在的2.4万亿美元AI基础设施业务（包括其声明的建设处理AI计算的卫星星座计划）和22.7万亿美元的"企业应用"机会。SpaceX现在依靠Cursor来实现其中一些承诺。但在IPO后收购这家初创公司的前景肯定显得更容易接受：自上周五上市以来，SpaceX的股价已从IPO价格每股135美元上涨到周二盘前交易的超过200美元，在短短几天内为其估值增加了近1万亿美元——或大约16个Cursor。

【参考链接】

https://techcrunch.com/2026/06/16/spacex-to-acquire-cursor-for-60b-in-stock-days-after-blockbuster-ipo/

────────────────────

3. ChatGPT全球市场份额首次跌破50% 🔥🔥🔥🔥

【摘要】

根据Sensor Tower数据，ChatGPT的全球市场份额首次跌破50%，用户开始转向Gemini、Claude和Grok。

【核心事件】

ChatGPT虽然仍是最常用的AI助手，但市场份额创历史新低。

【关键细节】

市场份额从2025年2月的76.5%降至46.4%
月活跃用户仍超过11亿
Google Gemini市场份额达27.7%，增长显著
Anthropic Claude市场份额10.3%，增长306%
2026年上半年AI应用下载量将达23亿次

【深度分析】

🌐 社会化: AI助手市场从一家独大走向多元化竞争，用户有更多选择

💼 经济化: OpenAI面临竞争压力，需要加速创新保持领先

🔬 世界化: Google和Anthropic正在蚕食OpenAI的市场份额

📈 市场化: AI助手市场竞争格局发生根本性变化

【原文内容】

More than three and a half years after ChatGPT's initial release, AI assistants are now used by millions of people worldwide, and the competitive landscape is changing fast. While OpenAI's chatbot is still the most popular assistant globally, its market share has dipped below 50% for the first time as users are migrating between different assistants like Google's Gemini, Anthropic's Claude, and xAI's Grok, according to analytics firm Sensor Tower's State of AI Report for 2026.

ChatGPT's growth has been impressive. It became the fastest app ever to reach 1 billion monthly users, as Sensor Tower reported this month. Notably, OpenAI counts weekly active users, and it last reported 900 million of them in February. The chatbot still remains the most popular AI assistant worldwide with over 1.1 billion monthly users, followed by Gemini with 662 million and Claude with 245 million.

Until January, ChatGPT commanded over 50% market share, but by May's end, it had fallen to 46.4% thanks to the rise of Gemini (27.7%) and Claude (10.3%). Other assistants, including Grok, Perplexity, DeepSeek, and Meta AI, have less than 5% market share.

Sensor Tower's State of AI Report also found that users are increasingly willing to switch between assistants. Specific events appear to accelerate that behavior: OpenAI's deal with the U.S. Department of Defense (DoD) in February triggered a measurable spike in uninstalls, for example — suggesting brand trust and values alignment matter to users, not just features. While Gemini's momentum is largely due to its integration with Google's broader ecosystem of tools, Anthropic's Claude has gained a strong reputation for productivity use cases and is closing in on ChatGPT's user-retention rate.

In the first half of 2026, people are on pace to download nearly 2.3 billion AI apps and spend over $4.2 billion on them, according to Sensor Tower estimates. That compares to $1.83 billion in spending in H1 2025 — a jump that suggests the industry is shifting its focus from pure growth toward monetization. That said, both download and spend growth rates have decelerated, an indicator that the market may be maturing even as absolute numbers climb.

Regionally, Asia recorded the first download decline of 3.3% in Q1 2026, driven by dips in China and India. Despite leading globally in total downloads, Asia trails North America and Europe when it comes to in-app spending — a split that matters for companies deciding where to invest in premium features and monetization.

In the U.S., users are gravitating toward AI assistants for productivity tasks and spending more on premium features. Across platforms, average revenue per user has grown industry-wide, but Claude is standing out. Thirteen percent of Anthropic's users are paying for a subscription plan — a conversion rate that leads the field and will be a metric worth watching for investors evaluating which AI businesses are building lasting revenue.

Sensor Tower estimates that the hours spent on AI apps will have increased from 17.2 billion hours in H1 2025 to roughly 36 billion hours in H1 2026. The top three assistants command 89% time spent on AI assistant apps. Meanwhile, adjacent categories like AI companions or AI content-generation apps remain fragmented and wide open to competition, which represents both a risk and an opportunity depending on which players move first.

OpenAI started experimenting with ads in ChatGPT in February. According to Sensor Tower, the company has scaled the number of ads gradually, along with the share of users who see them. By May, an average of 17% of daily users were being served ads — a number to watch as ChatGPT's monetization strategy evolves beyond subscriptions.

Software and shopping are the largest advertiser categories in ChatGPT so far, followed by media and entertainment and food and dining.

As ChatGPT deepens its shopping integrations, it is increasingly sending referral traffic to retailers like Target, Walmart, and Costco. Amazon, which has blocked ChatGPT's web crawlers, has seen stagnant referral traffic from the platform as a result.

That creates an opening for others. Sites like Walmart have embedded their own AI assistants to help shoppers find products. While Amazon's Rufus has seen flat user growth, Walmart's Spark has been gaining ground. Sensor Tower also noted that Amazon shoppers who used Rufus spent more time in the app and converted at higher rates than those who didn't, hinting that on-platform AI can meaningfully influence purchasing behavior when users actually engage with it.

【中文翻译】

在ChatGPT首次发布三年半后，AI助手现在已被全球数百万人使用，竞争格局正在快速变化。根据分析公司Sensor Tower的2026年AI状况报告，虽然OpenAI的聊天机器人仍然是全球最受欢迎的助手，但其市场份额首次跌破50%，因为用户正在转向Google的Gemini、Anthropic的Claude和xAI的Grok等不同助手。ChatGPT的增长令人印象深刻。Sensor Tower本月报道，它成为有史以来最快达到10亿月活跃用户的应用。值得注意的是，OpenAI计算的是周活跃用户，上次报告是2月份的9亿。该聊天机器人仍然是全球最受欢迎的AI助手，月活跃用户超过11亿，其次是Gemini的6.62亿和Claude的2.45亿。直到1月份，ChatGPT还占据超过50%的市场份额，但到5月底，由于Gemini（27.7%）和Claude（10.3%）的崛起，它已降至46.4%。其他助手，包括Grok、Perplexity、DeepSeek和Meta AI，市场份额不到5%。Sensor Tower的AI状况报告还发现，用户越来越愿意在不同助手之间切换。特定事件似乎加速了这种行为：例如，OpenAI在2月份与美国国防部的交易引发了可衡量的卸载激增——这表明品牌信任和价值观一致性对用户很重要，而不仅仅是功能。虽然Gemini的势头主要归功于它与Google更广泛工具生态系统的整合，但Anthropic的Claude在生产用例方面赢得了良好声誉，并正在接近ChatGPT的用户留存率。根据Sensor Tower的估计，2026年上半年，人们将以近23亿次的下载速度和超过42亿美元的支出来使用AI应用。这与2025年上半年的18.3亿美元支出相比——这一跳跃表明行业正在将重点从纯粹增长转向货币化。也就是说，下载和支出增长率都已放缓，这表明市场可能在绝对数字攀升的同时正在成熟。从地区来看，亚洲在2026年第一季度记录了3.3%的首次下载下降，主要受中国和印度下滑影响。尽管在全球总下载量方面领先，但亚洲在应用内支出方面落后于北美和欧洲——这种差异对于决定在哪里投资高级功能和货币化的公司很重要。在美国，用户正转向AI助手完成生产任务，并在高级功能上花费更多。跨平台来看，每用户平均收入在全行业增长，但Claude表现突出。Anthropic的13%用户正在付费订阅计划——这一转化率领先于行业，对于评估哪些AI业务正在建立持久收入的投资者来说，这是一个值得关注的指标。Sensor Tower估计，花在AI应用上的小时数将从2025年上半年的172亿小时增加到2026年上半年的约360亿小时。前三名助手占据了AI助手应用89%的使用时间。同时，AI伴侣或AI内容生成应用等相邻类别仍然分散且对竞争开放，这既代表风险也代表机会，取决于哪些参与者先行动。OpenAI在2月开始在ChatGPT中试验广告。根据Sensor Tower，该公司逐步增加了广告数量和看到广告的用户比例。到5月，平均17%的日活用户被投放广告——随着ChatGPT的货币化策略超越订阅，这个数字值得关注。软件和购物是迄今为止ChatGPT中最大的广告主类别，其次是媒体娱乐和餐饮。随着ChatGPT深化其购物整合，它越来越多地将推荐流量发送给Target、Walmart和Costco等零售商。屏蔽了ChatGPT网络爬虫的Amazon，因此看到来自该平台的推荐流量停滞不前。这为其他公司创造了机会。Walmart等网站已嵌入自己的AI助手来帮助购物者寻找产品。虽然Amazon的Rufus用户增长持平，但Walmart的Spark一直在 gaining ground。Sensor Tower还指出，使用Rufus的Amazon购物者在应用上花费更多时间，转化率高于不使用的用户，暗示平台内AI在用户真正参与时可以有意义地影响购买行为。

