As the hype around artificial intelligence (AI) grows, investors are asking whether the introduction of generative AI technologies signals the start of the next transformational cycle in markets, or whether we are simply seeing the continuation—or perhaps culmination—of the information technology cycle that started in the late 1970s.

Forecasting structural change

Many previous cycles of fundamental structural change—for example, the invention of the steam engine, the introduction of railroads, and electrification—have led to higher productivity and the creation of significant wealth. However, these cycles have also often come at a cost, in terms of lower levels of employment and greater inequality. Who will benefit most from structural change, who will struggle, and over what time period, are the questions that our global equity research analysts spend a large part of their time integrating into their long-term corporate earnings forecasts.

It is clear to us that generative AI will have a profound impact on many industries. Having a common valuation process that can evaluate the basis for structural change within each industry gives our global equity funds a significant edge when it comes to breaking down how the structural and cyclical implications of AI may play out across the global economy. While it is difficult to invest in innovation cycles early, much of the return for investors will ultimately come from being able to identify those companies that have a sustainable competitive advantage once the cycle is underway. The aim is to pay a fair price for the “winners”, and to avoid overpaying for those companies that are just riding the hype wave.

We have honed our process to be selective when choosing which companies to invest in and to try avoiding falling into common pitfalls when navigating structural themes, such as the introduction of generative technologies. These include:

• Selecting companies where the range of outcomes is less binary
• Considering different ways to gain access to the theme
• Remaining focussed on long-term cash generation potential

How we navigate market inflection points

We aim to identify attractively valued stocks based on our proprietary, fundamental long-term analyst forecasts. At the moment, the spread between stocks that we forecast to be undervalued and those that we forecast to be overvalued is widening. Wide valuation spreads have been observed at various market inflection points in the past, such as the dotcom bubble, and the global financial crisis.

These peaks in our valuation spreads are a reflection of two things: the polarisation of market views into identifying winners and losers, and the allocation of capital regardless of valuation; and the widening of long-term outcome ranges as the business models around new technology emerge. Who would have thought back in 1998, for example, that Amazon would go from online book seller to the e-commerce giant it is today? Or that Sun Microsystems would eventually stop being the dot in dot com?

In AI, we are currently observing a wide range of outcomes, as we extrapolate current growth rates and trends for early winners (such as Nvidia) and significant risks of disruption for other businesses. When it comes to navigating structural themes, we try to maintain a strong valuation discipline and look for those companies where the range of outcomes is less binary. We believe it is important to be selective and also to consider different ways to gain access to the theme. In AI, this means looking at companies that are helping to enable the development of AI, as well as the companies that are developing AI products.

However, the distinctions between what constitutes enablers and a final product provider can be more blurred at this early stage. Take Nvidia, for example, which doesn’t just sell microchips that enable AI, but has also built an ecosystem around its own AI graphics processors that is allowing the company to increase sales of services and software in addition to hardware, and to increase the switching costs for users.

What we can learn from the dotcom bubble

While we may not be able to narrow down the range of outcomes at this early stage of AI adoption, we can learn from the similar structural shift that occurred during the dot.com bubble of the late 1990s/early 2000s. Similar to today’s AI trend, investors started to look at every company through the dotcom lens, asking how they would cope with the shift to, or disruption from, an online and digital world. Now, every CEO is being asked what they are doing about AI, and if their business is going to be disrupted.

Cisco, in particular, was one of the enablers of the internet, and at the peak of the bubble had the largest market cap in the world. While forecasts were not necessarily far off from reality 20 years on, it became impossible for the stock to remain at such stretched valuations and the multiples (on sales) fell from 20x to 1x. For many of the other stocks that made up the Nasdaq at its 2000 peak, the internet phenomenon did not end quite as well and the valuation discipline inherent in our investment process helped us avoid many of the unsustainable business that proliferated at the time.

Disciplined focus on cash generation

While we are not afraid to adapt our process when the market environment changes, we believe it is very likely that long-term cash generation potential will remain the most appropriate valuation factor, even if the drivers of cash flow change over time. Our methodology has been consistent and proven over many decades, and has been shown to handle these periods of dislocation and inflection. Even if they don’t allow us to time the turning point in markets precisely, our proprietary valuation spreads provide a guidepost for how far we are willing to lean into certain market signals, and give us a critical element of valuation discipline when markets approach extremes.