当资本市场只谈EPS，我更关心：这波AI究竟会改变什么？
When Capital Markets Only Talk EPS, What I'm Really Asking Is: What Will This Wave of AI Actually Change?
核心观点 Core Thesis
估值模型、EPS 修正、算力缺口、GPU 交期——这些变量确实决定股价节奏。但如果我们只讨论哪些公司会涨，我们可能正在忽略一个更大的结构性问题：AI 技术爆发，本质上到底能改变什么？到2026年底，预计将有至少50家 AI 原生企业突破2.5亿美元年度经常性收入，部分企业正逼近10亿美元门槛。但这只是量的积累，质的跃迁在更深的层面。
Valuation models, EPS revisions, computing capacity gaps, GPU lead times — these variables genuinely determine the rhythm of stock prices. But if we only discuss which companies will rise, we may be overlooking a larger structural question: at its core, what will this wave of AI technology actually change? By end of 2026, at least 50 AI-native businesses are projected to reach $250 million in annual recurring revenue, with several approaching the $1 billion mark. But this is quantitative accumulation. The qualitative leap operates at a deeper level.
一、生物学为什么难：不是理论问题，而是变量规模问题 I. Why Biology Is Hard: Not a Theory Problem, but a Variable Scale Problem
Anthropic 执行长 Dario Amodei 提出过一个核心观点：生物学之所以困难，不是因为理论匮乏，而是变量数量呈指数级增长。蛋白质交互组合是天文数字，传统科研依赖人类团队线性推进，本质效率极低。AlphaFold 在蛋白质结构预测上的突破已经提供了第一波验证，但这只是开始。
Anthropic CEO Dario Amodei has articulated a core insight: biology is hard not because of a shortage of theory, but because the number of variables grows exponentially. Protein interaction combinations reach astronomical numbers, and conventional research relies on human teams advancing linearly. AlphaFold's breakthrough in protein structure prediction has provided the first wave of validation, but this is only the beginning.
如果大型模型可以在数据中心同时运行上亿个假设实验，把数十年的研究周期压缩到几年，医学研发路径就可能被重写。这不是空谈，这是方法论层面的根本性变革。
If large-scale models can simultaneously run hundreds of millions of hypothetical experiments in data centers, compressing decades of research into years, the path of medical development could be rewritten. This is not speculation — it is a fundamental methodological transformation.
二、这更像 2007 年的 iPhone 前夕，而非 2000 年的互联网泡沫 II. This Looks More Like the Eve of the 2007 iPhone Than the 2000 Internet Bubble
很多投资人在问：AI 会不会重演 2000 年的互联网泡沫？从产业节奏来看，更贴切的参照是 2007 年 iPhone 问世前夕。当年没有人能精准预测 App 经济的规模，但底层算力和网络架构已经准备就绪。
Many investors are asking whether AI will replay the 2000 internet bubble. A more accurate reference point is the eve of the 2007 iPhone launch — no one could predict the App economy's scale, but the underlying infrastructure was already in place.
2026年 AI 基础设施的现状：超大型云服务提供商合计资本支出预计突破6,000亿美元，同比增长40%。GPU 扩产、HBM 供给提升、数据中心电力升级、先进封装产能扩张——这些都是基础设施层面的铺垫。ABB 与英伟达的合作已经宣布，制造商使用该技术可将调试时间缩短80%、成本降低40%。基础设施到位，应用才会爆发。
The 2026 AI infrastructure reality: combined hyperscaler capex is projected to exceed $600 billion, up 40% year-over-year. GPU capacity expansion, HBM supply growth, data center power upgrades, advanced packaging capacity expansion — all laying the infrastructure groundwork. ABB and NVIDIA's partnership, announced in March 2026, claims manufacturers using the technology can cut setup times by up to 80% and reduce costs by up to 40%. Infrastructure in place precedes the application explosion.
三、效率工具，还是文明级跃迁 III. Efficiency Tool or Civilisation-Level Leap
真正值得思考的分水岭在于：如果 AI 只让广告投放更精准、客服成本更低，那它只是效率工具。如果它能压缩科研周期、改写研发方法、重构创新速度，那它是文明级跃迁。
The dividing line truly worth examining: if AI only makes advertising more precise and customer service cheaper, it is an efficiency tool. If it can compress research cycles, rewrite development methodologies and restructure the pace of innovation, it represents a civilisation-level leap.
四、核心结论 IV. Core Conclusion
对投资者而言，这个判断会直接影响应该在哪个层级的公司上分配资本，应该给予怎样的估值溢价，以及应该用怎样的持仓周期去承载这个判断。不是某一家公司的季报节奏，不是某一个季度的资本开支，而是这套技术最终停留在效率层，还是突破到方法论层——这才是真正的分水岭。
For investors, this judgment directly affects which layer of companies to allocate capital to, what valuation premium to assign, and what holding horizon is appropriate. Not the quarterly earnings rhythm of any single company, not the capital expenditure of any single quarter — but whether this technology ultimately remains at the efficiency layer or breaks through to the methodology layer. That is the true dividing line.
如果你对上述分析感兴趣，想了解这些宏观趋势如何具体落地到个别标的，财富俱乐部的深度研究报告与交易机会报告将提供更完整的分析框架——包括具体建仓价位区间、三个阶段止盈目标、关键催化剂时间节点、风险评估，以及触发重新评估持仓的具体条件。配合我们18套量化交易系统的实时买入卖出信号，在关键转折点提前一步捕捉抄底与逃顶机会，系统性提升每一笔交易的胜率。
订阅会员可在研究报告及交易机会栏目获取完整内容。
If you found this analysis useful and want to understand how these macro trends translate into specific investment opportunities, the Wealth Club's deep research reports and trading opportunity reports provide a complete analytical framework — including specific entry price ranges, three-stage profit targets, key catalyst timelines, risk assessment, and the precise conditions that would trigger a position reassessment. Combined with the real-time buy and sell signals from our 18 quantitative trading systems, you gain an early edge at every critical turning point — systematically improving your win rate on every trade.
Full content is available to subscribing members in the Research Reports and Trading Opportunities sections.
财富俱乐部 / Wealth Club
祝你投资顺遂，生活丰盛。
May your investments prosper and your life flourish.
