Many economists believe that generative artificial intelligence (AI) is about to transform the global economy. According to an article published last year by Ege Erdil and Tamay Besiroglu, from the research company Epoch, an “explosive growth” in which GDP skyrockets will be “plausible with artificial intelligence capable of largely replacing human labor.” Erik Brynjolfsson of Stanford University has said that he estimates that AI will “drive a productivity boom in the coming years.”
For such an economic transformation to occur, companies will need to spend heavily on new software, communications, factories and equipment to enable AI to fit into their production processes. The investment boom was necessary to allow previous technological advances (such as the tractor or the personal computer) to spread throughout the economy. From 1992 to 1999, U.S. nonresidential investment soared by 3% of GDP, for example, driven largely by additional spending on information technologies. So far, however, there is little sign of a big AI splurge. Around the world, capital expenditures (or capex) by companies are remarkably low.
After slow growth in the years before the Covid-19 pandemic, capex increased with the end of lockdowns. At the beginning of 2022, it increased at an annualized rate of around 8%. Technological optimism had taken over some companies, while others tried to strengthen their supply chains. Capex slowed somewhat later that year due to the effects of geopolitical uncertainty and rising interest rates. In March 2023, leading up to the launch of OpenAI’s ChatGPT-4, global capex spending grew at an annualized rate of around 3%.
Today, some companies are raising them again to take advantage of what they see as a huge opportunity in the AI ??space. Forecasts estimate that Microsoft’s spending (including R&D) will likely increase by about 20% this year. Nvidia’s will skyrocket more than 30%. “AI will be our largest area of ??investment in 2024, both in engineering and computing resources,” Mark Zuckerberg, the owner of Meta, announced late last year.
Elsewhere, however, plans are more modest. If we exclude the companies (like Microsoft and Nvidia) driving the AI ??revolution, those in the S
One would think that, given the great enthusiasm generated by the potential of generative AI, spending on information technologies is skyrocketing. Well no. In the third quarter of 2023, investment by US companies in “computer equipment and programs for information processing” fell 0.4% year-on-year.
Similar trends are observed globally. According to national accounts data (up to the third quarter of 2023) from the OECD, a club of mostly rich countries, investment spending (including public investment) is growing more slowly than in the years before the pandemic. High-frequency measurements of global capex from JPMorgan Chase point to minimal growth. With weak capex, it is not surprising that there is little sign of productivity improvements, according to a real-time measure derived from surveys of purchasing managers.
In Japan, an official survey does point to much greater growth in capex in the future, after years of lethargy. However, the study is likely to reflect factors specific to that country, such as corporate governance reforms. In most places outside the United States, the situation is considerably less encouraging. In Europe, the worsening economic outlook is no help. The investment intentions of European Union service companies are not even half as ambitious as they were at the beginning of 2022. British companies plan to increase capex by just 3% over the next year, compared to to 10% when asked in early 2022.
These trends point to one of the following two possibilities. The first is that generative AI is what in poker is called a dirty straight. Big tech companies love technology, but they’re going to have to work hard to find customers for the products and services they’ve spent tens of billions of dollars creating. It would not be the first time in recent history that technologists overestimate the demand for innovations. Let’s think about cryptocurrencies and the metaverse.
The second interpretation is less bleak, and more probable. Adopting new mainstream technologies often takes time. Let’s go back to the personal computer example. Although Microsoft released a novel operating system in 1995, American companies did not increase spending on software until the end of the decade. A Goldman Sachs analysis indicates that while only 5% of CEOs expect AI to have a “significant impact” on their business within one to two years, 65% think it will have an impact in the next three to five years. AI is likely to change the economy, but with a grunt, not a roar.
