When AI Borrows from Wall Street: Tech Giants' CapEx Cycle Accelerates Financialization
- The article says AI infrastructure spending is becoming so large that even cash-rich tech giants such as Alphabet, Amazon, and Meta are increasingly turning to bond markets instead of relying only on internal cash flow.
- Morgan Stanley estimates global AI-related debt issuance could reach about $570 billion in 2026, more than double last year, after already reaching about $236 billion by the end of May.
- Alphabet, Amazon, Microsoft, and Meta are expected to spend about $700 billion on related capital expenditure this year, and hyperscaler capex could exceed $1 trillion by 2027.
- Alphabet, Amazon, and Meta have all issued large bond deals this year while raising capex guidance, and their free cash flow has weakened as AI spending becomes more front-loaded.
- Oracle is presented as a more stressed case, with fiscal 2026 capex of about $55.66 billion versus operating cash flow of about $32 billion, leaving free cash flow at negative $23.69 billion.
The article argues that the AI investment cycle is entering a more financialized phase. In the earlier software era, large tech companies could fund growth largely from operating cash. In the current AI buildout, however, GPUs, servers, networking, data centers, cooling systems, land, and related facilities require heavy spending long before revenue arrives.
Alphabet is used as the clearest example. Its latest quarter still showed strong growth, but capital expenditure rose sharply and quarterly free cash flow turned negative. The company also lifted its 2026 capex guidance. The article makes the broader point that AI has made tech companies more asset-heavy and less like the cash-generating software businesses of the past.
Amazon and Meta are described as following the same pattern, issuing large bond deals this year while increasing capex plans and seeing free cash flow compress. The article says borrowing itself is not necessarily a warning sign for firms with large businesses and strong credit, but it becomes more important to watch when capex persistently exceeds a company’s own cash generation.
For stocks, the likely impact is mixed: generally supportive for AI infrastructure spending, chip supply, and capital-markets activity, but potentially a drag on valuation for companies where rising debt and weaker free cash flow become a concern. The impact on crypto, gold, and forex is indirect and mostly neutral in the article’s framing.
当AI向华尔街借钱：科技巨头的CapEx周期加速金融化
文章认为，AI投资周期正在进入更强的金融化阶段。过去的软件时代，大型科技公司主要依靠经营现金流就能支撑增长；而在当前的AI建设中，GPU、服务器、网络、数据中心、冷却系统、土地和配套设施都需要在收入兑现之前先行投入大量资金。
文中以Alphabet为例说明这一变化。该公司最新季度仍保持强劲增长，但资本开支大幅上升，季度自由现金流转为负值，公司还上调了2026年资本开支指引。文章借此指出，AI正在把科技公司重新变成更重资产的企业，而不再像过去的软件公司那样偏向高现金流模式。
Amazon和Meta也被描述为采取了类似做法：它们今年发行了大额债券，同时提高资本开支计划，自由现金流则明显收缩。文章强调，借债本身并不一定是风险信号，对业务规模大、信用评级高的公司来说，长期债务可以用来匹配前置的建设支出；但如果资本开支持续超过自身现金创造能力，就需要更谨慎地看待。
对股票市场而言，这一变化的潜在影响偏中性到分化：一方面有利于AI基础设施、芯片和资本市场融资活动，另一方面也可能对那些债务上升、自由现金流走弱的科技股估值形成压力。对加密货币、黄金和外汇市场，文章没有给出直接驱动，整体更偏中性。