Dot-com, housing boom-bust pattern resurfaces in AI

A set of market signals that appeared before the dot-com and housing busts has returned during the recent surge in AI investment. Those signals include large valuation gaps, rapid speculative funding of early-stage AI firms, and concentrated retail and options trading in a small group of AI-related stocks.

The current wave began after public-facing releases of generative AI models in late 2022 and 2023. Venture capital firms have committed large sums to AI startups, with several companies securing rounds in the hundreds of millions or billions of dollars despite limited revenue. Public market gains have concentrated among a few chipmakers and cloud providers that supply computing power for AI models.

Measured indicators show stretched metrics in some corners of the market. Price-to-sales ratios for a subset of AI-focused companies sit above historical tech averages. Recent IPOs and special-purpose acquisition companies tied to AI listed at valuations that outpaced business fundamentals. Options volume and retail trading in select AI names have spiked, increasing short-term volatility. Banks and asset managers report higher demand for GPUs and cloud compute; supply constraints have pushed up hardware costs for firms building model capacity.

Corporations and investors have cited reasons for the funding surge. Many companies are adopting AI tools for tasks such as code generation, customer service automation and document review, and cloud providers have rolled out specialized AI services and hardware. Large funding rounds and strong share-price performance in a few public companies have attracted further capital into AI strategies in private and public markets.

A senior markets strategist at a large investment firm offered a note of caution: “We are seeing concentrated capital, high valuations without matching revenue and strong retail-driven momentum-components that historically have preceded sharp corrections,” cautioned the strategist.

Other industry participants point to differences with past episodes. Firms report pilot AI projects moving into production and some implementations producing measurable cost or time savings. Supporters emphasize that AI tools are being embedded into existing business processes rather than only creating new standalone consumer services.

Recent deal data show mixed signals. The number of AI-focused venture rounds and the size of late-stage deals have risen. At the same time, public-company revenue growth in AI-related firms has often been accompanied by higher research and infrastructure expenses, reducing near-term profitability. Lending for leveraged buyouts and margin borrowing linked to technology portfolios has also increased.

Regulators and institutional investors are monitoring concentration and systemic risk. Central bank researchers and financial-stability teams have reviewed potential impacts on margin requirements and risk-weighted assets. Several large investors have adjusted risk limits, expanded stress tests and rebalanced portfolios away from the most concentrated AI positions.

Market participants are watching specific indicators for stress: widening valuation gaps between revenue-generating and non-revenue firms, sharp drops in venture funding velocity, lower secondary private-market prices and spikes in margin borrowing and forced liquidations. Historical episodes in the late 1990s and mid-2000s are cited as reference points for those monitoring market developments.