When Matt Deitke, a computer scientist who had just started his own AI company, got a job offer from Meta, he turned it down. Meta doubled the offer and last month Deitke joined the company for a $250-million, four-year deal. Not bad for a 24-year-old.
Most of the discussion about the economic effects of AI concerns the fear that it will cause mass unemployment. I don’t think that’s going to be the problem. During previous periods of innovation, the overall demand for labour has risen rather than fallen. Technological progress creates new jobs that we hadn’t dreamt of, and higher productivity enables us to pay for things we didn’t know we needed, like nail bars and gyms. The latest revolution will similarly create previously unimagined industries, and we could do with more workers in sectors such as health, social care and education.
The issue is not whether there are jobs to be done but how they are to be paid for when the tech giants are cornering more and more of the value the economy creates. Concern about this issue is rising up the agenda and uniting such disparate institutions as the Trades Union Congress and the International Monetary Fund.
The central problem with tech is that it tends towards monopoly. The tech giants have vast stores of data that allow them to tweak their algorithms and keep one step ahead of any competition; they have cutting-edge expertise that insurgents struggle to emulate and huge piles of capital. The only company that has recently broken into the market is TikTok, which had the advantage of a captive Chinese market in which American companies can’t compete.
The lack of competition leads to huge profit margins. Nvidia, which designs chips for the AI industry, this week reported a 72 per cent profit margin on revenues that were up by 56 per cent in the second quarter of its financial year. It has just become the first company to be worth $4 trillion; only seven years ago, Apple became the first $1 trillion company.
Monopoly means that the wealth the tech industry produces stays, in large part, in the tech industry. It’s going to bosses, shareholders and a very small number of employees like Deitke who have the skills needed to improve AI.
Monopoly is not a new problem. In the past, society has mitigated its consequences through regulation and taxation. The US government brought down the robber barons in the Gilded Age of the late 19th century by passing anti-trust legislation to limit the power of firms such as JD Rockefeller’s Standard Oil, and in the 1980s to break up the telephone monopoly. Taxation of both share ownership and company profits ensures that a slice of the wealth a company creates goes into the public purse.
But this time is different, because the tech companies aren’t tethered to particular places in the way that the oil and phone companies were. They make their money in the cloud — from data centres anywhere and everywhere. That has allowed them to shop around the world to find the lowest tax rate available. And the only government that they are subject to is that of the country where they are headquartered, and which is also their biggest market — the US. Other states have little power over them.
In order to stop the tech giants from playing governments off against each other, the OECD club of rich countries agreed among themselves in 2021 on a minimum tax rate of 15 per cent on corporate income. But Donald Trump has reneged on that agreement and is also impeding other governments’ efforts to regulate the tech giants.
“Digital Taxes, Digital Services Legislation, and Digital Markets Regulations are all designed to harm, or discriminate against, American Technology,” Trump posted on his Truth Social platform on Tuesday. This is a reference to Europe’s Digital Services and Digital Markets Acts, which attempt to regulate tech firms, and to the digital services taxes that some governments have imposed. Given that the trade deal between the US and the EU is not finalised, European officials regard it as a threat to use tariffs to promote the tech companies’ interests.
This imbalance of power between the US government and tech giants on the one hand, and the governments of the rest of the world on the other, is already showing up as a growing prosperity gap. Twenty years ago, GDP per person in the US was about the same as in Britain. Now it is more than half as high again. Much of the disparity is down to the tech industry, which is reckoned to have been responsible for about a third of American growth in the past decade. This year, the value of data centre construction in the US is expected to be higher than that of investment in offices. Trump is doing his best to mess up the American economy but its dynamism seems to be counteracting his efforts; and, if AI goes anywhere near to fulfilling expectations, the size of the gap between the US and the rest of the world is likely to increase.
Although the roots of this problem lie in technology and economics, its consequences are political, and any solution to it must be too. Yanis Varoufakis, a former Greek finance minister who coined the word “technofeudalism” to describe the relationship between tech giants and people, proposes collective ownership of companies by workers, and thus the end of capitalism. Tony Curzon Price, an economist who used to work at No 10, takes a less radical approach: he has created the First International Data Union, to allow people to take control of their data. More ideas will, no doubt, emerge. But for any solution to work, people need to be aware of the problem. Stop scrolling, sheeple!