Betting on Technological Progress – Long-Term Investment Thinking in the Age of AI
Artificial intelligence has become one of the strongest investment narratives of recent years. The hundreds of billions of dollars flowing into AI infrastructure, the rise of Nvidia, the rapid growth of OpenAI and Anthropic, and the data center investments of Google, Microsoft, Amazon, and Oracle have fundamentally reshaped the technology sector and, through it, the global equity markets.
However, this also raises an important question:
what happens if the extremely high expectations regarding AI are not met—or not met as quickly—as the market currently assumes?
A recent train of thought by investor Steve Eisman highlights precisely this risk. According to Eisman, the various players in the AI ecosystem have become interconnected to such an extent today that a problem emerging in certain segments of AI could easily ripple through the entire tech sector, and even beyond it.
At Duna Capital, we take this risk seriously. At the same time, we draw an important distinction between two questions:
predicting what will happen in the next 6–12 months is not the same as taking a stance on where the world is heading over the next 10–20 years.
The Three Dominoes of the AI Ecosystem
The first element of Eisman's argument is the vulnerability of the business model of LLM (Large Language Model) providers—primarily OpenAI and Anthropic.
Hyperscalers—such as Microsoft, Google, Amazon, or Oracle—are building massive data center and computing infrastructure. LLM providers, on the other hand, develop the models that run on this infrastructure.
One fundamental problem could be that the economic "moat" (durable competitive advantage) forming around LLMs is not necessarily as strong yet as many investors previously assumed.
Enterprise clients can switch between models relatively easily, while an increasing number of cheaper alternatives—including Chinese and open-source options—are emerging. This could lead to a price war, putting pressure on the profitability of LLM providers.
The second domino is that these companies do not operate in isolation.
A significant portion of the future contract backlog announced by hyperscalers is linked to the AI ecosystem, specifically to LLM providers. For example, in the second quarter of its 2026 financial year, Microsoft reported a commercial RPO (Remaining Performance Obligation) of $625 billion, approximately 45% of which was tied to OpenAI. This clearly illustrates that companies developing AI models and tech giants providing the infrastructure are financially deeply interconnected. If the future computing capacity demand of LLM providers were to fall significantly short of expectations, the consequences could also manifest at the hyperscalers.
The third domino is the concentration of the investment market.
The world's largest technology companies hold a dominant weighting in leading equity indices and the ETFs that track them. According to Eisman's argument, this means that many investors who otherwise consider their portfolios diversified actually have significant AI and tech exposure.
A potential AI correction, therefore, would not necessarily remain confined to the ecosystem of OpenAI, Anthropic, or Nvidia. It could impact hyperscalers, tech indices, ETFs, and, through them, global capital markets.
But Are We in a Bubble?
Currently, we do not have a reliable answer to this.
We do not know whether tech stocks will trade higher or lower in six months or two years. We also do not know for certain whether current valuation levels will, in hindsight, prove to be undervalued, fairly priced, or the peak of a speculative bubble.
And in our view, as long-term investors, this is not necessarily the question we need to answer perfectly.
There is, however, another question in which we have significantly greater conviction: do we believe that humanity will continue to advance technologically in the coming decades?
At Duna Capital, our answer to this is an unequivocal yes.
Betting on Technological Progress, Not a Single AI Company
We are convinced that we are on the verge of extraordinary technological advancement in the coming decades.
We do not know whether the ultimate winner of this will be OpenAI, Anthropic, Google, Microsoft, or a company currently barely known. Furthermore, it is highly conceivable that some of the defining technology companies of the next twenty years do not even exist today.
This is one of the reasons why we find index-based investing highly compelling for the long term.
A prime example of this is the Nasdaq-100 index and the investment products tracking it—such as XNAS (ISIN: IE00BMFKG444)—through which an investor does not stake everything on the success of a single company, but rather can share in the performance of a broader universe of innovative and technological companies.
