AI developers themselves are going even further. They estimate that the US economy could grow from over 30 to 44 trillion dollars by 2030 – which amounts to as much as 10% per year. Such growth cannot come solely from higher wages and increased spending. A large part must come from lower prices. Knowledge is becoming a mass-market commodity, and what is in abundant supply becomes cheap.

China illustrates what this looks like. There, production has grown strongly, whilst domestic demand has lagged behind. Industrial prices are 9% lower than in 2012. Companies lost their pricing power, profits stagnated and share prices underperformed. The 10-year interest rate fell to 1.6%. For consumers, that is good news. For those who have lent money, it is a different story.

Why banks are vulnerable

A bank finances the future. A loan to a software company, a consultancy firm or the owner of an office building is based on cash flows that are yet to materialise. The bank assumes that, in five or ten years’ time, the customer will be able to charge roughly the same prices as today. But what happens if AI halves the price of a service? Then those cash flows will never materialise. The debt, however, remains just as large.

The economist Irving Fisher described this mechanism as early as 1933. When prices fall, a debt effectively becomes a heavier burden. Turnover shrinks, whilst interest and repayments remain the same. This is precisely why deflation is so dangerous for an economy that runs on credit.

Moreover, a bank always has a weak balance sheet. It borrows short-term funds from savers and invests them long-term in loans that are difficult to sell. Equity is thin, often amounting to just a few per cent of total assets. A bank can usually absorb a setback in a single sector – think of the shipping industry or commercial property following the credit crisis. That is painful, but the losses remain limited because other sectors simply carry on as normal. Diversification is every bank’s insurance.

However, that insurance does not work if all sectors are hit at the same time. And that is precisely what AI does. It affects solicitors, accountants, call centres, software suppliers, logistics, and ultimately industry as well. Banks’ credit models are built on historical correlations between sectors. But a shock that strikes everywhere at once is not found in the history books. A bank can survive a problem in a single sector. It cannot survive a problem across all sectors.

This is not a scenario for tomorrow. The US is still grappling with inflation rather than deflation, and the debate centres on interest rate rises. But loans taken out today often run well beyond 2030. They are priced on assumptions that may be outdated in a few years’ time.

From factor investing to AI investing

In asset management, AI actually works to the investor’s advantage. Over thirty years ago, I met Joseph Lakonishok at the Amsterdam Institute of Finance. He was one of the pioneers of quantitative value investing. According to him, there was a significant gap between theory and practice. On paper, a systematic value strategy delivered a clearly higher return. In practice, however, transaction costs alone ate up a large part of that advantage.

Much has improved since then, particularly following the so-called ‘quant winter’ – the years in which factor strategies proved deeply disappointing. Fund managers began to operate more efficiently, costs fell and data quality improved. AI is the logical next step in this development. It processes far more information, including text such as annual reports and conversations with directors. It does so faster and more cheaply than a team of analysts. As a result, active investment decisions are increasingly being made by AI.

The result is that, for the first time, an active asset manager can genuinely promise a better return than the index, without costs cancelling out that advantage. There is, however, one caveat. If everyone starts investing in the same way, the advantage disappears. A market can never be completely efficient, because then it would no longer be worthwhile for anyone to gather information. But that equilibrium is still a long way off.

Why wealth management is becoming more valuable

Does this mean that investment advisers will disappear? On the contrary. Advisers in wealth management excel particularly at relationship management. That is the essence of private banking. A client does not come for a model, but for trust. That trust is most valuable at times when markets are falling and emotions take over. A good adviser then protects a client from classic behavioural errors. This is so human that AI cannot replace it.

AI does, however, make that adviser better. Reports, preparing for meetings, and checking whether a portfolio suits the client are carried out more quickly and accurately. The adviser has more time left for what really matters: the conversation.

Furthermore, AI will create a great deal of new wealth. Founders, early employees with share options, and investors in successful companies will become millionaires. This new group of high-net-worth individuals will benefit from lower costs and better investment returns. Their numbers will grow significantly. For the Netherlands, where many live off their investments, this could become an economic driving force – though not under the current tax bracket system.

Scale is everything

AI requires substantial investment in data, computing power, security and compliance. These are largely fixed costs. The more assets a firm manages, the lower those costs will be per euro. AI will therefore trigger a wave of economies of scale in wealth management. Assets under management will become an even more crucial factor in the valuation of an asset manager than they are today.

For investors, this signifies a shift within the financial sector. Value is shifting from the balance sheet to assets under management. From organisations that earn money by financing future cash flows to those that earn money through trust and scale. Banks lend at yesterday’s rates. Asset managers benefit from tomorrow’s abundance.

More columns by Han Dieperink