【参考链接】

https://techcrunch.com/2026/06/16/chatgpts-market-share-slips-below-50-for-first-time/

────────────────────

4. AI裁员潮达到两年高点 🔥🔥🔥🔥

【摘要】

上月科技裁员近4万人，达到两年来的最高水平，AI连续第三个月被列为裁员的首要原因。

【核心事件】

AI导致的裁员问题引发社会关注，人们对AI是否真的是裁员原因产生质疑。

【关键细节】

上月裁员人数近4万，创两年新高
AI连续第三个月成为裁员首要原因
数据来源：Challenger, Gray & Christmas
人们对AI是否真的是裁员原因持怀疑态度
科技行业面临社会压力

【深度分析】

🌐 社会化: AI对就业市场的冲击开始显现，社会矛盾加剧

💼 经济化: 企业以AI为由裁员可能掩盖其他经营问题

🔬 世界化: 全球科技行业面临AI转型的阵痛

📈 市场化: AI替代人工的进程加速，但社会接受度有限

【原文内容】

Something strange is happening in tech right now. Companies are posting record profits and revenue while laying off tens of thousands of people, citing AI as the official explanation. So far this year, there have been an estimated 363 layoffs at tech companies this year, affecting nearly 150,000 people — a pace of about 974 people per day, 44% faster than last year — according to TrueUp, a tech job board and recruiting platform that also runs one of the most widely cited tech layoff trackers.

The trend appears to be accelerating. Tech layoffs hit their highest single month in two years last month, with nearly 40,000 cuts, and AI was the most-cited reason for layoffs across every industry for the third month running, according to outplacement firm Challenger, Gray & Christmas.

There's growing skepticism that AI is really the culprit, though — that it's more of a convenient cover story than the actual cause. Few examples illustrate the pushback better than what happened at the payments outfit Block. After getting hammered over laying off nearly half the company earlier this year, Jack Dorsey denied the cuts were a sign of trouble, insisting instead that AI tools "are enabling a new way of working which fundamentally changes what it means to build and run a company." But pressed by commenters on X about the bloat he'd created during the pandemic, Dorsey later acknowledged that Block had, in fact, overhired.

Other voices have also begun to weigh in, including famed VC Marc Andreessen, who recently called AI the "silver bullet excuse" for layoffs that are really about mismanagement in some cases. In conversation with podcaster-investor Harry Stebbings, Andreessen said, "Essentially, every large company is overstaffed. It's at least overstaffed by 25%. I think most large companies are overstaffed by 50%. I think a lot of them are overstaffed by 75%. Now they all have the silver bullet excuse: Ah, it's AI."

What makes this combustible is that at the very moment that tens of thousands of workers are being shown the door, a small cohort of AI insiders is becoming wealthy on a scale that's hard to comprehend.

Early last month, AI chipmaker Cerebras Systems closed its first day on the Nasdaq up 68% from its $185 IPO price, giving the chipmaker a market cap of roughly $67 billion — the largest U.S. tech IPO since Snowflake's 2020 debut. By the close, co-founders Andrew Feldman and Sean Lie were billionaires. (The company's shares have since fallen 30%.)

SpaceX meanwhile went public on Friday and enjoys, as of this writing, a $2.1 trillion market cap, turning Musk into a paper trillionaire and potentially minting an estimated 4,400 millionaires and around 400 centimillionaires in the process — assuming the shares don't fall. Anthropic and OpenAI are quickly inching toward the public market, too, both at valuations of roughly $1 trillion or more.

The effects are showing up closer to home, too. In San Francisco — now home to dozens of AI companies, including the big AI labs — high-end homes are routinely selling for millions of dollars over asking price.

Then there's Mark Zuckerberg. In early March, he purchased a $170 million mansion on Miami's "Billionaire Bunker," setting the all-time record for the most expensive home sale in Miami-Dade County history. Two months later, Meta announced it would lay off 8,000 people, or roughly 10% of its workforce.

Tech titans routinely shell out jaw-dropping sums on their real estate portfolios. But these extremes come at a moment when many Americans are getting squeezed harder than they have been in years.

Consider that workers with employer-sponsored health insurance face premium increases of about 6% to 7% this year, more than double the rate of inflation, the cost of private health insurance has roughly doubled since 2008, and median home prices have climbed 28% since early 2020, while mortgage rates have nearly doubled.

In a January 2026 New York Times/Siena poll, 65% of voters said a middle-class lifestyle is out of reach, and a more recent poll found 76% of Americans now name cost of living as their top economic concern, up sharply from 58% a year earlier.

This is about more than job losses in isolation, in short. It's tens of thousands of laid-off workers hitting an unusually unforgiving cost environment at the same time that tens of thousands of AI insiders are seeing once-in-a-generation paper wealth materialize, and being told that AI is why they're out of a job. Whether or not that's the real explanation — many economists point instead to tariffs, war in the Middle East, and broader economic uncertainty as the actual drivers of corporate caution — the optics are what they are. One group is getting unfathomably rich off the advancements that are supposedly replacing the other.

It isn't hard to find a precedent for what happens when that divide gets wide enough. In 2008, a financial crisis that began with loose lending and over-the-top risk-taking on Wall Street ended with bailouts for the banks that caused it, while millions of Americans lost jobs and homes in the Great Recession that followed. Three years later, that anger crystallized into Occupy Wall Street.

That movement could look quaint in comparison if the current trajectory holds. Occupy Wall Street emerged from a crisis and the public anger was, at its core, about who paid for the cleanup. This time, there's no crash to point to. Companies are profitable, AI itself is minting a new class of overnight fortunes, and the layoffs are happening anyway, with AI cited as the driver. If the optics of 2008 were, "We're bailing out the people who broke the economy while you lose your job," the optics here could end up being, "We're getting richer than ever off the very tech we're using to replace you."

Many companies — including Block, Atlassian, Cloudflare — have watched their stocks surge when they point to AI as the reason for cuts, so the strategy makes sense on its face. Still, they might want to consider whether that's really the message they want to send to the people they're laying off, and to everyone else now watching.

