As with any investment, your capital is at risk.

Over the past 20 years, investors have grown accustomed to dramatic change. The rise of ecommerce, social media and mobile computing overturned business models that had stood for decades.

Yet I believe the period we are entering will dwarf those shifts. The next two decades will be marked by a speed and intensity of transformation that makes the last 20 look steady by comparison.

Artificial intelligence is at the heart of this. It is not just another technology trend but a new foundation for the global economy.

The challenge for investors is to recognise that standing still is not an option when the ground beneath you is moving.

Scale, inevitability and asymmetry

Equity markets are not a level playing field. History shows that wealth does not compound evenly across thousands of listed companies. Instead, it gathers disproportionately around a small number of extraordinary businesses. What is different today is that the outliers of the past decade are not just incrementally ahead but operating at a scale that makes them virtually unassailable.

Consider the investment required to compete in AI infrastructure. In 2025 alone, Alphabet, Amazon, Microsoft and Meta are expected to spend more than $400bn in total on AI datacentres, chips and cloud infrastructure. That’s equivalent to two-thirds of the UK’s total annual investment spending. These are not simply large enterprises but corporate equivalents of nation-states.

For investors, this is not about chasing AI market hype. It is about recognising inevitability. When a business is both vast and still compounding at double-digit rates, the arithmetic of future market returns is tilted heavily in its direction. That is why the idea of ‘mean reversion’ feels increasingly outdated. The companies shaping the future are not drifting back to average. They are pulling further away from it.

The growth investor’s task then is not to diversify endlessly in search of safety, but to accept that the path to long-term wealth will be defined by asymmetry. A small number of companies will account for the bulk of value creation. The job is to own and hold them through the volatility that inevitably accompanies leadership.

A new technology paradigm

The internet and mobile devices defined the last computing era. These technologies reshaped how supply met demand, enabled entire categories – including ride-hailing, social media, ecommerce and home-sharing – and gave rise to a handful of platforms that now dominate the global economy.

But with the launch of ChatGPT three years ago, we crossed into a new era. To borrow from the philosopher of science Thomas Kuhn, this is a paradigm shift. In his account of scientific revolutions, Kuhn described how knowledge progresses not in a smooth line but in leaps between frameworks. Newtonian mechanics framed our thinking about what was possible until the theory of relativity supplanted it and forced us to reconsider those limits. In the same way, the internet-mobile-cloud paradigm is giving way to an intelligence paradigm. AI unlocks new capabilities that legacy systems cannot deliver or compete with.

AI is not an incremental enhancement but a new foundation altogether. The frameworks of the past two decades, such as search, ecommerce and social networks, were powerful because they connected people at scale. The new era is powerful because it connects intelligence at scale. It allows machines to reason and to collaborate to accomplish tasks that were not previously possible.

Paradigms matter because they determine where value accrues. In the internet-mobile era, the greatest rewards flowed to the application layer. To businesses like Uber, Airbnb and Instagram that built services on top of the new infrastructure. AI is following the same pattern. The infrastructure layer is being secured by a handful of cloud giants with the resources to build datacentres, buy chips and train models. However, the applications that sit on top of this infrastructure will create vast new companies of their own.

Applications

The first breakout application has been conversational AI. ChatGPT reached 50 million users in just two months, making it the fastest-growing consumer app in history, faster than TikTok or Instagram. Today, it is one of the five most-visited websites in the world and generates tens of billions in annual revenues. It is not just capturing attention in the way social media platforms did – it is delivering utility, which is a more durable basis for value creation.

The second application wave is coding. Software engineers equipped with AI tools are already at least 20 per cent more productive. Companies such as Anthropic are turning this into billion-dollar revenue streams in a matter of months. We are moving towards ‘agentic’ systems that don’t just assist coders but act as autonomous members of development teams. The implications are enormous, given that software underpins almost every sector of the modern economy.

Beyond the digital realm, the effects are equally profound. In transportation, autonomous vehicles are no longer experimental. Waymo already provides hundreds of thousands of paid rides weekly in US cities, and exponential growth should be the default expectation. In logistics, firms such as Zipline use drones to deliver goods in minutes, not days, collapsing delivery times and unlocking entirely new categories of demand.

Zipline’s commercial drone deliveries over time

Zipline reached one million commercial drone deliveries in April 2024, becoming the first company in the world to achieve this milestone:

Source: ©Zipline

These services exemplify what economists call Jevons’ Paradox: when technology increases efficiency, consumption rises rather than falls. Faster delivery doesn’t mean a more efficient system of the same scale – it means much greater use of delivery. More efficient coding doesn’t mean costs fall. It means vastly more software is written. We should expect to see the same phenomenon in every domain where AI will be applied. Efficiency creates new markets rather than satiating old ones.

Even in seemingly low-tech industries, the same pattern emerges. Samsara builds software to manage vehicles and equipment. It began with simple dashcams for truck fleets. Today, it provides a suite of AI-enabled services that optimise routes, give feedback to drivers and schedule predictive maintenance. What started as a camera in a cab is evolving into an intelligence platform for physical operations.

From science fiction to reality

At the frontier, we are witnessing possibilities that border on the surreal. Colossal Biosciences has already achieved the first ‘de-extinction’ and plans to bring back the dodo and woolly mammoth in the coming years. It sounds fantastical, but it is enabled by AI’s ability to process and manipulate genomic data at a scale that would be impossible otherwise. Whether or not one agrees with the ethics, the commercial and scientific implications are vast.

These examples demonstrate the same underlying truth: AI is not a single product or service. It is a general-purpose technology that will ripple through every corner of the economy. From the obvious (coding, transport, logistics) to the unexpected (cosmetics, conservation and beyond), it lays the foundation for new categories, business models and winners.

