AI Boom Explained

Generative AI has triggered one of the largest corporate investment cycles in history and reshaped which companies drive the market — whether it's a durable productivity shift or a speculative narrative is still an open question.

From a chatbot to a capital cycle

OpenAI's release of ChatGPT in November 2022 turned generative AI from a research curiosity into a mainstream business priority almost overnight. Within a year, every major technology company had committed to building or buying the infrastructure needed to train and run large AI models: specialized chips, data centers, and the power capacity to run them. Nvidia, whose graphics processors turned out to be the preferred hardware for training these models, became one of the most valuable companies in the world as demand for its chips outstripped supply.

A capex boom in chips and cloud

The hyperscalers — Microsoft, Amazon, Alphabet, and Meta — have poured tens of billions of dollars a quarter into data centers and AI infrastructure, betting that AI capabilities will become a core input to their businesses and that being under-provisioned is the bigger risk. That spending flows directly into semiconductor makers, cloud infrastructure providers, and a widening circle of AI-adjacent companies, from power utilities serving data centers to firms building the networking equipment that ties it all together.

A reshaped market

This buildout has concentrated market leadership in a small number of companies. A handful of AI-linked megacaps now account for an outsized share of the S&P 500's total market capitalization and, in many recent years, an outsized share of its gains. That concentration means the broad market's performance has become unusually tied to the fortunes of a few firms and to the continued willingness of investors to pay premium valuations for AI-linked growth.

Productivity shift or speculative narrative?

The bull case treats this as comparable to earlier infrastructure buildouts — electricity, railroads, the internet itself — where heavy upfront investment eventually unlocked broad productivity gains, and points to the fact that today's leading AI companies are, unlike most dot-com-era darlings, highly profitable with real revenue growth to show for the spending.

The bear case draws the opposite historical parallel: past infrastructure booms, including the dot-com era's fiber-optic buildout, often overbuilt capacity ahead of demand, and some observers worry about circular financing arrangements between chipmakers, cloud providers, and AI startups, alongside genuine uncertainty about how quickly AI spending translates into durable profit for the companies buying the infrastructure rather than just the ones selling it. Neither case is resolved yet, and that's the honest state of the debate.

How to think about it going forward

Whatever the eventual verdict, the near-term test each quarter is straightforward: do earnings and guidance from the companies at the center of this cycle keep validating the spending, or does the story start outrunning the numbers. That's the same discipline that separated dot-com survivors from the companies that vanished — watching what the results say, not just what the narrative promises.

It's also worth watching positioning alongside fundamentals. Even a genuinely transformative technology can produce a painful drawdown if expectations run ahead of what any near-term earnings report can satisfy — that gap between a good story and a good print is where most of the historical volatility around new-technology cycles has come from, AI included.