· 方法论

认为 AI 已引发不可逆的岗位替代,投资相关基础设施和瓶颈是对冲失业冲击的主要方式。

涉及标的:

中文翻译

这个消息真的很令人心碎: $META 裁员 20% $ORCL 裁员 $AMZN 长期将有 60 万名工人被机器人和 AI 取代。 这是一种反乌托邦未来。 企业在没有人工劳动成本的情况下,仍然能赚创纪录的利润。 唯一能从中受益的办法: 把 AI 当成对冲工具去投资。 未来几年,想逃离 AI 造成的永久底层阶级,感觉最主要的办法就是这样。 AI 产生的股东权益回报,将流向股东。 而那些没投资股票、只能靠工资过活的人之间的差距会继续拉大。 这不是未来,而是已经在发生。 - Opus 4.6 已经足够替代今天大多数软件工程师。 - Waymo 已经开始在旧金山这样的地方替代出租车司机。 - 我们知道 $TSLA 的人形机器人也快来了,因为它们在中国已经普及。 这件事现在就在发生。 伊朗的扰动只是 AI 建设加速过程中的短期事件。 AI 已经到达拐点,而且似乎不可避免。 你已经看到美国就业修正接近减少 100 万,这非常惊人。 而且我们正在看到,新一代 LLM 正在由它们前一代模型自己构建,AI 正在逼近奇点(AI 递归增长)。 如果你要投资那些运行 AI 所需的算力和硬件: 数据中心 / 电力 / 电网板块: $NBIS、$XLU、$VRT、$BE 光子板块,负责扩展 AI: $LITE、$COHR、$AAOI、$TSEM 半导体板块,负责芯片: $NVDA、$TSM、$ASML、$INTC 存储板块,负责芯片: $MU、$SNDK、SK 海力士、三星 ASIC,负责超大规模云 AI 推理: $AVGO、$MRVL、Mediatek 良率 / 测试板块,确保芯片能工作: $TER、$AEHR、Advantest 以及 AI 所需的原材料或基板: $AXTI、$COPX、$SOI 还有很多其他公司,都在成为对抗大规模 AI 失业的单一、最大对冲工具。 谁掌握了算力的生产资料(瓶颈、材料、数据中心): 谁就掌握了 AI 的未来。

英文原文

The news is pretty heartbreaking: $META 20% layoffs $ORCL layoffs $AMZN 600,000 workers long term layoffs as they get replaced by robotics and AI. This is a dystopian future. Companies end up with record profits, without the cost of human labor. The only way to benefit: Investing in AI as a hedge. The next few years feels like the main way to escape the permanent underclass, caused by AI displacement. The return on equity derived from AI will go to the shareholders. While the gap between those who live paycheck to paycheck, not invested in stocks. Will continue to grow. This is not the future. - Opus 4.6 is good enough to replace most software engineers today. - Waymo has started to replace taxi drivers in places like SF today. - We know $TSLA Humanoids are coming next as they’re widespread in China, today. This is happening now. Disruptions in Iran are only temporary to the accelerating AI buildout. AI has hit the inflection point, and looks inevitable. You’re already seeing US job revisions down close to 1 Million, which is staggering. And we’re seeing the newest LLMs be built by their previous models, as AI approaches the singularity (AI led recursive growth). Investing in where the compute and hardware needed to run the AI: From the datacenter/power/grid sector: $NBIS, $XLU, $VRT, $BE Photonics sector needed to scale AI: $LITE, $COHR, $AAOI, $TSEM Semi sector needed for the chips: $NVDA, $TSM, $ASML, $INTC Memory sector for the chips: $MU, $SNDK, SK Hynix, Samsung ASICs for hyperscaler AI inference: $AVGO, $MRVL, Mediatek Yields sector to make sure the chips work: $TER, $AEHR, Advantest Along with the raw materials or substrates needed for AI: $AXTI, $COPX, $SOI And many others become the single, largest, hedge against widespread AI displacement. Whoever owns the means of compute (bottlenecks, materials, datacenters): Owns the future of AI.

在 X 上查看原推 ↗