In the late 90s — when the author of this article was just a child with a shiny new modem connection, which for God’s sake you were not allowed to switch on until after regular landline phone hours (that is 8 pm, in case anyone has forgotten) — the seed of what we know today as the dot-com bubble was planted.
The dot-com companies had several things in common. They all promised to change the world, and they had extremely high valuations because investors poured money into them in the hope that they would one day become profitable.
The third thing they had in common was that the dot-com companies’ business models in the 90s were still new, untested and, in most cases, completely unprofitable.
Promises to change the world. High valuations. Untested business models.
It could just as well be the 2023 AI companies we were talking about.
Several economists have also warned that the current AI boom risks turning into an AI bubble. Just like in the 90s, when things first had to go really, really wrong, before the internet companies finally took shape and got a handle on their business up through the 00s and 10s.
Are we heading for a new bubble, this time with artificial intelligence at the centre?
The Organisation for Economic Co-operation and Development (OECD), of which Denmark is also a member, wrote already in 2021 that investments in AI are growing at a “dramatic pace.” There may be something to that.
So far, 24 percent of this year’s venture capital investments have gone into AI companies. Yes, that is right, 24 percent.
These are investments such as the USD 10 billion Microsoft threw at OpenAI back in January, barely two months after the AI company went from being best known as yet another of Elon Musk’s obscure side projects to being the centre of the whole generative AI revolution.
Next followed Amazon’s investment of USD 4 billion in the American AI startup Anthropic, as well as NVIDIA’s several investments totalling USD 1.3 billion in Inflection AI, which has a somewhat vague vision of building a “personal AI for everyone.”
In addition to the high valuation, the AI companies — just like dot-com companies — promise to change the world with the new technology.
In retrospect, the dot-com companies were absolutely right. The Internet has changed the world. But back then, in the 90s, it was still somewhat unclear what that change would look like. Therefore, it did not go as quickly, nor were the changes nearly as all-encompassing as the dot-com companies’ own and self-serving predictions claimed they would be.
Once the dot-com bubble finally burst, trust in the new internet and tech industry was also irrevocably shaken, and as we know, it takes time to regain trust.
Likewise, today’s AI companies and gurus promise in vague terms that there is almost no aspect of everyday life and the world that artificial intelligence will not change. Be it work, education, or the very existence of humanity.
So far, the Danish education system has indeed been thoroughly shaken to its foundations when ChatGPT took over the country’s primary schools, high schools, and universities, but the reactions have been as different as there are different education programmes.
Some ban it, others embrace it, and most are bewildered. So, we do not yet know how artificial intelligence will change the education system, only that it is happening.
If we look at the labour market, it is still difficult to spot the changes. In any case, the predicted mass layoffs are yet to happen. The job statistics show neither any massacres nor significant shifts in job functions. And if we look at the existence of humanity itself, there is still no evidence that it is threatened by anything other than, of course, climate change.
It could all mean that we will soon hit another tech bubble, this time caused by artificial intelligence. In any case, there are many similarities between the ongoing AI hype and the dot-com bubble of the 90s. Neither commentators, economists nor the media have been slow to point that out.
But if we take two steps back and look at the development from the beginning, it is clear that there are also significant differences between the dot-com bubble and the AI boom.
First of all, investments in AI are actually turning around. No matter how you calculate it, overall global venture capital investment in AI companies fell last year. It was the first time in ten years that this had happened, according to OECD’s figures.
Although global investments in AI are falling again this year, there has been a serious slowdown in that fall, which we cannot deny. However, this can be attributed to one particular type of artificial intelligence, namely generative AI.
It turns out that 40 percent of venture capital investments in 2023 have gone to the development of generative AI. And, as we know, generative AI is quite expensive and difficult to develop, so it is primarily the tech giants who can afford to pour money into it.
At the same time, the stock price increases are primarily observed in the American mega-cap tech stocks — i.e. the tech giants’ stocks, as Matthew Bartolini, managing director at State Street Global Advisors, says to CNN:
“You also have a pretty narrow, myopic marketplace in terms of companies leading on AI. A lot of major tech stocks have done well, but smaller companies haven’t. In a bubble, you’d see a significant amount of stocks throughout the area benefiting.”
And that is not the case, at least not right now. But Matt Bartolini also emphasises that this does not mean that an AI bubble cannot form:
“What is happening is that AI has been around for a really long time, and now there’s a branding bubble, so to speak, where it’s the subject of a lot of press and general conversations around the dinner table. But I don’t think from a stock perspective we’ve reached a full scale bubble.”
At the same time, after a year of AI bonanza, it must be noted that virtually no one has actually presented a well-functioning business model for generative AI. In that way, the AI hype and the dot-com bubble are absolutely similar.
Microsoft-owned company GitHub has recently announced that it has an annual revenue of USD 100 million from the coding assistant Copilot, but apart from that, no one has managed to make a profit out of chatbots, image generators, and the like. But generative AI is not all artificial intelligence.
We must not forget that other less glamorous types of AI — typically just called “machine learning” — have for years supported our search engines, diagnostics, logistics, digital security, etc. It is not damn sexy, but there is damn good business in it.
Of course, there will also be some smaller AI companies in the mix if it turns out that generative AI is the new blockchain — practical enough for some things, but not at all as revolutionary as was predicted in 2018. And, of course, this will hurt the investors who have bet heavily on generative AI.
But things will be fine. Financial markets are unlikely to collapse. Nor is generative AI that important to the world economy.
So long that it is just the tech giants with diversified business portfolios that are trying their hand at generative AI. Amazon, Microsoft, Meta, and Google have their basic business models under control — among other things through the old “unsexy” types of AI — and they can afford to pour money into the chase for the next big tech phenomenon.
So, things will be fine. As long as the “branding bubble” does not cause us to collectively start panic-investing in any company that adds the word “AI” to their marketing materials, because we just have to and must go in that direction at all costs.
