Macroeconomic experts at BCA Research are deeply divided over the trajectory of the stock market’s artificial intelligence trade. Analysts are currently debating whether the surge in AI-related equity valuations represents a sustainable shift in enterprise computing or a highly concentrated earnings bubble.
The macroeconomic catalyst driving this internal debate is the hundreds of billions of dollars Big Tech hyperscalers are currently pouring into data centers, specialized silicon, and energy infrastructure to secure computing dominance. The disagreement centers on whether this sustained capital expenditure will generate long-term cash flow or merely inflate immediate revenues for semiconductor manufacturers.
What We Know
- BCA Research chief economist Peter Berezin and Arthur Budaghyan classify the current U.S. stock market environment as an “earnings bubble” driven by artificial profit surges.
- Equities head Noah Weisberger and portfolio construction head Juan Correa argue the bull case, stating that hyperscalers are trading at their cheapest valuations in a decade relative to their improving returns.
- Hyperscaler cloud backlogs—representing enterprise orders received but not yet fulfilled—ballooned by roughly $750 billion over the past two quarters.
The bear case focuses on the mechanics of capital expenditure. Massive hardware purchases by companies like Meta and Alphabet register instantly as revenue for chipmakers. Berezin argues that these underlying profit surges are unsustainable, a position he outlined in a recent MarketWatch analysis detailing the ongoing BCA Research debate.
Berezin compared current AI profit expectations to the real estate and banking sectors just prior to the 2008 financial crisis. In both scenarios, surface-level price-to-earnings ratios appeared reasonable to investors, echoing previous historical warnings regarding profitability bubbles that fail to produce matching long-term cash flow.
On the opposing side, Weisberger and Correa maintain that AI compute remains physically scarce. They argue the physical demand for infrastructure justifies the aggressive capital outlays, even as semiconductor profit expectations draw increased scrutiny from institutional investors.
Hyperscaler Capex and the S&P 500 Profit Margin Limit
The core friction in the AI trade lies in historical valuation metrics. According to BCA Research data, if the S&P 500 were adjusted to reflect 2019 profit margins, the index would currently trade at 27 times forward earnings. This figure officially surpasses the dot-com bubble’s peak valuation of 26.5 times forward earnings recorded in March 2000.
The sustainability of the broader stock market rally now depends on a strict timeline. The market reality is no longer a question of whether artificial intelligence will reshape the economy, but rather how long hyperscaler spending can outpace market doubts before a potential reckoning. These companies must begin translating their $750 billion cloud backlogs into direct enterprise revenue before shareholder patience for hardware spending expires.