“Few economic concepts have proved more durable than the ‘gravity model of trade’, which predicts that countries trade most actively with their nearest neighbours. Canada would have to defy that force to realise Prime Minister Mark Carney’s vision of it becoming an ‘Associate Member’ of the European Union.”
Read more: Mark Carney’s EU pitch has strikingly strong logic.
The untrustables
In November 1871, the tycoon John D. Rockefeller met a group of executives in a New York hotel. Together, they hatched a secret plan for leading US railroads and oil refineries to give each other preferential treatment, thereby taking effective control of two crucial industries. “All in all, it was an astonishing piece of knavery, grand-scale collusion such as American industry had never witnessed,” Ron Chernow wrote in his of Rockefeller.
Railroads and oil were the motors of the US economy in the late 19th century. Artificial intelligence plays that role today. So when Dario Amodei, the CEO of Anthropic, calls to “slow the pace” at which AI models are improving, and OpenAI boss Sam Altman and Elon Musk of SpaceX SPCX.O agree, some detect a Rockefeller-style attempt to rig the market. “Stop pretending antitrust law has to be suspended so you can form a cartel,” wrote David Sacks, President Donald Trump’s former AI and crypto czar.
Amodei and his peers have good reasons to slow the pace of AI development. The recent security breach where autonomous OpenAI agents joined forces to break out of a test environment and hack another platform is cause enough to slam on the brakes. Anthropic and other so-called frontier labs are also fighting against tougher external oversight. Warnings from AI researchers that the technology could “kill us all by the end of the decade” hardly help the case for self-regulation.
A co-ordinated slowdown would also bring financial benefits. Anthropic and OpenAI have burned through tens of billions of dollars in their race to develop superior models. As Robyn Mak points out, the pair devote roughly two-thirds of their available computing capacity to training. An orchestrated pause in development might make their finances a bit more sustainable ahead of what are likely to be record-breaking stock market listings.
The problem with this argument, though, is that Anthropic and OpenAI will struggle to form an effective cartel. The building blocks of the AI frenzy are semiconductors, data centres, energy, and scientists. The frontier labs do not control the first three inputs, and AI researchers seem happy to jump ship when they get a more interesting offer. They’re unusually free to do so, too: the industry’s effective home state of California restricts the use of non-compete agreements. Any slowdown at Anthropic and OpenAI would therefore probably benefit AI laggards, including Chinese labs whose bargain-basement models are gaining traction around the world.
To have any hope of controlling the market, US labs would have to rope in chipmaker Nvidia NVDA.O and tech giants like Microsoft MSFT.O and Amazon.com AMZN.O, as well as the Trump administration. That seems unlikely. Nvidia CEO Jensen Huang this week played down prospects for a co-ordinated pause and then interrupted an on-stage discussion to take a call from Trump, who declared that opposing AI development only helps China.
Cartels tend to be secret, so we cannot know for sure what is happening behind closed doors. But creating the Standard Oil of AI will require an even greater amount of knavery than the AI labs have so far displayed.
Chart of the week, podcasts, and parting shot
It’s been a big week for central banks. The U.S. Federal Reserve and Bank of Japan both hiked interest rates, following the European Central Bank’s move a week earlier. Only the Bank of England shifted in a slightly less hawkish direction. Higher official borrowing can do little to counter the inflationary effects of rising oil prices. Nonetheless, investors expect central banks to keep going. Jon Sindreu argues markets are getting carried away.
The average chief executive spent most of his or her career in an era of relative stability. Now geopolitics are rewriting the rules of power. On The Big View, I talked to the strategist Tina Fordham about the unexpected shocks companies should be preparing for – and why there will be no return to the old status quo.
The Iran conflict has flared up again and Brent Crude oil is back above $100 a barrel. Has the much-predicted energy supply shock finally arrived, or is this another hiccup that the world economy will take into its stride? Yawen Chen and Jon Sindreu joined Aimee Donnellan and Jonathan Guilford in the Viewsroom this week to unpack the debate.
Prediction markets have made their social and financial mark. Companies like Kalshi, Polymarket and Robinhood Markets HOOD.O have enabled coin-toss bets on anything from the winner of the BMW PGA golf championship to who will be Donald Trump’s next press secretary. Trading volumes are on track to exceed $1 trillion. Yet the industry faces big regulatory, financial and technological questions. Jeffrey Goldfarb dusts off his crystal ball.
Five things learned from Breakingviews this week
Half of American mortgage borrowers still pay less than 4% interest. (Rent control in disguise)
A data signal from a satellite is 48,000 times weaker than a signal from a telecom tower 1 mile away. (Elon Musk meets physics)
Over half of the A$200 billion lent by private credit funds in Australia went to the country’s property sector. (Cockroaches Down Under)
The Ukraine war added €1,000 to the average cost of a German-made car. (Margins are the victim)
African billionaire Aliko Dangote’s Nigerian refinery has been the biggest supplier of jet fuel to Europe for the past three months. (A well-timed IPO)
QQQ•
