Boom or bust, AI remains an American phenomenon
‘‘ An inescapable question has dominated economic and markets discourse in recent weeks: are we in an AI stock market bubble? Last week, two of the world’s most powerful central bankers weighed in on the question, when the Bank of England warned of the possibility of a “sudden correction” in equity valuations. The worry is that mammoth market capitalisations for the world’s largest AI-centric tech firms are not commensurate with future returns. Also last week, US companies’ quarterly earnings beat expectations by the largest margin in four years, powered almost entirely by the world’s most valuable companies, which continue to pour money into the AI arms race.
The phenomenon is novel enough to have led to the creation of a separate category to describe said firms and their investment activity: the “hyperscalers”. The five largest AI hyperscalers are Amazon, Alphabet, Microsoft, Meta and Oracle. They are broadly defined as the firms spending hundreds of billions of dollars on building AI computing power — through data centres and software — to meet expectations of exploding demand for AI over the next decade.
Their capital investment has risen from $125 billion two years ago, to $200 billion in 2024, and is on course to reach close to $350 billion this year.Their stock market valuations have reached record highs.
The numbers are gigantic. The five largest listed US tech companies are worth 16 per cent of the entire global stock market — more than the combined equity valuations of Europe’s largest 50 firms, and the leading quoted companies in the UK, India, Japan and Canada. The largest ten listed US companies make up a quarter of the global market and are worth $25 trillion, having grown by $5 trillion this year. Chipmaker Nvidia hit a $5 trillion market valuation last week
The AI spending bonanza is largely a US phenomenon. The hyperscalers, and rapidly growing private companies like Anthropic, are all American. There are close to 5,500 data centres in the US, compared with just under 550 in Germany and about 520 in the UK. That means that the AI investment boom, its impact on the economy, jobs and workers, is becoming a preoccupation for the Federal Reserve, the US central bank, which is struggling to make sense of how it should respond.
Unlike the Bank of England, the Fed has not warned about an AI stock market bubble that could be inflating under its watch. Christine Lagarde, president of the European Central Bank, has hinted at the risks of a boom in AI spending on the continent, despite no clear evidence that this is happening anywhere in the eurozone this year. European imports of computers and IT related equipment have been stagnant since the release of OpenAI’s ChatGPT in late 2022. Imports have more than doubled in the US, by contrast.
Lagarde and Andrew Bailey, governor of the Bank of England, are not contending with the challenges that face the Fed’s Jerome Powell. The AI hype is becoming a seriously complicating factor for the central bank chief, whose job and institution is under fire from the White House.
To start, the investment boom has been immune to the Fed’s monetary policy action. The US had been yanking up interest rates since 2022 in an attempt to make borrowing more expensive. Yet Fed policy has not touched the hyperscalers, which have been funding investment through their bumper earnings, rather than borrowing. The five hyperscalers are spending 60 per cent of their cash flows on building data centres. Borrowing costs have next to no impact on their business decisions.
“I don’t think interest rates are an important part of the AI data centre story,”Powell said last week.
Respected economists such as Jason Furman at Harvard think the AI gold rush has been the only thing keeping the US economy from slipping into recession this year.
Even on more muted measures, the spending boom accounted for about 1.6 per cent out of the 3.8 per cent annualised GDP growth in the second quarter, according to Bank of America. That raises the question of whether the US economy’s resilience in the face of President Trump’s tariffs, migrant deportations and a government shutdown is built on sand or on the most transformative technological innovation since the internet. Nobody — including Powell, economists and investors — knows the answer.
But while Fed’s interest rate policy doesn’t touch the fastest growing part of the stock market, the hyperscalers’ impact on the US labour market and economy is huge. Amazon has said it will cut 14,000 roles as a result of its transformation for the AI age. Meta and Salesforce have also announced cuts at a time when they are reporting bumper corporate profits.
The Fed is now cutting rates to support a slowing jobs market, which should help to cushion the blow. But soaring stock valuations, creating a “wealth effect” for the richest Americans exposed to the fortunes of equity markets, are deepening economic divides in the country.
Households with median incomes above $150,000 a year are the most optimistic about their fortunes, while those over $50,000 reported the lowest rates of consumer confidence last month. The Fed’s monetary policy is a blunt instrument.
Annualised growth of close to 4 per cent and 3 per cent inflation does not usually make an environment ripe for monetary loosening. But the cooling jobs market, dislocations from AI and more tech layoffs are pushing in the other direction.
Beyond the question of whether the AI bubble will burst or not, central bankers are struggling to understand whether artificial general intelligence will be permanently inflationary or disinflationary. For the optimists, the potential AI productivity gains in white and blue collar professions could be a silver bullet for stagnant economies and falling living standards in rich countries and they will keep a lid on price growth.
Mehreen Khan is Economics Editor of The Times"
No comments:
Post a Comment