In my lifetime, besides blowing bubbles, I have lived through three economic periods called bubbles. Like the bubbles I shake and blow for my granddaughter, these floating spheres never last very long. The same can be said about economic bubbles. They get big and then burst. Did economic bubbles happen before the onset of capitalism? It turns out the answer is yes.
Historic Bubbles
Bubble episodes have been a part of human history going back to Imperial China, the Roman Empire, and other ancient civilizations. Wherever humans established exchange rates involving land, goods and services, speculation abounded.
In 33 CE, several Roman banks failed, as agricultural land values crashed. It would have been deja vu for any Romans who lived then if they were still around to witness the 2008 CE financial crisis, when bad paper fuelled by subprime mortgage-backed securities and credit default swaps caused the investment bank, Lehman Brothers, to fail and created a global recession.
Successive dynasties of the Chinese Empire were beset by economic cycles of boom and bust similar to recent historic bubbles, mostly arising in the Western World. These include the Dutch Bubble in the 1630s caused by speculation on the price of tulip bulbs, the British South Sea Bubble in 1720, the Railway bubble of the 19th century that culminated in the Gilded Age, the 1920s stock market bubble and crash of 1929 that caused the Great Depression, the Dot-com Internet bubble of the 1990s, and finally, the 2008 Subprime mortgage bubble.
Paul Krugman, who won the 2008 Nobel Prize in Economics, in a recent conversation, described a bubble as a situation where people invest “in something that has no realistic chance of paying off, not socially but just commercially, to an extent that justifies the amount of money being thrown at it.” He continued, stating, “A bubble is something that people do because everyone else is doing it.”
In some respects, a bubble is like a self-driving “Ponzi scheme,” a “money game” that preys on both sophisticated and unsophisticated investors.
The Current Bubble
Artificial intelligence (AI) is at the heart of the current bubble. AI companies like OpenAI, Anthropic, Meta, Google, Microsoft, and Xai, joined by chip makers like NVIDIA, have had investors throw billions of dollars at them. For the AIs, realizable profits are nonexistent despite hundreds of billions changing hands.
If history repeats, the AI bubble will be like all those that preceded it. The market will crash similarly to what occurred in the Dot-com bubble. A few companies will emerge as highly profitable. The rest will be left in the dust.
In other respects, however, the AI bubble may be different, because the product itself may influence the outcome. Currently, any company purporting to be doing AI is riding on the bubble membrane as it expands. But what if an AI tool could help distinguish between different AIs, helping to determine the difference between realizable value propositions and those companies that have unrealistic expectations and are overvalued?
In some sense, AI may be the only technology to mitigate the growing AI bubble, providing a valuable service to separate facts from hype and provide measurable data to support the billions currently being thrown at the technology.
If the Bubble Bursts
The hundreds of billions being invested in AI technology are largely going to build capacity. The bubble is the economic growth arising from infrastructure investments. The concentration of AI activity is in China and the United States, the competitive players in this race to the top. NVIDIA, centred in Taiwan, has seen enormous growth in value.
In terms of earnings, however, based on recent disclosures, it appears the main AI players like OpenAI and Xai are bleeding red ink. Google, Meta, and Microsoft continue to make money, but not from AI. Anthropic is making some from its AI coding tools.
Investor money in these companies is going into the building of data centres to increase AI computing capacity. If and when the AI bubble bursts, these data centres will remain with servers filled with NVIDIA chips that may eventually be used.
The analogy to the Dot.com bubble is the billions that investors gave companies to build fibre optic networks. When that bubble burst, the excess was denoted as dark fibre. Eventually, many years later, that fibre got lit. The same will likely happen with the data centres. But for the labour force engaged by AI companies and then let go, and for those whose jobs were replaced by nascent AI, the bad taste will linger for a long time.
