Advertisement
Stocks
OpinionWorld Opinion
Opinion
Andrew Sheng

Investors in AI-driven bubble risk forgetting painful lessons

AI-driven stock market hype is not independent of what is happening in the rest of the economy

3-MIN READ3-MIN
Listen
A man walks past a digital stock display in Tokyo on May 7. Photo:  EPA
Andrew Sheng is a former central banker and financial regulator, currently distinguished fellow at the Asia Global Institute, University of Hong Kong.

We all suffer from nostalgia bias, thinking back to a golden past, remembering the good and forgetting the bad. In looking to the future, we worry about the risks and are fearful of the unknown. This tendency makes us forget how the last financial crisis wiped out quite a few investors and created new wealth.

McKinsey Global Institute marked the consultancy’s centennial year by publishing a book called A Century of Plenty, meant to show that mankind never had it so good. It puts forward a bold scenario: by 2100, global gross domestic product could be 8.5 times larger than at present. The advances of technology can overcome resource limits, and human ingenuity could bring bigger and better progress and well-being.

However, the road to nirvana is paved with many pitfalls. Accepting that we should be optimistic and work towards the goal of greater progress, we must also survive the short-term bumps in the road and diversions.

Advertisement
The current euphoria around artificial intelligence (AI) is a key risk. Goldman Sachs has looked into the trillions of dollars in AI capital spending, which is fuelling an almost relentless rise in tech stock valuations. The key trends show up in this review: macroeconomic corporate dominance and stock market equity risk.

First, the AI business is today highly concentrated in selected players, making the S&P 500 one big trade in the “Magnificent Seven” rather than a diversified benchmark. These seven technology stocks have driven 85 per cent of the S&P 500’s gains in May after contributing 50 per cent in April. The top 10 largest companies now account for around 40 per cent of the S&P 500’s total market capitalisation.

Advertisement
Second, a key characteristic of the current concentration is the massive upfront physical infrastructure costs. The baseline projection is that the hyperscalers and tech giants will spend US$765 billion in AI capital expenditure in 2026, reaching a cumulative US$7.6 trillion by 2031. That is nearly one quarter of US GDP in 2025, which is significant because it is current capex that is driving GDP growth and expectations of future earnings, hence tech stocks’ high valuations.
Nvidia CEO Jensen Huang meets with Samsung, Hyundai chiefs over beer and fried chicken
Advertisement
Select Voice
Select Speed
1.00x