A bubble in talks of an AI bubble?

If there’s ever been a ‘million dollar’ question, it’s the one posed by the title of this article, with this being ever-present in the market’s psyche since OpenAI launched Chat GPT on November 30, 2022.

AI has been presented as a technological breakthrough which in essence changes the way we interact with computers, breaking the barrier of translation previously needed via coding and user-friendly data aggregators or platforms, allowing us to command these machines, which are now an integral part of our lives, using plain, simple English.

The immediate winners were, in hindsight, obvious, as forward-looking expectations skyrocketed for ‘compute demand’ and chips, which are necessary in the training and development of large language models such as Chat GPT, or Anthropic’s Claude, or Google’s Gemini.

So too did expectations for the profits generated by semiconductor companies, most notably Nvidia, with stock prices following suit. In fact, Nvidia’s shares grew in value by about 13 times since OpenAI’s launch, while the broader iShares Semiconductor Index has more than quadrupled since then. As the saying goes, during a gold rush, sell shovels.

But is the equity market getting ahead of itself? And have expectations become too optimistic for a technology which, all in all, is still in its infancy, and which is still in its early stages of adoption, and is still to be stress-tested from a regulatory, privacy, and security standpoint?

The naysayers also express reasonable skepticism in relation to the promised AI-led gains in productivity, and the broader effect on labour market dynamics.

These topics would all deserve their own separate articles, but let us zoom in on recent developments and observe both sides of the coin.

Undoubtedly, the observed rally has been well-supported fundamentally. The S&P500 experienced considerable growth in both its top and bottom line, along with an improvement in profitability margins to well-above historical averages. In fact, for Q1 2026’s earnings season, S&P500 companies registered 27% year-over-year earnings growth on an aggregate basis.

Forward-looking guidance has been affirmed or improved, and while the S&P500 and its tech-heavier counterpart, the Nasdaq100, are at all-time highs, both are now cheaper from a valuation perspective, versus the end of 2025 or the end of 2024, and not at levels that could be deemed exuberant.

A feather in the cap of the bullish camp. As such, with fundamentals delivering, the comparison versus the multiple-led boom and bust of the early 2000s, the dot-com bubble, loses applicability.

Recent developments, along with the launch of a variety of agentic AI models specifically designed to tackle certain tasks or needs, further bolstered optimism. A material increase in the number of new businesses in the US seemingly reinforces the narrative of AI allowing less people to do more.

Jevons paradox further fans the flames for AI believers, with this stating that demand for a resource increases as the cost for the same resource decreases through gains in efficiency, promising continued demand for AI as adoption is scaled up.

So the initial signs have been positive, and the numbers, at a broad level, have been adding up, but are they sustainable?

The Magnificent Seven have considerably decreased buybacks, with free cash flows now allocated in major part to AI investments. These companies also recently sought financing through international capital markets via the issuance of debt.

While the funding runway appears lengthy when considering this group’s strength in generating cash from operations, along with how robust their balance sheets are, one can reasonably conclude that it would be difficult to maintain such levels of investment in perpetuity.

The bears, or those which are pessimistic, also question the return being generated on such capital expenditures.

As AI adoption increases, sceptics also question the lack of improvement in per unit economics, which is thus far not improving through economies of scale, as one would expect.

Infrastructure and energy constraints are also other stumbling blocks, along with a declining rate of marginal improvement in model performance in spite of ever-increasing computing power.

We are still in what feels like the early-innings, with any definitive answer to the questions posed above being near impossible to ascertain with confidence.

What is sure, however, is that the ‘AI boom vs AI bust’ debate is set to continually dictate investor sentiment and market performance.

The information presented in this commentary is solely provided for informational purposes and is not to be interpreted as investment advice, or to be used or considered as an offer or a solicitation to sell/buy or subscribe for any financial instruments, nor to constitute any advice or recommendation with respect to such financial instruments.

Curmi & Partners Ltd, with its registered address at Finance House, Princess Elizabeth Street, Ta’ Xbiex XBX 1102, is a member of the Malta Stock Exchange and is licensed by the MFSA to conduct investment services business.

Kieran Degiorgio is a senior portfolio manager and research analyst at Curmi & Partners Ltd.