The GenAI Boom Is Only Slightly Louder Than The Dot Com Boom

September 29, 2025 Timothy Prickett Morgan

In most markets, revenues and hopefully profits go from exponential growth to log growth to linear growth in the fullness of time. We are living in the exponential phase of AI, and for GenAI in particular, where the x in y=a^x is a pretty big number and the resulting revenue growth looks like the proverbial hockey stick.

The numbers can be a bit surreal, as they seemed back in the Dot Com boom, where we turned on mainstream stores into warehouses with delivery services. With AI, we are turning the digital data of our lives into data warehouses and creating an algorithm that sure feels like a kind of sentience that can do a lot of menial tasks performed by people, and some that are not so easy.

IT spending figures for the Dot Com boom are hard to isolate because the Y2K crisis, where the world’s applications had to be coded for four digit year designations instead of two digits because the millennia was incrementing, caused a big bump – somewhere around $300 billion to $600 billion – in spending according to estimates by Gartner. Companies doubled up their systems and often installed third party applications in 1998 and 1999 to solve the problem. This spending coincided with the rise of the commercial Internet, so we had an exponential and a big linear hitting at the same time.

In that Dot Com boom 1995 and 2000, according to this report, worldwide information and communications technology spending rose from $1.61 trillion to $2.33 trillion, a factor of 1.44X increase. If you take out the telecom spending, which was having a bubble of its own as we all got cell phones and ironically we needed more fiber optics and as the world was stitching Pangea back together with fiber optic cables and Internet technologies, then the core IT spending rose from $940 million in 1995 to $1.35 trillion in 2000, the growth rate was the same 1.44X of that period. Which stands to reason because the Dot Com boom was a Web infrastructure boom in the datacenter as much (but no more than) as it was a telecom boom.

We do not have precise breakdown spending of Web infrastructure spending versus other infrastructure for back office and remote office functions, which is a shame, but it is pretty safe to say that Web interfaces drove some back office system growth, but not necessarily an increase of revenues because of intense competition from IBM in the Unix market against Sun Microsystems and Hewlett Packard and because of the rise of X86 servers in the datacenter. That back office capacity growth was driven by an increasing number of people were given access to data in those systems as part of a Web transaction. And then, of course, there was spending on all of the Web infrastructure that wrapped around those back office systems to provide Web applications that started collecting telemetry on Web sites and users, leading to the Big Data boom a few years later. Of the $1.35 trillion in IT spending in 2000, it is not hard to imagine that $300 billion of that was purely for Webby stuff. That means Web infrastructure drove about 22 percent of IT spending in 2000, just before the Dot Com bubble and the Y2K bubble both burst. (And a year later, 9/11 happened and the world was definitely in a recession.)

Let’s fast forward to the GenAI boom and compare and contrast IT spending and AI spending. The GenAI boom started in earnest in late 2022 and is still ongoing. A new AI spending forecast from IDC shows data bookending 2004 and 2029, with $430 billion last year and $1.3 trillion four years from now. That is the same five year span as the Dot Com boom, but maybe we should start the comparison in 2021. In a separate report we dug up, you can calculate that IT spending in 2021, which you can think of as the steady state of machine learning and statistical AI spending before GenAI took off, was $362 billion. If you further take the compound annual growth rate and work backwards from 2029 to 2027, you can get a similar data point to the end of the Dot Com boom to compare that 2021 AI spending level to, and that works out to $747 billion in AI spending worldwide in 2027. That represents a factor of a little less than 2.1X growth for AI spending (across hardware, software, and services just like in the Dot Com spending above) across those six years.

So how is this GenAI boom driving IT spending overall, which is the only way we can infer a comparison to the Dot Com boom? In the most recent forecast, IDC hinted that by 2029, GenAI would drive more than 26 percent of total worldwide IT spending. If you do the reverse math on that, you get just shy of $5 trillion in overall IT spending in 2029.

That is a slightly richer mix for AI infrastructure in the GenAI boom than Web infrastructure was during the Dot Com boom, but not crazily so.

The other thing that is loud is the magnitude of the numbers. If you take the core IT spending numbers from IDC for 1995 ($940 million) and 2000 ($1.35 trillion), they inflation adjust to $1.98 trillion in 1995 and $2.51 trillion in 2000 in 2025 dollars. So IT spending overall, and in the broadest sense including telecom services, will have doubled in terms of real spending from 2000 to 2029.

And if you really want to reckon this growth in IT spending correctly, you have to measure it against global GDP. In 2000, Global GDP was around $33.7 trillion (we are averaging a bunch of tightly scattered estimates). If you inflation adjust this, this is $66.75 trillion in 2025 dollars, and inflation adjusted IT spending was 3.76 percent of that and inflation adjusted Web infrastructure spending (around $560 billion) represented was 0.84 percent. Less than a point. It sure felt like a lot more than that, didn’t it?

Let’s continue. There will be some inflation between here and 2029, of course, and the World Bank says take an average of 2.5 percent per year of global inflation for GDP between 2025 and 2030. So you have to shave that inflation off each year for the projections and for cumulatively between 2025 and 2029, you have to shave off 12.5 percent to get a figure in 2025 dollars. Thus, a projection of $137.8 trillion global GDP in 2029, which is in the ballpark of estimates we surveyed, spends more like $120.6 billion does today. And that 2029 AI spending forecast and IT spending forecast from IDC have to be shaved down to 2025 dollars as well. So that is $4.38 trillion in overall global IT spending in 2029 as expressed in 2025 dollars and $1.14 trillion in AI spending in 2029 in 2025 dollars. It is the same 26.3 percent ratio of AI spending to IT spending as in the unadjusted data, but IT spending will be 3.63 percent of GDP after both are inflation adjusted and that is a tenth of a point lower than the IT/GDP ratio in 2000. But inflation adjusted GenAI spending will represent 0.95 percent of overall GDP adjusted down to 2025 dollars.

Again, the GenAI boom in 2029 will only be a slightly higher component of a complex and vast economy. But it sure does feel bigger than this, doesn’t it?

This Issue Sponsored By

Table of Contents

Content archive

Recent Posts

Subscribe

Pages

Search