AI is entering our homes and workplaces, becoming a matter of national interests and technological sovereignty. We can sense it is opening an enormous range of potential applications, potentially life-changing. It is logically now a central topic of social, political and even philosophical debate.
To make this revolution possible, an entire technological and industrial complex has emerged. Hundreds of hyperscale data centres are being built, filled with increasingly powerful GPUs and supported by the infrastructure needed to connect, cool and power them.
To fuel this complex, a new financial ecosystem has developed. AI-related infrastructure investment is moving towards the trillion-dollar-a-year range, making this one of the largest investment cycles in modern history.
At its core are the hyperscalers Amazon, Google, Microsoft, Oracle and Meta. They initially financed much of this buildout through their enormous free cash flows. They are now increasingly issuing debt and helping other players raise financing, backing them with their long-term capacity commitments.
They also invest in AI labs, NeoClouds and other parts of the ecosystem. More than half of all this investment flows into semiconductors. Earnings across the chip sector have skyrocketed, while companies such as Nvidia have also become a key investor in the ecosystem itself.
This AI complex has been the dominant engine of the equitable market of the past four years. Beyond corporate earnings and stock market performance, it has generated a fantastic wealth effect, particularly in the United States. And more broadly, it has been a massive tailwind to broader economic growth.
AI usage is real. The infrastructure spending is real. The revenues are real. Tomorrow’s profit are still expectation, though. For markets to continue higher, the return generated by this investment cycle must eventually justify the extraordinary amount of capital being deployed, and the timing of the expected monetisation will matter too.
Railways, electricity and the internet transformed the world. Yet many of the investors who financed their initial buildouts suffered large losses.
So the question is no longer whether AI is real. It clearly is, in this form or the next. It’s not the point. The question is whether its economic returns will arrive quickly enough and at a sufficient scale to justify the capital being committed today.
We will, of course, follow these developments to understand the broader picture. More importantly, we will watch the signals that tend to move first: public equities and the credit markets, as well as the appetite investors show in new funding rounds.
For now, humility compels us to give due credit to the most capable companies in the world. Experience reminds us to stay alert to our own behavioural biases. Our minds are not wired to grasp a transformation of this scale, and that is precisely when judgement is most easily distorted.
We therefore stay the course, but not blindly. History teaches that even genuine revolutions can be accompanied by excess. We continue to advocate diversification and will monitor the global geoeconomic context closely.
