As 2025 draws to a close, the debate over whether artificial intelligence represents a market bubble continues to divide investors and analysts. John Belton, a portfolio manager at the $35 billion Gabelli Funds, offers a nuanced perspective by distinguishing between two types of bubbles: earnings-driven and valuation-driven.
Belton draws parallels to the dot-com era, noting that while top tech stocks today trade at elevated prices relative to profits, they haven’t reached bubble territory. He points out that the median forward price-to-earnings ratio for the “Magnificent 7” (the seven largest tech companies) stands at approximately 25x today, compared to nearly 90x at the end of 1999. This significant difference suggests that current valuations reflect strong fundamentals rather than irrational exuberance.
The portfolio manager identifies two key reasons why AI hasn’t entered an earnings bubble. First, despite high capital expenditure spending on AI infrastructure, the majority of these investments are being deployed to strengthen already-profitable business operations rather than speculative ventures. Second, the ecosystem of AI use cases continues to expand beyond initial applications, with promising developments in autonomous driving, robotics, life sciences, scientific discovery, and agentic software.
However, recent market volatility has raised concerns. Oracle and Broadcom have both experienced declines as investors scrutinize the returns on massive AI investments and question whether capex spending justifies current valuations. Despite this turbulence, Belton maintains confidence in both companies’ ability to weather the storm, specifically highlighting Broadcom as among the top chip stocks to own in the current environment.
Belton acknowledges that investor caution regarding massive spending from companies like OpenAI is warranted. He emphasizes that AI infrastructure development will inevitably follow a cyclical pattern, stating that the question isn’t whether the market will reach a peak, but rather when that peak will occur and how high it will climb. While maintaining a bullish outlook on AI’s near-term strength, the portfolio manager advocates for measured optimism tempered with realistic expectations about the infrastructure cycle’s natural progression.
Key Quotes
It is very difficult to argue we are in a valuation bubble currently. As an example, median ‘Mag 7’ forward P/E was close to 90x at YE99 vs. ~25x today. Valuations seem to simply reflect strong fundamentals without obvious excess. Are we in an earnings bubble? Time will tell.
John Belton, portfolio manager at Gabelli Funds, provides quantitative evidence comparing today’s AI market to the dot-com bubble, suggesting current valuations are far more reasonable than the late 1990s tech bubble.
Second biggest reason to feel confident we are not in an earnings bubble: additional large-scale use cases are beginning to present themselves. Autonomous driving, robotics, life sciences, other scientific discovery, agentic software. We do need at least some of these use cases to become commercial — early, but there is promise.
Belton outlines the expanding ecosystem of AI applications that could justify current investment levels, emphasizing that commercial viability of these use cases will be crucial for sustaining the AI market’s growth trajectory.
Like any infrastructure cycle this too will be a cycle (question is not if we reach a peak, but when and how high the peak is).
The portfolio manager acknowledges the inevitable cyclical nature of AI infrastructure investment, providing a realistic framework for investors to understand that while growth is expected, peaks and corrections are natural market dynamics.
Our Take
Belton’s framework distinguishing between valuation and earnings bubbles is intellectually honest and practically useful for investors navigating AI markets. The 25x versus 90x P/E comparison is compelling evidence that we’re not in dot-com 2.0, at least not yet. However, the real test lies in whether AI capex translates into revenue growth and profitability at scale. The recent volatility in Oracle and Broadcom stocks signals that markets are beginning to demand proof of returns, not just promises. The expansion into autonomous driving, robotics, and life sciences represents AI’s maturation from a narrow technology into a general-purpose platform, similar to how electricity or the internet transformed industries. If even a fraction of these use cases achieve commercial scale, current valuations may prove conservative. The key risk remains timing—infrastructure cycles can overshoot dramatically before correcting, and investors who enter late in the cycle often suffer the most severe losses.
Why This Matters
This analysis matters because it provides institutional investors’ perspective on one of the most critical questions facing technology markets today. The distinction between valuation and earnings bubbles offers a framework for understanding AI investment risks more precisely than binary bubble/no-bubble debates. With trillions of dollars flowing into AI infrastructure and development, understanding whether current valuations are sustainable has profound implications for tech company stock prices, venture capital allocation, and the pace of AI innovation.
The comparison to the dot-com era is particularly significant, as that period saw massive wealth destruction when the bubble burst. However, Belton’s data showing dramatically lower P/E ratios today suggests the market may be on firmer footing. The expansion of AI use cases beyond initial applications—from autonomous vehicles to drug discovery—indicates the technology is finding real-world commercial applications, which supports the thesis that current investments may generate genuine returns rather than speculative losses. For businesses planning AI strategies and investors allocating capital, this nuanced view suggests continued opportunity while acknowledging cyclical risks ahead.
Related Stories
- AI Lessons from the Dot-Com Era: Steve Case’s Perspective on the AI Revolution
- Meta and Nvidia Billionaires’ Wealth Soars $152B in AI Boom
- Global Funds Turn to Indian Stocks as Hedge Against AI Risks
- CEOs Express Insecurity About AI Strategy and Implementation
Source: https://www.businessinsider.com/ai-bubble-valuations-earnings-tech-magnificent-7-avgo-orcl-2025-12