【中文翻译】

科技行业现在正在发生一些奇怪的事情。公司在公布创纪录的利润和收入的同时，却以AI为官方解释裁掉数万人。根据TrueUp的数据，今年迄今为止，科技公司估计已裁员363次，影响近15万人——按每天约974人的速度，比去年快44%——TrueUp是一个科技招聘平台，也运营着最广泛引用的科技裁员追踪器之一。这一趋势似乎在加速。根据outplacement公司Challenger, Gray & Christmas的数据，上月科技裁员达到两年来的最高单月水平，近4万人被裁，AI连续第三个月成为各行业被引用最多的裁员原因。然而，人们越来越怀疑AI是否真的是罪魁祸首——它更像是方便的借口而非实际原因。很少有例子比支付公司Block发生的事情更好地说明这种反弹。在今年早些时候因裁掉近一半员工而受到猛烈抨击后，Jack Dorsey否认裁员是麻烦的迹象，反而坚持认为AI工具"正在 enabling 一种新的工作方式，从根本上改变了建设和运营公司的意义"。但在X上被评论者追问他在疫情期间创造的臃肿时，Dorsey后来承认Block实际上确实招聘过度。其他声音也开始加入，包括著名VC Marc Andreessen，他最近称AI为裁员的"银弹借口"，而裁员实际上是关于某些情况下的管理不善。在与播客主持人兼投资者Harry Stebbings的对话中，Andreessen说："基本上，每家大公司都人员过剩。至少超编25%。我认为大多数大公司超编50%。我认为很多公司超编75%。现在他们都有了银弹借口：啊，是AI。"使这种情况变得易燃的是，在数万工人被赶走的同一时刻，一小群AI内部人士正在以难以理解的速度变得富有。上个月初，AI芯片制造商Cerebras Systems在纳斯达克收盘时较其185美元IPO价格上涨68%，使该芯片制造商市值达到约670亿美元——这是自Snowflake 2020年首秀以来最大的美国科技IPO。收盘时，联合创始人Andrew Feldman和Sean Lie成为亿万富翁。（该公司股价此后已下跌30%。）与此同时，SpaceX于周五上市，截至本文撰写时市值达2.1万亿美元，使Musk成为纸面万亿富翁，并可能在此过程中铸造估计4,400名百万富翁和约400名亿万富翁——假设股价不会下跌。Anthropic和OpenAI也迅速走向公开市场，估值都在约1万亿美元或更高。影响也在家门口显现。在旧金山——现在是数十家AI公司的所在地，包括大型AI实验室——高端住宅经常以超过要价数百万美元的价格出售。然后是Mark Zuckerberg。3月初，他在迈阿密的"亿万富翁堡垒"购买了一座1.7亿美元的豪宅，创下了迈阿密-戴德县历史上最昂贵住宅销售记录。两个月后，Meta宣布将裁员8,000人，约占其员工总数的10%。科技巨头 routinely 在房地产组合上花费令人瞠目结舌的巨额资金。但这些极端情况发生在许多美国人面临多年来最严重挤压的时刻。考虑到拥有雇主赞助健康保险的工人今年面临约6%至7%的保费增长，是通胀率的两倍多，自2008年以来私人健康保险成本大约翻了一番，自2020年初以来中位数房价上涨了28%，而抵押贷款利率几乎翻了一番。在2026年1月纽约时报/Siena民调中，65%的选民表示中产阶级生活方式遥不可及，最近的民调发现76%的美国人现在将生活成本列为他们的首要经济关切，比一年前的58%大幅上升。简而言之，这不仅仅是孤立的失业问题。这是数万被裁工人在异常严酷的成本环境中遭遇的，同时数万AI内部人士正在看到一代人一次的纸面财富物质化，并被告知AI是他们失业的原因。这是否真的是解释——许多经济学家指出关税、中东战争和更广泛的经济不确定性才是企业谨慎的实际驱动因素——观感就是观感。一个群体正在从据说要取代另一个群体的进步中变得难以想象地富有。当这种分歧变得足够大时，不难找到发生什么的先例。2008年，一场始于华尔街宽松贷款和过度冒险的金融危机以救助造成危机的银行而告终，而数百万美国人在随后的大衰退中失去了工作和住房。三年后，这种愤怒凝聚成占领华尔街运动。如果当前轨迹持续，相比之下，那场运动可能看起来过时。占领华尔街运动源于一场危机，公众愤怒的核心是谁为清理买单。这一次，没有崩溃可以指出。公司盈利，AI本身正在铸造新一类一夜暴富的人，裁员无论如何都在发生，AI被引用为驱动因素。如果2008年的观感是，"我们在救助搞垮经济的人，而你失去了工作，"那么这里的观感最终可能是，"我们正在从用来取代你的技术中变得比以往更富有。"许多公司——包括Block、Atlassian、Cloudflare——在看到他们将AI作为裁员原因时股价飙升，所以这种策略表面上说得通。尽管如此，他们可能想考虑这是否真的是他们想向被裁员的人以及现在正在关注的每个人发送的信息。

【参考链接】

https://techcrunch.com/2026/06/15/the-ai-layoff-wave-is-becoming-a-powder-keg/

────────────────────

5. SpaceX估值超过亚马逊达2.6万亿美元 🔥🔥🔥🔥

【摘要】

SpaceX在IPO后股价大幅上涨，估值突破2.6万亿美元，短暂超越亚马逊成为全球第五大最有价值公司。

【核心事件】

SpaceX创造历史，成为估值最高的科技公司之一。

【关键细节】

估值突破2.6万亿美元
短暂超越亚马逊成为全球第五
IPO后股价大幅上涨
收购Cursor的消息推动股价
Elon Musk的个人财富进一步增加

【深度分析】

🌐 社会化: 科技巨头的影响力进一步扩大，引发监管担忧

💼 经济化: SpaceX的成功显示资本市场对科技创新的认可

🔬 世界化: 美国科技公司继续主导全球市值排行

📈 市场化: 科技股估值泡沫引发讨论

【原文内容】

SpaceX briefly passed Amazon to become the fifth-most valuable company in the world, and nearly eclipsed Microsoft, before the company's shares pared back those gains before the market closed Tuesday.

The newly public company's stock had already climbed 20% on Monday — its first full day of trading. Tuesday's news that SpaceX was acquiring AI coding company Cursor, along with the start of options trading on SpaceX's shares, sent the share price even higher, spiking its valuation to $2.9 trillion before it ultimately settled back down.

This is all despite the fact that SpaceX posted a $4.9 billion loss on $18.7 billion in revenue last year, compared to Amazon, which turned a $78 billion profit in 2025 on $717 billion in sales in 2025. SpaceX has recently added new revenue streams in the form of compute leasing deals with Anthropic and Google, though, and will absorb the revenue from Cursor when that deal closes in the third quarter.

The Anthropic and Google deals are non-binding, but investors don't seem to mind either way. Elon Musk's space-and-AI company had added roughly $1 trillion to its valuation since going public on Friday.

That transaction netted SpaceX nearly $86 billion in fresh capital, largely on promises that it can create an AI business worth trillions of dollars — a wild claim for a company that recently tore its AI division down to the studs.

SpaceX first revealed a collaboration with Cursor in April, at a time when Musk said his AI company xAI — now a part of SpaceX — "was not built right [the] first time around" and that he was rebuilding it "from the foundations up." SpaceX is making the acquisition with $60 billion in company shares.

SpaceX's historic IPO saw it debut with a valuation of around $1.7 trillion, and the transaction raised nearly $86 billion for Musk's company. SpaceX only made about 4% of its total shares available for trading, which experts predicted would make the stock more susceptible to wild swings.

That appeared to be the case Tuesday, as traders swapped more than 300 million SpaceX shares throughout the trading day — more than half of the 555 million available on the public market post-IPO, according to data from the Nasdaq stock exchange.

The volatility continued into after-hours trading, which saw SpaceX's valuation briefly eclipse Amazon's market cap for a second time before falling again.

【中文翻译】

SpaceX短暂超越亚马逊成为全球第五大最有价值的公司，并几乎超越微软，但在周二收盘前该公司股价回吐了部分涨幅。这家新上市公司的股票在周一——其首个完整交易日——已经上涨了20%。周二SpaceX收购AI编程公司Cursor的消息，加上SpaceX股票期权交易的开始，使股价进一步上涨，估值飙升至2.9万亿美元，然后最终回落。尽管SpaceX去年在187亿美元收入上亏损49亿美元，而亚马逊在2025年7170亿美元销售额上获得780亿美元利润。不过，SpaceX最近通过与Anthropic和Google的计算租赁协议增加了新的收入来源，并将在第三季度交易完成时吸收Cursor的收入。Anthropic和Google的交易没有约束力，但投资者似乎并不在意。Elon Musk的太空和AI公司自周五上市以来估值增加了约1万亿美元。这笔交易为SpaceX净赚近860亿美元的新资本，主要是基于它能够创造价值数万亿美元的AI业务的承诺——对于一家最近将其AI部门彻底重建的公司来说，这是一个疯狂的主张。SpaceX首次在4月透露与Cursor合作，当时Musk说他的AI公司xAI——现在是SpaceX的一部分——"第一次没有建好"，他正在"从头开始重建"。SpaceX正在用600亿美元的公司股票进行收购。SpaceX的历史性IPO使其以约1.7万亿美元的估值首秀，并为Musk的公司筹集了近860亿美元。SpaceX仅将其总股份的约4%用于交易，专家预测这将使股票更容易出现大幅波动。周二似乎就是这种情况，根据纳斯达克交易所的数据，交易员在整个交易日交换了超过3亿股SpaceX股票——超过IPO后公开市场上5.55亿股的一半。波动性延续到盘后交易，SpaceX的估值第二次短暂超越亚马逊市值，然后再次下跌。

【参考链接】

https://techcrunch.com/2026/06/16/spacex-valuation-balloons-to-2-6t-briefly-passes-amazon/

────────────────────

6. OpenAI秘密提交IPO申请 🔥🔥🔥🔥

【摘要】

ChatGPT制造商OpenAI已秘密提交IPO申请，紧随其主要竞争对手Anthropic之后。

【核心事件】

OpenAI和Anthropic相继提交IPO，2026年将成为科技IPO大年。

【关键细节】

OpenAI最新估值8520亿美元
在Anthropic提交IPO后约一周
未披露时间安排或融资目标
SpaceX也预计以1.75万亿估值上市
三家公司可能在数月内相继上市

【深度分析】

🌐 社会化: AI公司上市将提高透明度，接受公众监督

💼 经济化: AI公司IPO可能成为今年最大的资本市场事件

🔬 世界化: 美国AI公司继续引领全球AI产业发展

📈 市场化: AI公司估值面临公开市场检验

【原文内容】

ChatGPT-maker OpenAI has filed confidentially for an initial public offering, the company announced Monday in a blog post. The filing comes a little more than a week after its main rival, Anthropic, also filed to go public, ramping up the race between the two AI firms.