The illusion of safety

One of the most dangerous assumptions in investing is that you can find stability in companies deemed ‘safe’. Time and again, real capital destruction has come not from volatile growth stocks, but from incumbents whose earnings and valuations were built on the illusion of permanence.

The supposed safe havens can be the most exposed when technological change accelerates. They suffer a double blow: earnings erode as their models are disrupted and valuations compress as the market recognises they were never as dependable as assumed. Some of my worst investment mistakes have come from backing companies that looked steady, only to discover that steadiness was a mirage.

This is why I believe the framework of mean reversion is increasingly unhelpful. Much market commentary still rests on the idea that sectors and valuations revert to long-run averages. But in a world of exponential change, ‘the long run’ itself is shifting. The average of yesterday may bear little resemblance to the future.

The cost of a technology falls predictably as cumulative production rises. This pattern has held true in semiconductors, solar and batteries. The companies deploying those technologies have displaced the incumbents tied to the inflating costs of alternative approaches. 

The dramatic fall in solar panel, battery, storage, and AI inference costs

Use the arrows below to see the exponential cost decline in action:

AI is no different. Since 2010, the compute used to train leading models has doubled roughly every six months. Costs per unit of intelligence are falling so fast that new models have limited shelf lives. Incumbents built on yesterday’s cost curves will not revert to the mean because they cannot compete with the phenomenal increases in productivity in this vibrant new industry.

Societal consequences

It is tempting to talk about AI purely in terms of efficiency, productivity and profits. But its consequences run much deeper, and they will be felt unevenly.

At the frontier, the competition for talent already illustrates the point. The very best AI engineers now command packages in the hundreds of millions. Meta reportedly offered up to $1bn pay packages to lure researchers from rivals, an almost unimaginable price for individual talent. At the same time, entire categories of routine work are at risk of being automated away. Within companies, this creates vast disparities in pay. Between companies, those who can harness AI surge ahead, while others fall irretrievably behind. Between countries, those with affordable energy and computing infrastructure become hubs of value creation, while others are relegated to consumers of imported intelligence.

With its high energy costs, the UK falls into the latter camp. We will ‘pay for it’ rather than ‘build it’. That is not a position of strength.

Inequality at these scales does not just show up in income data. It reshapes politics. As the benefits of AI accrue narrowly, populism finds fertile ground. We have already seen democratic systems strain under far less pressure. The following decades will test them further.

And then there are the more disquieting edges: malign uses of generative media, the potential for autonomous weapons, the societal shock of de-extinction and synthetic biology. These are not distant hypotheticals. They are already happening on a small scale. The world is about to get ‘weird’ more quickly than many of us can comfortably imagine.

Yet I remain, on balance, an optimist. For a century, limited access to intelligence has constrained human progress. If we can augment human capability with machine intelligence on an exponentially declining cost curve, the potential to enhance people’s quality of life is immense. The challenge is not whether the technology works but whether our institutions, cultures and policies can adapt quickly enough to channel it productively.

Implications for investors

What does all of this mean for equity investors?

First, equities are not the place to minimise risk. Investors may rightly wish to preserve wealth, but bonds and cash are better tools for that job. You should own equities to capture the asymmetric returns that accrue to the winners of structural change.

Second, avoiding the US is an active bet against the most dynamic part of the global economy. Nowhere else – not Europe, emerging markets or even China – has the same density of AI infrastructure builders and application innovators. The US is not ‘expensive’ in any meaningful sense if what you are buying is unique.

Third, not all exposure is created equal. It is tempting to own the Magnificent Seven passively and declare the job done. But history suggests that the application layer will throw up entirely new giants – think Amazon, Uber and Airbnb. Identifying them early and holding them through volatility is the task of active investment.

Finally, culture matters. The companies best placed to thrive are not simply those with capital, but those with ambition, adaptability and founder-led urgency. Toby Lütke’s decision at Shopify to abandon delivery operations and redeploy the organisation towards AI is a case in point. That willingness to pivot, to stop being the best implementation of the old paradigm and commit fully to the new, distinguishes long-term winners.

Investing in the inevitable

Today’s real danger lies in assuming things will revert to how they have always been. They will not. The coming decades will bring extraordinary advances and extraordinary dislocations. They will reward scale, ambition and adaptability. They will penalise complacency, inertia and the illusion of safety.

For investors, the prudent course is not to spread capital thinly across what looks comfortable, but to ensure meaningful exposure to what is inevitable.

In today’s market, that means owning the companies, almost all of them American, that are shaping the intelligence paradigm and creating the next wave of applications on top of it.

Avoiding US growth is not just a missed opportunity. It is a failure to insure portfolios against the upheavals ahead. In a world of accelerating change, it is, quite simply, perilous.

Risk factors 

The views expressed should not be considered as advice or a recommendation to buy, sell or hold a particular investment. They reflect opinion and should not be taken as statements of fact nor should any reliance be placed on them when making investment decisions.

This communication was produced and approved in October 2025 and has not been updated subsequently. It represents views held at the time of writing and may not reflect current thinking.

Potential for Profit and Loss 

All investment strategies have the potential for profit and loss, your or your clients’ capital may be at risk. Past performance is not a guide to future returns.

This communication contains information on investments which does not constitute independent research. Accordingly, it is not subject to the protections afforded to independent research, but is classified as advertising under Art 68 of the Financial Services Act (‘FinSA’) and Baillie Gifford and its staff may have dealt in the investments concerned.

All information is sourced from Baillie Gifford & Co and is current unless otherwise stated. 

The images used in this communication are for illustrative purposes only.

172512 10057677

Read the other articles in our latest Long View series: