8:01 AM Krypton | Apple Launches New Siri AI; ROKID Responds to 'Smart Glasses Secretly Filming Stewardess' Incident; OpenAI Secretly Submits IPO Documents 8点1氪丨苹果推出全新Siri AI；ROKID回应“智能眼镜偷拍空姐”事件；OpenAI秘密提交IPO文件

This anxiety and urgency are not unique to Apple. Look at how Alibaba swiftly established its Token Foundry Division, merging the Tongyi large model with the Future Life Lab, with CEO Wu Yongming personally leading the charge. This is clearly a signal to the market and internally: AI is no longer an "experimental business" but must rapidly generate returns and integrate into the core arteries of e-commerce and cloud computing. Alibaba's financial report, for the first time, disclosed the transition from an "investment phase" to a "commercialization and return phase," which is less a technological victory and more a triumph of organizational and strategic focus. The pragmatism and aggressiveness of domestic tech giants in AI implementation form an interesting contrast with Apple's plight of seeking external aid for core technology. One side is eager to prove AI can make money, while the other is eager to prove AI can be used effectively.

Zooming out, news of NVIDIA and LG co-building an AI factory paints another picture: the AI battlefield is solidly moving from cloud algorithms to the physical world. Robotics, autonomous driving, and data centers—heavy-asset, long-cycle domains—are precisely the critical leap for AI to move from "chatting" and "content generation" to "perception, decision-making, and execution." Jensen Huang's compute empire is deeply intertwining with global manufacturing giants, aiming to define the next generation of industrial infrastructure. This touches the roots of industrial transformation far more profoundly than simply competing on model parameters or flashy applications.

However, technology's relentless advance always leaves shadows in the corners. The incident of secret filming with ROKID smart glasses is like a bucket of cold water poured over wearable AI. As hardware becomes so invisible and powerful, traditional boundaries of privacy protection instantly blur. The emergence of the "indicator light cover sticker" black industry chain nakedly reveals the race between human nature and technological oversight. While manufacturers' "post-incident handling" statements are necessary, they can never fully repair the cracks in trust. Every leap in AI hardware must simultaneously build more robust ethical and legal guardrails; otherwise, the so-called "smart life" might first become a "life of constant apprehension."

WeChat opening its AI ecosystem capabilities to developers is another quiet "infrastructure expansion." WeChat's vast ecosystem has long been a model of a "super app," and now it aims to offer AI capabilities as ubiquitously as utilities. Although still in beta, this move sends a clear message: in the AI era, platforms must not only use AI themselves but also make all participants in the ecosystem dependent on your AI. This is a deeper form of binding and dominance. As developers need to "actively authorize access" to WeChat's AI, a new, platform-defined AI development paradigm is taking shape.

Meanwhile, OpenAI has secretly submitted IPO documents, with its valuation surging toward a trillion dollars. This marks the official entry of AI fervor into the capital harvesting phase. AI companies are no longer content to merely depict the future; they urgently need real money from secondary markets to sustain this expensive race. This will bring more capital but may also foster more short-sighted commercial goals, posing a test to long-term technological commitment.

Returning to Apple, the Cook era has officially come to an end. His successor will not only inherit the legacy of chips and design but also face an Apple that needs to prove itself anew in the AI race. The improvements to Siri AI, the speed enhancements in macOS, the compromises in liquid glass design—these details piece together the image of a giant striving to maintain elegance and stability amid transformation. But the real challenge lies here: as AI becomes the new operating system kernel, will Apple leverage its integrated hardware-software ecosystem advantage to catch up and surpass, or will its obsession with "user experience" and "privacy" continually leave it constrained by others in terms of AI's openness and native capabilities? Whether this partnership represents Apple's far-sighted strategy or a forced compromise, time will provide the sharpest answer.