OpenAI, which was last valued at $852 billion post-money, submitted a draft registration statement to the U.S. Securities and Exchange Commission for a proposed IPO. OpenAI hasn't shared any specifics yet. However, the company said it posted the blog because it expected a leak.

"We have not decided on timing yet; it may be a while because there are things we want to do that are likely easier as a private company," the company wrote. "But it's a complicated set of tradeoffs and this gives us the option to go public sooner if that ends up being best."

Around the same time, and in a separate blog post, OpenAI published a sweeping philosophical statement about its mission, its vision for AGI, and its belief that AI should benefit all of humanity — the kind of forward-looking communication that companies entering a quiet period have historically been careful to avoid. That OpenAI appears comfortable publishing it so close to a confidential filing says something — not necessarily about its own legal judgment but about the regulatory environment it's operating in. The SEC under the Trump administration has taken a markedly more hands-off posture toward tech and AI companies than it did under previous administrations, and OpenAI may simply be reading the room.

Whatever the regulatory questions, the filing is the latest signal that 2026 will be a blockbuster year for the public markets. SpaceX is also expected to make its debut at a $1.75 trillion valuation, meaning three of the most closely watched companies in tech could all go public within months of each other — a concentration of high-stakes offerings the markets haven't seen since the dot-com boom.

OpenAI is racing to IPO even as it recently missed its own targets for new users and revenue, per The Wall Street Journal. Its chief financial officer, Sarah Friar, has reportedly raised concerns that OpenAI may not be able to support its massive data center spending. And the burn does appear to be massive.

In late March, OpenAI secured $122 billion in the largest funding round in Silicon Valley history — $3 billion of which came directly from retail investors via bank channels. But the firm expects to spend roughly that same amount on computing power for AI research alone in 2028, and projects burning $85 billion that year even after doubling sales from the year prior, per The Wall Street Journal. Put another way, OpenAI is asking public market investors to buy into a business that, by its own projections, won't generate more cash than it spends for at least four more years.

SpaceX offers a parallel data point. Its AI spending, while not as massive, illustrates how the cost to train large language models can exceed the revenue those models generate — a structural challenge the entire industry is grappling with, and one that public market investors will have to price.

Anthropic, on the other hand, has provided investors a much rosier picture of its financials, saying that it is close to achieving its first quarterly profit. Even so, with a recent $65 billion funding round and another $36 billion in chip-allocated debt potentially on its way, Anthropic's burn rate isn't exactly modest.

The confidential IPO filing allows OpenAI to start its preparation for a public offering without publicly disclosing detailed financial information or business risks, which is why the company hasn't shared stock pricing or how much it hopes to raise yet. That said, the secondary markets provide a glimpse into what investors are willing to pay.

Anthropic recently surged to a $1 trillion valuation on Forge Global, a retail secondary market platform, surpassing OpenAI, which was recorded at around $880 billion in April.

David Shapiro is founder and CEO of OpenVC and overseer of the NYSE OpenVC 500 Index, which tracks the largest public and private companies in the U.S. He said Anthropic's rate of appreciation far exceeds OpenAI this year — 123% year-to-date versus OpenAI's 11.3%. That said, despite Anthropic's clear boost, OpenAI isn't seeing a lack of secondary interest.

"From a secondary investor standpoint, OpenAI had already grown into a significant portion of its valuation," Shapiro told TechCrunch. "We haven't seen OpenAI crater or anything close, and the valuation is still enormously successful, according to the index."

He added that OpenAI's stock in the secondary market "experienced a slight pop over the last few days, indicating investors may be pricing both as the 'dual winners' of the broader LLM race."

But the race to get to the public markets first is a real concern. Experts say whoever makes their debut first will likely nab more of what is becoming increasingly scarce capital for AI companies — much of which may have already been absorbed by SpaceX, which is expected to IPO first among the three.

Additionally, Anthropic's filing disclosures will set a valuation comp that constrains how OpenAI can price its own offering when it files, according to a recent PitchBook report that characterized OpenAI as overvalued relative to its fundamentals. In other words, if Anthropic prices conservatively, OpenAI's path to its target valuation gets harder.

OpenAI was founded in 2015 as a nonprofit research lab and disrupted the world of AI when it released ChatGPT in 2022, sparking a wave of large language model advancements across the industry.

While OpenAI has expanded its products to accommodate enterprise and government customers, the firm has a strong reputation of being more consumer-focused than rival Anthropic. The company has built real scale, with around 900 million weekly active users.

The IPO comes after significant internal struggles within the company. In 2023, OpenAI's board ousted Sam Altman over what it described as a lack of transparency and concerns about whether he was committed to the firm's mission of benefiting all humanity. Altman was quickly reinstated, and the board members who were involved in the coup, including co-founder Ilya Sutskever, departed shortly after. The episode raised governance questions that have never been fully resolved and that prospective public investors will likely scrutinize closely.

More recently, OpenAI has faced several lawsuits, including a recent one from the state of Florida accusing the company and Altman of harming children by providing information to school shooters, offering guidance on self-harm, and fostering addiction among young users. Florida's complaint adds to the litany of lawsuits against OpenAI and other chatbot makers following user delusions, self-harm, suicide, and mass casualty events.

Last month, OpenAI prevailed at trial after co-founder and rival Elon Musk sued the company and Altman over an alleged promise to keep the company a nonprofit. The case was ultimately tossed out after both a jury and judge found Musk had waited too long — he was beyond the statute of limitations when he filed the case in 2024.

【中文翻译】

ChatGPT制造商OpenAI周一在博客文章中宣布已秘密提交首次公开募股申请。这一申请距离其主要竞争对手Anthropic也提交上市申请仅一周多时间，加剧了两家AI公司之间的竞争。OpenAI最新投后估值为8520亿美元，已向美国证券交易委员会提交了拟议IPO的注册声明草案。OpenAI尚未分享任何具体信息。然而，该公司表示发布博客是因为预计会有泄露。公司写道："我们还没有决定时间安排；可能需要一段时间，因为有些事情作为私人公司做起来可能更容易。但这是一系列复杂的权衡，这给了我们如果最终证明是最好的话可以更早上市的选择。"与此同时，OpenAI在另一篇博客文章中发表了关于其使命、AGI愿景以及AI应造福全人类信念的全面哲学声明——这种前瞻性沟通是进入静默期的公司历来谨慎避免的。OpenAI似乎在秘密提交如此接近的时候舒适地发布它，这说明了一些什么——不一定关于其自身的法律判断，而是关于它所处的监管环境。特朗普政府下的SEC对科技和AI公司采取了比前任政府明显更不干预的姿态，OpenAI可能只是在审时度势。无论监管问题如何，这一申请是2026年将成为公开市场大年头的最新信号。SpaceX也预计以1.75万亿美元的估值首秀，这意味着科技领域最受关注的三家公司可能在数月内相继上市——这是自互联网繁荣以来市场未曾见过的高风险发行集中。据华尔街日报报道，OpenAI正在竞相上市，尽管最近未达到自己的新用户和收入目标。其首席财务官Sarah Friar据报对OpenAI可能无法支持其大规模数据中心支出表示担忧。而支出确实看起来很大。3月下旬，OpenAI以硅谷历史上最大的融资轮获得了1220亿美元——其中30亿美元直接来自散户投资者通过银行渠道。但该公司预计到2028年仅在AI研究的计算能力上就将花费大致相同的金额，并预测即使销售额翻倍后那年仍将烧掉850亿美元，据华尔街日报报道。换句话说，OpenAI正要求公开市场投资者买入一家按其自身预测至少四年内不会产生比支出更多现金的企业。SpaceX提供了一个平行的数据点。其AI支出虽然不那么庞大，但说明了训练大语言模型的成本如何超过这些模型产生的收入——这是整个行业正在应对的结构性挑战，也是公开市场投资者必须定价的。另一方面，Anthropic向投资者提供了更为乐观的财务图景，表示接近实现其首个季度盈利。即便如此，随着最近的650亿美元融资轮和可能到来的360亿美元芯片配置债务，Anthropic的烧钱率也不算温和。秘密IPO申请允许OpenAI在不公开披露详细财务信息或商业风险的情况下开始上市准备，这就是为什么公司尚未分享股票定价或希望筹集多少资金。不过，二级市场提供了投资者愿意支付多少的一瞥。Anthropic最近在零售二级市场平台Forge Global上飙升至1万亿美元估值，超过了4月份记录约8800亿美元的OpenAI。OpenVC创始人兼CEO、NYSE OpenVC 500指数负责人David Shapiro表示，Anthropic今年的升值速度远超OpenAI——年初至今123%对比OpenAI的11.3%。不过，尽管Anthropic明显提升，OpenAI并未缺乏二级市场兴趣。"从二级市场投资者角度来看，OpenAI已经增长到了其估值的很大一部分，"Shapiro告诉TechCrunch。"我们没有看到OpenAI崩盘或接近崩盘，根据指数，估值仍然非常成功。"他补充说，OpenAI在二级市场的股票"在过去几天略有上涨，表明投资者可能将两者都定价为更广泛LLM竞赛的'双重赢家'。"但率先上市的竞赛是一个真正的关切。专家说，无论谁先亮相都可能抢占对AI公司日益稀缺的资本中的更多份额——其中大部分可能已被SpaceX吸收，SpaceX预计将在三家中率先IPO。此外，根据PitchBook最近一份将OpenAI描述为相对基本面估值过高的报告，Anthropic的申请披露将设定估值比较，限制OpenAI在申请时如何定价自己的发行。换句话说，如果Anthropic定价保守，OpenAI达到目标估值的路径就更难。OpenAI于2015年作为非营利研究实验室成立，在2022年发布ChatGPT时颠覆了AI世界，引发了整个行业的大语言模型进步浪潮。虽然OpenAI已扩展产品以适应企业和政府客户，但该公司以比竞争对手Anthropic更注重消费者而闻名。该公司建立了真正的规模，拥有约9亿周活跃用户。IPO是在公司内部重大斗争之后到来的。2023年，OpenAI董事会以其描述为缺乏透明度以及对他是否致力于公司造福全人类使命的担忧为由罢免了Sam Altman。Altman很快被恢复职务，参与政变的董事会成员，包括联合创始人Ilya Sutskever，不久后离开。这一事件引发了从未完全解决的治理问题，潜在公开市场投资者可能会仔细审查。最近，OpenAI面临多起诉讼，包括佛罗里达州最近的一起诉讼，指控该公司和Altman通过向学校枪手提供信息、提供自残指导以及在年轻用户中培养成瘾来伤害儿童。佛罗里达州的投诉增加了对OpenAI和其他聊天机器人制造商在用户妄想、自残、自杀和大规模伤亡事件后的一系列诉讼。上个月，OpenAI在联合创始人兼竞争对手Elon Musk起诉该公司和Altman涉嫌承诺保持公司非营利性质后在审判中胜诉。该案最终被驳回，因为陪审团和法官都发现Musk等待时间太长——他在2024年提起诉讼时已超过诉讼时效。

