The conversation around whether artificial intelligence stocks are caught in a bubble just got more interesting.
Global banking powerhouse Citi has taken a closer look at the numbers and concluded that, while AI valuations are undeniably stretched, they haven’t yet crossed into dangerous territory.
In a recent internal review of tech-sector pricing, Citi analysts said most AI-related companies still trade within a “rational range” supported by strong profit growth and demand for digital infrastructure. Only a handful of areas, particularly those tied to capital-intensive industries, are showing early warning signs of excess.
“The market’s enthusiasm for AI remains intense, but it’s not irrational,” Citi noted in its latest valuation report. The bank’s analysts observed that the current rally, though rapid, continues to align with improving earnings, unlike the speculative surges seen during past technology manias.
Cracks Forming Beneath the Surface
Still, Citi cautioned that parts of the ecosystem — especially hardware and infrastructure suppliers — could be reaching unsustainable levels. These segments, heavily reliant on physical assets and massive capital expenditures, have posted some of the biggest price gains in recent months. The firm recommended investors consider locking in profits from these plays while maintaining exposure to diversified AI portfolios.
Citi reiterated its preference for a “growth at a reasonable price” strategy, highlighting companies that balance innovation with consistent cash generation. Analysts also underscored the growing divide between asset-light models, such as software and data platforms, and asset-heavy ones like chip foundries and data center developers — the latter being more vulnerable to cyclical slowdowns.
The study comes amid growing comparisons between today’s AI surge and the late-1990s internet boom. Market watchers have pointed out that the ten largest U.S. companies — many of which are riding the AI wave — now account for nearly 40% of the S&P 500’s total market value, compared with about 25% during the dot-com peak.
Meanwhile, traditional valuation metrics such as the price-to-book ratio have surpassed their 2000 highs, reigniting concerns that investor behavior is again drifting into speculative territory. Capital continues to flow into AI-linked stocks regardless of short-term profitability, echoing the sentiment-driven rush of two decades ago.
A More Mature Market
But this cycle may not end the same way. Unlike many dot-com startups, the current generation of AI leaders — from chipmaker Nvidia to major cloud service providers — are profitable, cash-rich businesses funding their expansion without excessive leverage. Their products are already embedded in global supply chains and productivity systems, suggesting a stronger foundation than the internet stocks of the past.
Citi’s analysis ultimately strikes a middle ground: while investor excitement around AI has created hot spots of overvaluation, the overall sector still appears grounded in fundamentals. The bank believes the real risk lies in selective overheating, not a full-blown bubble — at least not yet.
