Infrastructure: A volatility-lite approach to AI investing

Source: Trustnet

Much of the financial world breathed a sigh of relief when Nvidia CEO Jensen Huang reported yet another set of impressive revenues and profits.

With a market capitalisation of $5trn, Nvidia has become the defining poster child of the AI boom. Huang’s latest upbeat summary of the company’s performance and prospects once again eased concerns about an AI bubble.

It’s worth considering for a moment, though, what might have occurred if Nvidia had fallen short of expectations.

In my view, the lesson here isn’t that AI is in grave danger of becoming a busted flush. It’s not inconceivable that the commotion surrounding its adoption and impact could persist for years, given the extraordinary ‘hyperscaling’ already underway.

I believe all the nervousness that now routinely precedes Nvidia’s announcements instead underscores the prudence of diversifying AI-related investments. After all, it isn’t especially healthy to feel your fate rests almost entirely – if not exclusively – in the hands of a single business.

So how can investors tap into arguably the most significant theme of our age without finding themselves unduly exposed to the usual suspects? What lies beyond the volatility risk increasingly associated with the high-growth stocks that dominate headlines, indices and so much investment thinking?

One answer lies in recognising what the aforementioned hyperscaling ultimately involves. Very simply put: it involves infrastructure.

A picks-and-shovels perspective on AI

The infrastructure needed to cope with ever-larger datasets demands genuinely extraordinary levels of investment. This much is broadly appreciated.

Perhaps less understood is that it also requires collaboration between multiple stakeholders. This is where many investors remain unaware of the substantial array of AI-linked opportunities available to them.

By way of illustration, consider that OpenAI – which originally triggered AI mania by launching ChatGPT in 2022 – intends to splash out $1.4trn on data centres over the course of the next eight years. Meanwhile, collectively, Amazon, Google, Meta and Microsoft are set to lavish approximately $400bn on AI before the end of this year.

By any standard, these are serious commitments to the cause. Yet it would be wrong to assume the tech titans plan to do all this work themselves.

The hyperscaling effort instead relies on ‘picks and shovels’ businesses. Like the merchants who sold the equipment that prospectors used during the gold rushes of the 19th century, these are in many respects the true enablers of transformation.

Crucially, like those merchants, they’re also among the most likely success stories. They include network operators, specialists in small nuclear reactors and other novel power sources, manufacturers of cooling systems, transport companies and – above all as far as our fund is concerned – energy suppliers and other utilities.

An added potential attraction is that many of these businesses aren’t just helping to facilitate the AI revolution: they’re also already taking advantage of it. For example, we’re now encountering ever more evidence of utilities using AI to enhance their own networks’ capacity and efficiency.

Same theme, different drivers

This gives us two reasons to invest in infrastructure from an AI perspective. The first is its contribution to the positive disruption to come. The second is the range of benefits it can derive from the positive disruption that has taken place to date.

But remember that we’re also interested in a third reason: relatively limited volatility. To highlight infrastructure’s appeal in this regard, let’s compare the recent volatility of all the holdings in our own portfolio with the recent volatility of Nvidia.

Over a three-year period, according to our analysis, there was a correlation of just 0.03. In other words, there was practically zero overlap between the two.

This can be explained relatively easily. We invest in companies whose earnings are shaped by long-term contracts, regulated returns and physical service demand – drivers that are very different from those that influence the earnings of, say, chip manufacturers and cloud-computing companies.

Similarly, the volatility correlation between our holdings and the constituents of the tech-heavy Artificial Intelligence & Technology (IAIQ) Index over the same period was only 0.3. This underlines that the portfolio is positioned some distance from the regular ups and downs of an AI-themed benchmark.

The lesson: it’s quite possible to invest in AI without being at the mercy of the dynamics that continue to buffet this theme’s best-known names. Such an approach could offer measured participation in AI-linked upside while at the same time delivering resilience during rotations or risk-off periods.

It might be perceived as less exciting, less glamorous, less spectacularly cutting-edge. But it’s a long way from hanging on Jensen Huang’s every utterance – and that, on balance, may be no bad thing.

Tim Humphreys is co-manager of the IFSL Marlborough Global Essential Infrastructure Fund. The views expressed above should not be taken as investment advice.