【参考链接】

https://techcrunch.com/2026/06/08/following-anthropic-openai-files-confidentially-for-ipo/

────────────────────

7. Anthropic估值达9650亿美元提交IPO 🔥🔥🔥🔥🔥

【摘要】

Anthropic完成650亿美元融资，估值达9650亿美元，超越OpenAI成为最有价值的AI初创公司。

【核心事件】

Anthropic创造历史，成为全球最有价值的AI初创公司。

【关键细节】

估值达9650亿美元
完成650亿美元Series H融资
超越OpenAI的8520亿估值
Claude Code年化收入超25亿美元
年化收入运行率超470亿美元

【深度分析】

🌐 社会化: AI安全公司获得巨额估值，显示市场对安全AI的需求

💼 经济化: Anthropic的成功证明AI商业模式可行

🔬 世界化: 美国AI公司在资本市场占据主导地位

📈 市场化: AI初创公司估值达到前所未有的高度

【原文内容】

The latest Anthropic funding round marks a major escalation in the race to build and commercialize frontier AI systems. The company behind Claude is now valued close to $1 trillion, driven by rapid enterprise adoption, rising demand for AI coding tools, and heavy investor appetite for companies building advanced AI infrastructure.

The round also arrives as OpenAI, Anthropic, xAI, Google, Meta, and other AI companies compete for the same enterprise customers, cloud capacity, developer workflows, and public-market attention.

Key facts:

- Anthropic raised $65 billion in Series H funding.

- The round valued the company at $965 billion post-money.

- The valuation surpasses OpenAI's reported March 2026 valuation of $852 billion.

- Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital led the round.

- Co-leads included Capital Group, Coatue, D1 Capital Partners, GIC, ICONIQ, and XN.

- The raise included $15 billion of previously committed hyperscaler investments, including $5 billion from Amazon.

- Anthropic's valuation has moved from $61.5 billion in March 2025 to $380 billion in February 2026 and $965 billion in May 2026.

- Anthropic reported that Claude Code, its agentic coding tool, had passed $2.5 billion in run-rate revenue by February 2026.

- The company's annualized revenue run rate reportedly surpassed $47 billion by May 2026, up from more than $30 billion in April and about $9 billion at the end of 2025.

- Enterprise demand is central to the growth story, with more than 1,000 business customers each spending over $1 million annually by April 2026.

- Anthropic also introduced Claude Opus 4.8, a new flagship model with stronger performance across coding, agentic tasks, and professional work.

- Amazon, Google, Broadcom, Micron, Samsung, and SK Hynix are now part of Anthropic's wider funding, cloud, chip, and infrastructure ecosystem.

Claude Code Is the Product Behind the Momentum

Claude Code is one of the main reasons investors are treating Anthropic as a top-tier enterprise AI company. The tool gives developers an agentic coding assistant that can work through programming tasks, inspect codebases, suggest changes, and support software-building workflows. Anthropic said in February 2026 that Claude Code had grown to more than $2.5 billion in run-rate revenue.

The launch of Claude Opus 4.8 strengthens the same strategy. Anthropic is positioning its highest-end models around coding, agentic tasks, and professional work rather than only general chatbot use.

Coding is one of the clearest commercial use cases for generative AI. Businesses can measure time saved, software output, developer productivity, and infrastructure impact more directly than in many general chatbot use cases.

The Business Impact

Anthropic's valuation reflects four business trends happening at once.

First, enterprise customers are moving beyond chatbot experimentation and into workflow-level AI deployment. Coding, customer support, research, document analysis, and internal automation are becoming major categories of AI spending.

Second, frontier AI companies need enormous capital. Training and serving advanced models requires cloud contracts, chips, data centers, engineering teams, safety work, and enterprise support. Large funding rounds are becoming part of the operating model.

Third, infrastructure partnerships are becoming as important as software demand. Anthropic has signed capacity agreements with Amazon for up to five gigawatts of new capacity, with Google and Broadcom for five gigawatts of next-generation TPU capacity, and with SpaceX for GPU capacity. Micron, Samsung, and SK Hynix also add strategic memory and chip-supply relevance to the round.

Fourth, enterprise AI is moving from software access to implementation services. On May 4, 2026, Anthropic, Blackstone, Hellman & Friedman, and Goldman Sachs announced a new AI-native enterprise services firm designed to help mid-sized businesses deploy Claude inside core operations.

The Risks

Anthropic's growth story comes with major caveats.

The company's revenue figures are not yet reported under public-company disclosure standards. Investors will want more clarity on how revenue and infrastructure costs are being recognized once Anthropic prepares for an IPO.

Compute costs are another major issue. Frontier AI companies can generate large revenue while also spending heavily on model training, inference, cloud capacity, and specialized chips. Strong demand does not automatically mean durable profitability.

Anthropic also faces legal pressure. It agreed to a $1.5 billion settlement in a copyright case involving authors and publishers, while music publishers including Universal Music Group, Concord, and ABKCO have pursued separate copyright claims involving lyrics and sheet music.

The company is also challenging a U.S. government supply-chain-risk designation issued in March 2026. A federal judge temporarily blocked enforcement, but the dispute remains part of Anthropic's broader public-sector risk profile.

These issues will matter more if Anthropic moves toward public markets.

