From software to shovels: Navigating market shifts in the next phase of AI

2025 was another exceptional year for equity market investors with fantastic returns. It was a year dominated by the themes of Trump and AI. But through the year, we also saw a lot of speculation creep into the market. This speculation created concern for us that while we may be in a bubble, we also need to watch how there are elements of this AI cycle that could continue for some time. And we need to be very cautious about how we can position our investors alongside the potential opportunity of this, but also not to be on the wrong side of the speculation taking place.

But as we came into 2026, while the equity market is up again around 4% in U.S. dollar terms, the shape or the leadership of the market has changed. There's been a distinct shift by the market towards cyclicals.

The best sectors this year have actually been materials, industrials and energy with it around benchmark at the moment. And part of that is because there are more concerns creeping in by investors around where we sit in the cycle for AI investment, but also where AI might start to disrupt some traditional business models.

Over the past two weeks, part of that concern has actually been related to software and the overall market narrative or question has been is software dead? 

Now if we think about software, a lot of these software companies are incredibly profitable companies. They provide code over data and they lock in the consumer, some of them being monopolies, by saying you must license all your staff to our software and we will charge you very large enterprise based fees and often subscription recurring revenue.

That is until AI proves that potentially it can displace software.

And up until a couple of weeks ago, despite people being aware of this risk, there was a there was the underlying belief that software could actually have AI embedded into it as another add on feature. However, there was a launch by Anthropic, their model clawed particularly something called Cowork that actually showed. We've now progressed to the AI layer being something that can actually replace software entirely.

This new development in AI actually enables the AI to act like software, performing end to end tasks and embedding digital humans that are called agents to actually do a lot of these things on behalf of software and actually replace it entirely. Investors have reacted by selling down software.

We've actually seen the software index down close to around 20% in a week and a half, and some individual software companies down even more. What are our views on this?

Our views are the end, what we call terminal value of these companies is definitely something that investors should question now. Is this the first sign that some of the monopolistic status, the competitive Moat of these companies is actually being eroded?

And if so, what does that mean for the future pricing power and subscription recurring revenue that these companies have been able to attract? There will be some losers from this, but also there will be some companies that have unique benefits that they can actually supply to their customers. That means they won't be disrupted.

But for now, it's a throw the baby out with the bathwater. And it's also something that this negative sentiment has actually been cast across a lot of the other components within AI and tech in general.

What we're actually watching right now is not just where there might be some opportunities from some of the oversold software companies in this basket, but also where this negative sentiment starts to creep too much into the AI ecosystem.

What we believe is that while there are some bubble like speculation taking on in the market and concerns around software, we're gaining more conviction that there's actually a stronger cycle taking place in demand for AI.

We saw amazing CapEx stats and new announcements released by the hyperscalers, again, surprising our numbers and the street in general. And for us, we're actually also starting to see bottlenecks where there are components that go into building AI capacity, such as data centers where we just frankly don't have enough, such as memory. And the time it's going to take to actually build out the capacity for the demand that exists is also going to take some time.

That starts to give us some conviction that not only other dollars substantial and enough for this cycle to continue, but the fact that we can't add that capacity in one hit means that it elongates and the duration of this cycle is expected to take place for longer.

So for us, as we move forward from here, we believe it is important for investors to identify where there are pockets of speculation taking place. Valuation is very, very important. It's important to identify where AI might actually be a disruptive force. And we've started to see that in software and the valuations of these companies being disrupted can become very, very vulnerable.

But we still do believe that there is a distinct opportunity within AI when it comes to hardware. The picks and shovels that are critical to building out this we believe is actually going to provide more of an opportunity for some time from here.