It is an important clarification that the Nasdaq-100 is not simply an index of "100 tech companies": it comprises the 100 largest non-financial companies listed on the Nasdaq stock exchange, yet, due to its composition, it provides strong technology and innovation exposure.
One of the most important features of this structure is that the index's composition changes over time. Former winners may drop out or receive a lower weighting, while new, fast-growing companies are included.
Therefore, we do not have to pick the single winner of the technological race thirty years in advance.
In a sense, the index continuously adapts to which companies become the dominant players in the market.
What Does History Show?
The Nasdaq-100 was launched in 1985.
According to the summary prepared by Nasdaq for its 40th anniversary, the index achieved a compound annual growth rate (CAGR) of 14.25% between 1985 and the end of 2024. By comparison, the annualized return of the S&P 500 during the same period was 11.57%.
Naturally, this does not mean that similar returns can be expected over the next forty years.
In fact, the road there included explicitly painful periods. According to Nasdaq's own historical summary, when the dot-com bubble burst, the Nasdaq-100 plummeted by roughly 83%.
This is an important lesson.
A highly successful long-term technological trend and a short-term bubble are not mutually exclusive. It is possible to simultaneously believe that a technology will fundamentally change the world and that, at a given moment, certain companies associated with it are overvalued. There is no contradiction between the two.
Our Biggest Problem: We Cannot Time the Market
If we accept that we believe in technological progress in the long term, but in the short term we cannot determine whether the market is 20% overvalued or 20% undervalued, the following question arises:
how should we invest?
One possible answer is systematic investing at predetermined intervals, known as Dollar Cost Averaging (DCA).
The essence of this is that we do not try to perfectly catch the bottom or the top of the market. Instead, we buy regularly. When prices are high, we buy fewer mutual fund units or shares with the same amount of money. When the market corrects, we buy more with the same amount.
DCA, of course, does not eliminate the risk of investment loss and does not guarantee a positive return. It can, however, help make the investment decision less dependent on whether one manages to time the market correctly at a single specific moment.
What Happens if the AI Correction Comes Tomorrow?
It might. We do not know.
It is possible that a portion of the capital currently invested in AI infrastructure will not generate adequate returns. It is conceivable that a severe price war will develop among LLM providers. It is possible that certain companies are trading at excessively high valuations today.
And it is also possible that after a significant correction, it will take years for certain share prices to reach their previous peaks again.
This is why Eisman considers monitoring the financial position of OpenAI and Anthropic to be particularly important. Should these companies go public, their public financial statements will provide a much clearer picture of how sustainable one of the most critical financing links in the AI ecosystem truly is.
All of this, however, does not change our core long-term premise.
Duna Capital's Investment Thesis
We do not want to bet on knowing exactly which AI model will be the best in five years. We do not want to predict which tech company will be the biggest winner of the next thirty years. And we do not believe that we can consistently call when the next 20–30% market correction will occur.
However, we are willing to bet on the long term that humanity's technological advancement will continue.
Artificial intelligence, robotics, automation, semiconductors, cloud computing, biotechnology, autonomous systems, and technologies we cannot even foresee today may emerge.
A long-term, systematic investment in a broader tech index, therefore, is not merely about "buying AI stocks" for us. Rather, it is about buying a stake in humanity's innovation and technological progress.
We cannot predict next quarter's share prices. Nor the timing of the next correction. We do believe, however, that the world will be more technologically advanced twenty years from now than it is today. As long-term investors, this is the thesis that truly matters to us.
Legal Disclaimer
This article is for general informational and educational purposes only and does not constitute investment advice, an investment recommendation, financial analysis, an offer, or a solicitation to buy or sell any financial instrument.
The opinions and findings contained in this article do not take into account the unique financial situation, investment objectives, risk tolerance, or time horizon of individual investors. Past performance is no guarantee of future results. The value of equity and ETF investments can fluctuate significantly, and investors may even suffer losses.
Before making any investment decision, it is always recommended to thoroughly review the documentation of the given financial instrument, consider the associated risks, and, if necessary, seek the advice of a duly authorized professional.