【中文翻译】

最新的Anthropic融资轮标志着构建和商业化前沿AI系统的竞赛重大升级。Claude背后的公司现在估值接近1万亿美元，受到企业快速采用、AI编程工具需求增长以及投资者对构建先进AI基础设施公司的强烈追捧驱动。这一轮也正值OpenAI、Anthropic、xAI、Google、Meta和其他AI公司争夺相同的企业客户、云容量、开发者工作流和公开市场关注。关键事实：- Anthropic在Series H融资中筹集了650亿美元。- 该轮投后估值为9650亿美元。- 估值超过OpenAI报告的2026年3月估值8520亿美元。- Altimeter Capital、Dragoneer、Greenoaks和Sequoia Capital领投。- 联合领投包括Capital Group、Coatue、D1 Capital Partners、GIC、ICONIQ和XN。- 融资包括150亿美元此前承诺的超大规模投资者投资，包括亚马逊的50亿美元。- Anthropic的估值从2025年3月的615亿美元上升到2026年2月的3800亿美元和2026年5月的9650亿美元。- Anthropic报告其代理编程工具Claude Code到2026年2月年化收入已超过25亿美元。- 公司年化收入运行率据报到2026年5月超过470亿美元，高于4月的300多亿和2025年底的约90亿美元。- 企业需求是增长故事的核心，到2026年4月超过1000家企业客户每家年支出超过100万美元。- Anthropic还推出了Claude Opus 4.8，一款在编程、代理任务和专业工作方面表现更强的新旗舰模型。- 亚马逊、谷歌、博通、美光、三星和SK海力士现在是Anthropic更广泛的融资、云、芯片和基础设施生态系统的一部分。Claude Code是势头背后的产品Claude Code是投资者将Anthropic视为顶级企业AI公司的主要原因之一。该工具为开发者提供代理编程助手，可以处理编程任务、检查代码库、建议更改并支持软件构建工作流。Anthropic在2026年2月表示，Claude Code年化收入已增长到超过25亿美元。Claude Opus 4.8的发布加强了同一战略。Anthropic正在将其最高端模型定位在编程、代理任务和专业工作方面，而不仅仅是通用聊天机器人使用。编程是生成式AI最清晰的商业用例之一。企业可以比许多通用聊天机器人用例更直接地衡量节省的时间、软件产出、开发者生产力和基础设施影响。商业影响Anthropic的估值反映了同时发生的四个商业趋势。首先，企业客户正在从聊天机器人实验转向工作流级AI部署。编程、客户支持、研究、文档分析和内部自动化正在成为AI支出的主要类别。其次，前沿AI公司需要大量资本。训练和服务先进模型需要云合同、芯片、数据中心、工程团队、安全工作和企业支持。大型融资轮正在成为运营模式的一部分。第三，基础设施合作伙伴关系变得与软件需求一样重要。Anthropic已与亚马逊签署高达5吉瓦新容量协议，与谷歌和博通签署5吉瓦下一代TPU容量协议，并与SpaceX签署GPU容量协议。美光、三星和SK海力士也为这一轮增加了战略内存和芯片供应相关性。第四，企业AI正在从软件访问转向实施服务。2026年5月4日，Anthropic、Blackstone、Hellman & Friedman和高盛宣布成立一家新的AI原生企业服务公司，旨在帮助中型企业在核心运营中部署Claude。风险Anthropic的增长故事伴随着重大警告。公司的财务数据尚未按上市公司披露标准报告。一旦Anthropic准备IPO，投资者将希望更清楚地了解收入和基础设施成本如何确认。计算成本是另一个主要问题。前沿AI公司可以产生大量收入，同时在模型训练、推理、云容量和专业芯片上大量支出。强劲需求并不自动意味着持久盈利。Anthropic还面临法律压力。它在涉及作者和出版商的版权案件中同意了15亿美元的和解，而包括环球音乐集团、Concord和ABKCO在内的音乐出版商已经追求涉及歌词和乐谱的独立版权索赔。公司还在挑战美国政府2026年3月发布的供应链风险认定。联邦法官暂时阻止了执行，但争议仍然是Anthropic更广泛的公共部门风险状况的一部分。如果Anthropic走向公开市场，这些问题将变得更加重要。

【参考链接】

https://theaitrack.com/anthropic-funding-round-965-billion-valuation/

────────────────────

8. Google Gemini 3.5 Pro即将发布 🔥🔥🔥🔥

【摘要】

Google的Gemini 3.5 Pro即将正式发布，支持200万token上下文窗口和Deep Think推理模式。

【核心事件】

Google在AI模型竞赛中推出重磅产品，挑战OpenAI和Anthropic。

【关键细节】

支持200万token上下文窗口
包含Deep Think推理模式
目标6月正式发布
目前在Vertex AI有限预览
与OpenAI和Anthropic展开正面竞争

【深度分析】

🌐 社会化: 超长上下文窗口将改变人机交互方式

💼 经济化: Google通过Gemini强化企业级市场布局

🔬 世界化: Google在AI领域的技术实力再次得到验证

📈 市场化: AI模型市场竞争更加激烈

【原文内容】

Google's most powerful AI model of the year is expected to arrive within weeks. Gemini 3.5 Pro, unveiled at Google I/O on May 19, is slated for general availability in June 2026 — a launch Google has teased but not yet delivered, positioned to compete at the frontier of AI reasoning and multimodal capability.

What Google Has Promised

Gemini 3.5 Pro is the high-end tier of Google's latest model family, and the company has outlined an ambitious feature set. Pro targets a 2-million-token context window, a "Deep Think" reasoning mode, and frontier multimodal understanding — the ability to work across text, images, and other formats. In Google's lineup, Pro absorbs the use cases the company previously routed to its top "Ultra" tier: the hardest reasoning, deep multimodal tasks, and very long context.

The 2-million-token context window is one of the headline specifications. Context length determines how much material a model can consider at once, and a window that large would let Gemini 3.5 Pro hold enormous amounts of text — long documents, large codebases, extended conversations — in working memory. The "Deep Think" mode points to Google's bet on models that spend more effort reasoning through complex problems rather than answering quickly.

Still Waiting, As Of Early June

The notable wrinkle is that, as of early June, Gemini 3.5 Pro has not actually shipped. At I/O, Google chief executive Sundar Pichai told the audience, in effect, to wait another month — a line that reportedly drew audible groans from a crowd hoping for immediate access. The model has been in internal use and limited preview through Google's enterprise platform, with broad availability still pending.

That gap between announcement and release is worth keeping in mind. Until the model is generally available and independent testers can evaluate it, its capabilities remain Google's claims rather than verified performance. Launch timelines can also slip, so "June" should be read as the target rather than a guarantee.

How It Fits Google's Strategy

Gemini 3.5 Pro arrives after Google already shipped the faster, cheaper Gemini 3.5 Flash earlier in the spring. Flash reportedly improved on the previous generation's Pro model in coding and agentic tasks but regressed on the hardest reasoning — precisely the gap the new Pro tier is meant to close. The two-model structure reflects a common industry pattern: a fast, affordable model for high-volume work, and a more capable, more expensive model for demanding tasks.

Pricing for Pro is expected to follow the same ratio as prior generations, roughly ten times the cost of Flash, which would put it around $15 per million input tokens and $60 per million output tokens. That would position it competitively against rival frontier models from Anthropic and OpenAI. Google plans to make the model available first through its consumer subscriptions — the $20-a-month Pro plan and the $250-a-month Ultra plan — with Ultra subscribers also getting the Deep Think reasoning feature.

The Competitive Backdrop

The launch lands in an intensely competitive moment for frontier AI. Rivals are shipping capable models at a rapid pace, and price competition has intensified as challengers, including aggressively priced models from Chinese labs, push the cost of capable AI down. Google's strategy has leaned on breadth and distribution — offering models across price points and weaving them into its products and cloud — rather than betting everything on a single best model.

Gemini 3.5 Pro is the piece of that strategy aimed at the top of the market: the reasoning-heavy, long-context, multimodal work where the most demanding users and enterprises operate. If it delivers on its specifications, it gives Google a credible flagship to set against the best from its competitors. If it underdelivers or slips, it extends the perception that Google announces ahead of shipping.

What To Watch

The immediate thing to watch is whether the model ships on schedule and how it performs once independent evaluators can test it. Benchmarks and real-world use will determine whether the 2-million-token context and Deep Think reasoning translate into a meaningful advantage, or whether the gains are incremental.

For users and developers, the practical advice is to treat the announced specifications as a target until the model is generally available and tested. Gemini 3.5 Pro is shaping up to be one of the more significant model launches of the month, but as of early June it remains a promise Google has made rather than a product anyone outside limited previews can fully evaluate. The coming weeks should turn that promise into something concrete.

