I've spoken before about the stages of the AI revolution. The first stage is the infrastructure build-out. The second is the rise of agentic AI, where systems begin acting independently rather than simply responding to prompts. The third is what I call "everything AI", where artificial intelligence becomes embedded in almost every product, service and industry. We are still firmly in the first stage.
In my opinion, AI will prove to be the most significant technological advancement in human history. It will likely surpass even the Industrial Revolution in both its scale and its impact. Like every transformational technology before it, however, the path forward will not be a straight line. Investors should expect periods of extraordinary optimism followed by equally painful corrections. The innovation may be exponential, but investment cycles are not.
The first phase is the easiest to understand because it involves building the foundations. Think of it like constructing a new mine. It may take a thousand workers to build the roads, install the equipment, connect the power and prepare the site. Once the mine is operational, only a fraction of those workers are needed to keep it running. The same principle applies to AI. Today we are pouring huge amounts of capital into chips, data centres, networking equipment, cooling systems and electricity generation because that infrastructure is needed before the next phase of AI can prosper.
That has created an enormous investment opportunity, but investors also need to exercise caution. Every major technological revolution has experienced a period where investors assume today's shortages will last forever. Railways required steel. Electricity required power stations. The internet required fibre optic cables. AI requires computing infrastructure. These periods often lead to extraordinary investment returns, but they also result in overbuilding and excess capacity as companies race to meet seemingly insatiable demand.
Throughout history, technology has advanced at an exponential pace. Every generation of computing becomes more powerful and more efficient than the last. AI models continue to improve, software extracts more performance from existing hardware, and specialised chips deliver greater computing power in smaller and more efficient packages. Over time, demand for AI infrastructure will change. The market tends to extrapolate current conditions indefinitely. That doesn’t always hold up, especially with regards to technology.
None of this changes my long-term optimism. AI will transform businesses, industries and society over the coming decades. Productivity gains will be massive, entirely new business models will emerge and many of today's largest companies will look very different in ten or twenty years' time. The opportunity is real.
The challenge for investors is recognising that a technological revolution and an investment boom are not the same thing. The AI revolution is real. Parts of the infrastructure build-out may also prove to be overvalued over time. Understanding the difference will likely be one of the most important investment decisions of the next decade. Investors cannot blindly invest in AI. It is critical to understand the technology, the pace of change and consider the future revenue and profit that will ultimately justify and drive company valuations.