AI will change the world. That much feels certain.
What is far less certain is whether it will reward investors in the way many expect.
History offers a useful reminder.
In 1999, Cisco Systems was the most valuable company in the world. It built the infrastructure of the internet—routers and switches that made everything possible. The thesis was simple and compelling: the internet would reshape the world, and Cisco would be at the centre of it.
The logic was sound. The outcome was not.
The stock fell 86% the following year and took decades to recover to its peak.
Today, we are seeing echoes of that same conviction—this time in artificial intelligence.
Some of the most profitable companies ever created are now spending heavily just to remain competitive in AI. In many cases, they are investing more than they earn. That is a remarkable shift—and not an encouraging one.
For decades, software represented the ideal business model. Build once, sell endlessly. Minimal marginal cost. No factories, no inventory, no physical constraints. This model justified premium valuations and defined the gap between technology leaders and traditional industries.
That model is now changing.
Scaling AI requires vast physical infrastructure. Data centres the size of small towns. Continuous cooling systems. Energy demands that stretch national grids. These are no longer purely software businesses. They are capital-intensive with a software layer on top.
In simple terms, the factory has returned.
What makes this more complex is how these investments are being financed.
Many companies prefer not to carry the full cost on their balance sheets. Instead, they rely on structures such as special-purpose vehicles (SPVs), private credit arrangements, and long-term lease agreements. These approaches shift obligations out of immediate view, but they do not remove the underlying risk.
An estimated $120 billion in data centre spending has already been moved off balance sheets in this way.
The risk remains. It is simply less visible.
There is also a critical constraint that receives far less attention: time.
The advanced chips driving AI innovation have a useful life of roughly four to five years. Yet the infrastructure required to deploy them—power grids, planning approvals and construction—can take much longer to develop.
This creates a mismatch. Some chips may become obsolete before they are fully utilised.
Even industry leaders acknowledge the challenge. The constraint is no longer computing power alone—it is the ability to build, connect, and sustain the infrastructure quickly enough. At the same time, AI must demonstrate real, measurable value to justify its enormous energy consumption.
If it does not, the industry may face limits not just from economics, but also from broader societal expectations.
A significant hurdle for new US data centres is the lack of necessary electrical equipment—specifically transformers, switchgear, and batteries. This shortage is the primary factor leading to the anticipated delay or cancellation of nearly fifty percent of the data centre projects scheduled for this year.
These are not theoretical risks. They are practical, industrial constraints.
None of this diminishes the significance of AI. Transformational technologies have always reshaped the world. Railroads, electrification, and the internet all delivered lasting change.
But they also followed a familiar pattern.
They created enormous value—while simultaneously destroying capital for many investors who rushed in without discipline.
The technology succeeds. The investment outcome varies.
What stands out today is not only the scale of spending but also how much of it is structured to remain out of sight. When obligations move from the balance sheet to the footnotes, they do not disappear. They simply become harder to assess.
And most investors do not read the footnotes.
No one can predict precisely how this cycle will unfold.
What we can observe is this: companies once celebrated for being asset-light are now competing to build and control some of the most capital-intensive infrastructure imaginable.
Much of the financing supporting this expansion depends on a near-perfect scenario—strong demand, no delays, sustained relevance of technology, and infrastructure that keeps pace.
That is a demanding set of assumptions.
Some investments will succeed, and succeed dramatically. Others will not.
Our role is not to chase excitement. It is to assess risk with clarity and act with discipline.
As Warren Buffett observed:
“The less prudence with which others conduct their affairs, the greater the prudence with which we should conduct our own.”
This is the approach we apply in managing the fund—focused, selective, and grounded in long-term value rather than short-term narrative.
We would welcome the opportunity to share how we are positioning capital in this evolving landscape.
Awet Ghirmay, Head of Investment Research
If you want to learn more about this topic, please find the link here.
More information about our fund can be found here: Hargreaves Lansdown or AJ Bell.
Risk warnings
The value of investments and the income derived from them may fall as well as rise. You may not get back what you invest. This communication is for general information only and is not intended to be individual advice. You are recommended to seek competent professional advice before taking any action. All statements concerning the tax treatment of products and their benefits are based on our understanding of current tax law and HM Revenue and Customs practice. Levels and bases of tax relief are subject to change. This blog is based on my own observations and opinions.