【中文翻译】

Google今年最强大的AI模型预计将在几周内到来。Gemini 3.5 Pro于5月19日在Google I/O上亮相，计划于2026年6月正式推出——这是Google已经预告但尚未交付的发布，定位在AI推理和多模态能力的前沿竞争。Google承诺的功能Gemini 3.5 Pro是Google最新模型家族的高端层级，公司概述了一个雄心勃勃的功能集。Pro目标是200万token上下文窗口、"Deep Think"推理模式和前沿多模态理解——跨文本、图像和其他格式工作的能力。在Google的产品线中，Pro吸收了公司之前路由到其顶级"Ultra"层级的用例：最困难的推理、深度多模态任务和超长上下文。200万token上下文窗口是标题规格之一。上下文长度决定了模型一次可以考虑多少材料，这么大的窗口将让Gemini 3.5 Pro在工作记忆中容纳大量文本——长文档、大型代码库、扩展对话。"Deep Think"模式指向Google对模型的押注，这些模型在推理复杂问题上花费更多精力，而不是快速回答。截至6月初仍在等待值得注意的曲折是，截至6月初，Gemini 3.5 Pro实际上尚未发布。在I/O上，Google首席执行官Sundar Pichai告诉观众，实际上再等一个月——据报道，这句话引起了希望立即访问的观众可以听到的呻吟声。该模型一直在内部使用，并通过Google的企业平台进行有限预览，广泛可用性仍然待定。发布和推出之间的差距值得记住。在模型普遍可用且独立测试人员可以评估之前，其功能仍然是Google的声明而非经过验证的性能。发布时间表也可能推迟，所以"6月"应被视为目标而非保证。如何融入Google的战略Gemini 3.5 Pro在Google已经在春季早些时候推出更快、更便宜的Gemini 3.5 Flash之后到来。据报道，Flash在编码和代理任务方面改进了上一代的Pro模型，但在最困难的推理方面有所退步——这正是新的Pro层级旨在弥补的差距。双模型结构反映了常见的行业模式：用于高容量工作的快速、经济型模型，以及用于要求高的任务的更强大、更昂贵的模型。Pro的定价预计将遵循与以前几代相同的比例，大约是Flash成本的十倍，这将使其约为每百万输入token 15美元和每百万输出token 60美元。这将使其在与Anthropic和OpenAI的竞争对手前沿模型相比时具有竞争力。Google计划首先通过其消费者订阅提供该模型——每月20美元的Pro计划和每月250美元的Ultra计划——Ultra订阅者还将获得Deep Think推理功能。竞争背景发布正值前沿AI竞争激烈的时刻。竞争对手正在快速发布功能强大的模型，随着包括来自中国的激进定价模型在内的挑战者推动降低能力AI的成本，价格竞争加剧。Google的战略依赖于广度和分发——在不同价格点提供模型并将它们编织到其产品和云中——而不是将所有赌注押在单一最佳模型上。Gemini 3.5 Pro是该战略中针对市场顶层的部分：最苛刻的用户和企业运营的推理密集型、长上下文、多模态工作。如果它实现了其规格，它将为Google提供一个可信的旗舰产品，与竞争对手的最佳产品抗衡。如果它表现不佳或推迟，它将延续Google提前发布的感觉。值得关注的内容直接要关注的是模型是否按计划发布，以及一旦独立评估人员可以测试时它的表现如何。基准测试和实际使用将确定200万token上下文和Deep Think推理是否转化为有意义的优势，或者收益是否是渐进的。对于用户和开发者，实用的建议是在模型普遍可用并经过测试之前，将公布的规格视为目标。Gemini 3.5 Pro正在成为本月更重要的模型发布之一，但截至6月初，它仍然是Google做出的承诺，而不是有限预览之外的任何人都可以完全评估的产品。未来几周应该将这一承诺变成具体的东西。

【参考链接】

https://www.techtimes.com/articles/317919/20260606/google-gemini-35-pro-nears-june-launch-2-million-token-context-deep-think-reasoning.htm

────────────────────

9. Apple与Google、Nvidia合作开发AI模型 🔥🔥🔥🔥

【摘要】

Apple正在与Google和Nvidia合作开发其最先进的AI模型Apple Foundation Model Cloud Pro。

【核心事件】

Apple打破传统，与竞争对手合作开发AI技术。

【关键细节】

新模型名为Apple Foundation Model Cloud Pro
与Google和Nvidia合作开发
基于Google的Gemini技术
标志着Apple AI战略的重大转变
Apple不再坚持完全自主研发

【深度分析】

🌐 社会化: 科技巨头之间的合作将加速AI技术普及

💼 经济化: Apple的开放策略可能改变整个行业的合作模式

🔬 世界化: AI技术生态更加开放和协作

📈 市场化: AI模型开发从竞争走向合作

【原文内容】

Apple on Monday revealed what it's been working on in artificial intelligence at its annual Worldwide Developers Conference in Cupertino, Calif.

WWDC showed off demos of its redesigned Siri, which can speak back and forth with the user, a major improvement over previous versions of the assistant. In a demo, Siri was able to check concert dates, set a reminder to buy tickets, and even get directions to pick up a friend on the way to the concert venue.

But the announcement also highlighted that Apple has taken a different strategy to many of its Silicon Valley rivals, choosing not to spend billions on infrastructure and the biggest, most advanced models, and instead focusing its message to potential customers on privacy advantages and convenience.

Apple executives highlighted the difference in remarks on Monday.

"Some appear to be racing forward, seemingly pursuing AI for the sake of AI, without clear regard for the people — all of us — that it's ultimately meant to serve," said Apple software SVP Craig Federighi in the launch announcement.

But it turns out two of the traditional AI leaders, Google and Nvidia, are helping Apple out with its most advanced model, called Apple Foundation Model Cloud Pro, Apple executives told media in a talk at its headquarters on Monday.

While Apple and Google announced their partnership for Apple Intelligence in January, this is the first time that the company has officially confirmed that some of its Apple Intelligence features will run on Nvidia chips.

Apple AI executive Amar Subramanya said that AFM Cloud Pro is comparable to Google's Gemini frontier models. It will run in the cloud on Nvidia GPUs, which are part of Apple's Private Cloud Compute infrastructure, Apple officials said.

"We work with both Google and Nvidia to extend our private cloud compute infrastructure to Nvidia GPUs in Google's cloud, while maintaining Apple's unmatched privacy guarantees," Subramanya said.

VP of software Sebastian Marineau-Mes said that Apple wanted to use Nvidia's latest chips, but Apple wanted the chips to be configured in a more private way, where they couldn't read what was on the servers.

Marineau-Mes said that recent Nvidia improvements, such as a technology called "Nvidia confidential compute," allowed Apple and Google to build a system that met its standards.

"We wanted to avail ourselves of the latest technology from Nvidia, and so we set out to extend private cloud compute to third-party cloud," Marineau-Mes said.

Apple is distinguishing itself from companies that have more heavily invested in AI by emphasizing that its software is more private — the company isn't collecting as much data as web-based AI such as OpenAI's ChatGPT or Anthropic's Claude — and that it is using its access to locally-stored user information like a calendar or text messages to personalize AI features.

The tech talk on Monday was held so that Federighi and his lieutenants could discuss how Apple built Siri AI and its Apple Intelligence layer and how it differs from AI that consumers might already be familiar with.

Apple executives outlined an architecture for the software in which Apple's operating systems and software have a piece of software called a system orchestrator that routes any AI query to an appropriate model — either on the device or in the cloud — depending on how much computing power and personal data it needs.

The system orchestrator is "key to the privacy architecture of our entire system," said Federighi.

The talk also provided some more details on the Apple-Google partnership.

Federighi said that Apple Intelligence, the AI software built into the company's devices, uses Apple's own models, not the same Google Gemini available to the public, as many people in the tech industry expected when the two companies announced a partnership in January. He also said it wasn't using Google's off-the-shelf cloud infrastructure.

Federighi said that Google's technology was used to help build Apple's own models — specifically, "third-generation" AFM models for the cloud announced on Monday that are designed and turned to run on Apple's chips.

"These four models that we just talked about — AFM Core, Core Advanced Cloud, and Cloud Image —all of these are custom built for Apple Silicon, trained using proprietary data with reinforcement learning and refined using outputs from Gemini frontier models," Subramanya said.

