Recently I heard the term “AI Bubble” spoken by my 13-year-old son. Like most inhabitants of this little green planet, I have been watching, with a slightly terrified eye, the inexorable march of the tech-nerds-come-billionaires self-made race to create a species more intelligent than ourselves. But it gave me a jolt that my kids were worrying about this too.
Debate about a potential AI bubble began last August when OpenAI CEO Sam Altman told reporters he believed AI could be in a bubble, comparing market conditions to those of the dotcom boom of the 1990s.
From traditional media like The Guardian and New York Times to leading podcasters such as Diary of a CEO’s Steven Barlett and technology ethicist Tristan Harris, artificial intelligence, and our lack of comfort with its unchecked progress, is a conversation we as humans are having everywhere.
So I was interested to see just how much more ‘airtime’ this potential AI bubble has been getting over the last month, including last week’s Guardian headline (25 February) A feedback loop with no brake’: how an AI doomsday report shook US markets.
The piece reports on how US stock markets have been hit by a wave of ‘AI jitters’, and shares in Uber, Mastercard and American Express fell on the back of apocalypse scenario posted on Substack by think tank Citrini Research, a little-known US firm that provides insights on ‘transformative megatrends. Their report went viral and – though it was completely speculative – warns about the impact of the technology on the world’s largest economy.
The New York Times ran a similar piece. Meanwhile investors were rattled by the report’s portrayal of a near future – the scenario begins now and ends in June 2028 – in which autonomous AI systems upend the entire US economy, from jobs to markets and mortgages. Share prices for Uber, American Express, Mastercard and DoorDash, all specifically named in Citrini’s report, all dropped between four and six percent as a result.
Another Guardian headline (on the same day), Meta agrees $60bn deal with chipmaker AMD despite AI bubble fears, reports on an agreement between Facebook and Advanced Micro Devices where Facebook has agreed to buy $60bn (American dollars) of artificial intelligence chips from the US semiconductor company despite fears about the vast sums committed to AI infrastructure projects.
The story reads: “It is one more massive deal in a year in which US tech companies are expected to spend $660bn on AI assets and may represent part of a broader pivot in Meta’s AI strategy, said Alvin Nguyen, an analyst at Forrester.”
The five-year deal involves Meta buying 10 percent of the California chip company, a similar arrangement to a partnership between OpenAI and AMD last year.
At the same time, American papers are reporting on proposed US data centres as they face a slew of problems amid grass roots protests against AI and with new constructions delayed or cancelled, raising questions about US’s ability to expand infrastructure to support the boom.
An investigative piece by the New York Times shows when Microsoft opened a data centre in central Mexico last year, nearby residents said power cuts became more frequent and water outages, which once lasted days, stretched for weeks. And their experiences are echoed elsewhere, as an artificial intelligence building boom strains already fragile power and water infrastructures in communities around the world.
Dozens of proposals for new ‘hyper-scale’ data centres, which house the infrastructure for artificial intelligence, have been proposed across the US. The centres can consume as much power as the country’s largest cities, meaning grids need to rapidly expand their infrastructure capacity, adding transformers, circuit breakers, high-voltage cables, steel poles and other pieces of equipment, to connect data centres to the grid.
Connecting to the grid is “the No 1 challenge we’re seeing”, Marsden Hanna, head of energy and sustainability for Google, said at a utility industry conference last month.
So all this sounds very far away.
But Auckland already has 29 data centres including two ‘hyper-centres’ in Silverdale and Hobsonville. Both are owned by CDC, who have similar large scale data centres across Australia.
And AI is not even here yet, at least not on the scale tech bosses and their investors are expecting it will be.
And amid our own rising cost of energy, one of the bigger factors in our cost-of-living crisis with our unregulated energy market, one has to ask, ‘how we are going to do it?’
Back in the US, the risk of an artificial intelligence bubble has become the top concern among credit investors and overshadowing geopolitical worries, according to a Bank of America survey last month. Of their investment-grade clients who buy and sell debt, 23 percent said the threat of an AI bubble was now their top concern, up from nine percent in December, according to the bank’s strategists Barnaby Martin, Ioannis Angelakis and analyst Mohit Agarwalla.
“[An] AI bubble is now seen as the No 1 investor risk for the first time ever,” they said in a survey report issued last Tuesday.
Huge amounts of money have been poured in developing AI, and so far with very little return. And if it cannot sustain itself power wise, it seems plausible the equation has hit a stalemate.
• Merrie Hewetson