苹果在WWDC上宣布与谷歌Gemini合作，这可能是近年来硅谷最戏剧性的一幕。一家以构建封闭花园、自研芯片和操作系统为傲的公司，如今在其最核心的AI战略上，选择了拥抱昔日的对手。这背后，不是简单的“技术互补”，而是一场关于速度、恐惧与商业现实的生存博弈。苹果的AI，显然在过去一两年被OpenAI、谷歌甚至国内厂商的突飞猛进衬托得步履蹒跚。那个曾经定义“智能手机”和“应用生态”的苹果，在生成式AI的浪潮中，正经历一场艰难的“追赶者”心态转变。Siri AI的独立App和跨设备同步功能是好的开始，但核心引擎的“借壳”，终究是权宜之计。这暴露了一个事实：在AI大模型的军备竞赛中，即便是苹果，也难以承受从零研发、慢慢打磨的时间成本。用户的期待窗口期，远比想象中短。

这种焦虑和紧迫感，并非苹果独有。你看，阿里巴巴迅速成立Token Foundry事业部，将通义大模型和未来生活实验室合并，CEO吴泳铭亲自挂帅。这分明是在向市场和内部宣告：AI不再是一个“实验性业务”，而是必须快速造血、融入核心电商与云计算动脉的主引擎。从“投入期”到“商业化回报期”，阿里的财报首次披露了这一转折，这与其说是技术胜利，不如说是组织与战略聚焦的胜利。国内大厂在AI落地上的务实和凶猛，与苹果在核心技术上寻求外援的窘迫，形成了有趣的对照。一边是急于证明AI能赚钱，一边是急于证明AI能用好。

而将视野拉远，英伟达与LG合建AI工厂的消息，则描绘了另一幅图景：AI的战场正从云端算法，坚实地迈向物理世界。机器人、自动驾驶、数据中心，这些重资产、长周期的领域，恰恰是AI从“聊天”和“生成内容”走向“感知、决策与执行”的关键一跃。黄仁勋的算力帝国，正在与全球制造业巨头深度捆绑，试图定义下一代工业基础设施。这比单纯比拼模型参数或应用花哨，更触及产业变革的根基。

然而，技术的狂飙突进，总在角落留下阴影。ROKID智能眼镜的偷拍事件，像一盆冷水浇在可穿戴AI的头上。当硬件变得如此隐形和强大，传统的隐私保护边界瞬间模糊。“指示灯遮光贴”这种黑色产业链的出现，赤裸裸地揭示了人性与技术监管的赛跑。厂商的“事后处理”声明，虽然必要，但永远无法完全挽回信任的裂痕。每一次AI硬件的跃进，都必须同步构筑更坚固的伦理和法律护栏，否则，所谓的“智能生活”可能先一步沦为“提心吊胆的生活”。

微信开放AI生态能力给开发者，则是另一场静默的“基建扩张”。微信的庞大生态，一直是“超级App”的典范，如今它要将AI能力像水电一样开放出去。虽然尚在内测，但这步棋的意味非常明确：在AI时代，平台不仅要自己用AI，更要让生态内的所有参与者都离不开你的AI。这是一种更深层次的绑定和统治力。当开发者需要“主动授权接入”微信AI时，一个新的、由平台定义的AI开发范式正在形成。

与此同时，OpenAI秘密提交IPO文件，估值冲向万亿美元。这标志着AI狂热正式进入资本化的收割阶段。AI公司们不再满足于描绘未来，它们急切地需要二级市场的真金白银来延续这场昂贵的竞赛。这会带来更多的资金，也可能催生更急功近利的商业目标，对技术长期主义是一场考验。

回到苹果，库克时代正式落幕。接任者将面临的，不仅是芯片和设计的传承，更是一个在AI赛道上需要重新证明自己的苹果。Siri AI的改进，macOS的速度提升，液态玻璃设计的妥协，这些细节堆砌起的是一个试图在变革中保持优雅与稳定的巨头形象。但真正的挑战在于，当AI成为新的操作系统内核，苹果是能凭借其软硬一体化的生态优势后来居上，还是会因为对“用户体验”和“隐私”的执念，在AI的开放与原生能力上持续受制于人？这场合作，究竟是苹果的深谋远虑，还是迫不得已的妥协，时间会给出最尖锐的答案。