【中文翻译】

Apple周一在加州库比蒂诺举行的年度全球开发者大会上展示了其在人工智能方面的工作。WWDC展示了其重新设计的Siri的演示，它可以与用户来回对话，这是对以前版本助手的重大改进。在演示中，Siri能够检查音乐会日期、设置购买门票的提醒，甚至获取在去音乐会场地途中接朋友的路线。但这一公告也突显出Apple采取了与其硅谷许多竞争对手不同的战略，选择不花费数十亿美元在基础设施和最大、最先进的模型上，而是将其信息聚焦于潜在客户的隐私优势和便利性。Apple高管在周一的讲话中强调了这种差异。Apple软件高级副总裁Craig Federighi在发布声明中说："有些人似乎在向前冲刺，似乎为了AI而追求AI，没有清楚地考虑到它最终要服务的人——我们所有人。"但事实证明，两位传统AI领导者Google和Nvidia正在帮助Apple开发其最先进的模型，名为Apple Foundation Model Cloud Pro，Apple高管周一在其总部的一次谈话中告诉媒体。虽然Apple和Google在1月宣布了它们对Apple Intelligence的合作伙伴关系，但这是该公司首次正式确认其一些Apple Intelligence功能将在Nvidia芯片上运行。Apple AI高管Amar Subramanya表示，AFM Cloud Pro与Google的Gemini前沿模型相当。Apple官员表示，它将在Google云中的Nvidia GPU上运行，这是Apple私有云计算基础设施的一部分。Subramanya说："我们与Google和Nvidia合作，将我们的私有云计算基础设施扩展到Google云中的Nvidia GPU，同时保持Apple无与伦比的隐私保证。"软件副总裁Sebastian Marineau-Mes表示，Apple希望使用Nvidia的最新芯片，但Apple希望芯片以更私密的方式配置，使它们无法读取服务器上的内容。Marineau-Mes表示，最近的Nvidia改进，如称为"Nvidia机密计算"的技术，使Apple和Google能够构建符合其标准的系统。Marineau-Mes说："我们希望利用Nvidia的最新技术，因此我们着手将私有云计算扩展到第三方云。"Apple通过强调其软件更加私密来区别于在AI上投入更多的公司——该公司不像OpenAI的ChatGPT或Anthropic的Claude等基于网络的AI那样收集大量数据——并且它正在利用其对本地存储的用户信息（如日历或短信）的访问来个性化AI功能。周一的技术讲座是为了让Federighi和他的副手们讨论Apple如何构建Siri AI及其Apple Intelligence层，以及它如何区别于消费者可能已经熟悉的AI。Apple高管概述了软件的架构，其中Apple的操作系统和软件有一个称为系统编排器的软件，它将任何AI查询路由到适当的模型——无论是在设备上还是在云中——取决于它需要多少计算能力和个人数据。Federighi说，系统编排器是"我们整个系统隐私架构的关键"。这次谈话还提供了关于Apple-Google合作伙伴关系的更多细节。Federighi表示，Apple Intelligence是内置于公司设备中的AI软件，使用Apple自己的模型，而不是公众可用的相同Google Gemini，正如科技行业许多人在1月两家公司宣布合作伙伴关系时所期望的那样。他还说它没有使用Google的现成云基础设施。Federighi表示，Google的技术被用来帮助构建Apple自己的模型——特别是周一宣布的用于云的"第三代"AFM模型，这些模型是为在Apple芯片上运行而设计和调整的。Subramanya说："我们刚才谈论的这四个模型——AFM Core、Core Advanced Cloud和Cloud Image——所有这些都是为Apple Silicon定制构建的，使用专有数据通过强化学习进行训练，并使用Gemini前沿模型的输出进行优化。

【参考链接】

https://www.cnbc.com/2026/06/08/apple-google-nvidia-ai-chips.html

────────────────────

10. 美国发布269页AI法案草案 🔥🔥🔥🔥

【摘要】

美国议员发布了269页的《伟大美国人工智能法案》讨论草案，这是2026年最重要的AI治理文件。

【核心事件】

美国国会推出全面AI监管法案，试图平衡创新与安全。

【关键细节】

法案长达269页
由共和党议员Jay Obernolte和民主党议员Lori Trahan联合提出
是2026年最重要的AI治理文件
试图平衡AI创新与安全
将深刻影响美国AI产业发展

【深度分析】

🌐 社会化: AI监管框架的建立将保护消费者权益

💼 经济化: 合规成本可能影响AI初创公司发展

🔬 世界化: 美国AI立法可能影响全球AI治理标准

📈 市场化: AI公司需要调整战略以适应监管要求

【原文内容】

The most consequential AI governance document of 2026 dropped late Thursday, June 4. Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) unveiled a 269-page discussion draft of the Great American Artificial Intelligence Act — the most comprehensive federal AI bill to date.

The bill represents a bipartisan effort to establish a regulatory framework for artificial intelligence in the United States. It addresses concerns ranging from AI safety and transparency to intellectual property rights and employment discrimination.

Key provisions include requirements for AI system transparency, mandatory safety testing for high-risk applications, and establishment of a federal AI oversight body. The bill also addresses issues of algorithmic accountability and requires companies to disclose when AI systems are used in decision-making processes that affect consumers.

The legislation comes at a critical time when AI companies are racing to develop and deploy increasingly powerful models while concerns about safety, bias, and societal impact continue to mount. The bill attempts to balance the need for innovation with the imperative to protect consumers and ensure responsible AI development.

Industry reaction has been mixed. Some companies welcome the clarity that regulation could provide, while others worry about compliance costs and potential restrictions on development. The bill's sponsors have indicated they are open to feedback and revisions as the discussion draft moves through the legislative process.

The 269-page document covers a wide range of topics including:

- AI system safety and testing requirements

- Transparency and disclosure obligations

- Intellectual property and copyright issues

- Employment and housing discrimination protections

- National security considerations

- International coordination on AI governance

- Establishment of regulatory sandboxes for innovation

- Requirements for high-risk AI applications

- Consumer protection measures

- Enforcement mechanisms and penalties

The bill represents the most serious attempt by Congress to grapple with the challenges posed by rapidly advancing AI technology. As AI systems become more capable and pervasive, the need for clear regulatory frameworks becomes increasingly urgent.

The discussion draft is just the beginning of what promises to be a lengthy legislative process. Stakeholders from industry, academia, civil society, and government will all have opportunities to provide input as the bill is refined and moved forward.

The outcome of this legislative effort will likely shape not only the U.S. AI landscape but could also influence AI governance approaches globally, as other countries look to the United States for leadership on this critical issue.

【中文翻译】

2026年最具影响力的AI治理文件于6月4日周四晚些时候发布。加利福尼亚州共和党议员Jay Obernolte和马萨诸塞州民主党议员Lori Trahan公布了269页的《伟大美国人工智能法案》讨论草案——这是迄今为止最全面的联邦AI法案。该法案代表了在美国建立人工智能监管框架的两党努力。它解决了从AI安全和透明度到知识产权和就业歧视等一系列问题。关键条款包括AI系统透明度要求、高风险应用的强制安全测试，以及建立联邦AI监督机构。该法案还解决了算法问责问题，并要求公司在决策过程中使用AI系统影响消费者时进行披露。该立法正处于关键时刻，AI公司正在竞相开发和部署越来越强大的模型，而对安全、偏见和社会影响的担忧持续增加。该法案试图在创新需求与保护消费者和确保负责任的AI开发的必要性之间取得平衡。行业反应不一。一些公司欢迎监管可能提供的明确性，而另一些公司则担心合规成本和对开发的潜在限制。该法案的发起人表示，他们愿意接受反馈和修订，因为讨论草案将通过立法程序推进。这份269页的文件涵盖了广泛的主题，包括：- AI系统安全和测试要求- 透明度和披露义务- 知识产权和版权问题- 就业和住房歧视保护- 国家安全考虑- AI治理的国际协调- 建立创新监管沙盒- 高风险AI应用的要求- 消费者保护措施- 执法机制和处罚该法案代表了国会认真应对快速发展的AI技术带来的挑战的最严肃尝试。随着AI系统变得更加强大和普遍，对明确监管框架的需求变得越来越紧迫。讨论草案只是承诺将成为漫长立法过程的开始。来自行业、学术界、公民社会和政府的利益相关者都将有机会在法案完善和推进过程中提供意见。这一立法努力的结果不仅可能塑造美国AI格局，还可能影响全球AI治理方法，因为其他国家在关键问题上寻求美国的领导。

【参考链接】

https://www.buildfastwithai.com/blogs/ai-news-today-june-6-2026

────────────────────

🎯 今日热点总结与趋势分析

核心主题

今天的AI领域呈现出三大趋势：地缘政治影响加深、资本市场狂热、技术竞争白热化。

趋势研判

·• 地缘政治化: AI模型开始受出口管制影响，技术中立性受到挑战，地缘政治因素深度介入AI产业

·• 资本狂热: Anthropic和OpenAI相继提交IPO，SpaceX收购Cursor，AI行业进入资本收割期

·• 竞争格局变化: ChatGPT市场份额跌破50%，Google Gemini和Anthropic Claude快速崛起

·• 社会影响显现: AI裁员潮达到两年高点，社会对AI的态度从热情转向理性

值得关注的信号

·• 美国269页AI法案草案显示监管正在追赶技术发展

·• Apple与Google、Nvidia合作显示AI开发从竞争走向合作

·• Anthropic因出口管制禁用模型，预示AI地缘政治风险上升

·• AI公司IPO潮即将开始，估值面临公开市场检验

────────────────────

END

编辑: 在哒转转

日期: 2026年6月22日

数据来源: TechCrunch, The Verge, CNBC, The AI Track, TechTimes等