Everyone wants to know what AI is worth. One investor looked to the past for the $20 trillion answer.

• Ben Miller, the CEO of Fundrise, analyzed AI's potential and estimated its value at $20 trillion.
• Miller's team compared AI's growth with past tech waves, finding consistent growth rates.
• Miller predicted that Nvidia wouldn't be the only big recipient of AI profits for long.

The prevailing narrative surrounding AI is that the power and the opportunity it presents are unprecedented and seemingly limitless. And yet some investors in companies at the heart of AI infrastructure are nervously wondering when that power is going to turn into profit. Investors and academics regularly look to past moments of technological transition and find cautionary tales.

Jensen Huang, the CEO of Nvidia, is reportedly paranoid about following the fate of Cisco , which overshot the dot-com boom and suffered in the bust. Intel offers the opposite warning , having sat out one too many technological shifts.

In his quest to back winners in AI, Ben Miller, the CEO of the investment platform Fundrise, started by documenting the past.

"Everybody's constantly talking about the waves," Miller told Business Insider. "And I was wondering, does the financial data match the waves?"

He put together a lean team of analysts and set out to determine what AI is likely to be worth. After months of analysis, he found that generative AI isn't quite as unprecedented as having a conversation with Anthropic's Claude or generating a podcast from a document in minutes with NotebookLM may feel. In fact, the generative-AI wave looks a lot more like past tech waves than its revolutionary capabilities would suggest.

The Fundrise model (a human-made financial model, not an AI one) ingested more than 40 years' worth of data from 250 tech firms with at least $1 billion in market capitalization, from Adobe and Amazon to Yelp. AI powerhouses like Nvidia, AMD, Meta, and Microsoft were in the mix too.

Segments within those firms' revenues were then broken down by technology wave — PC, internet, mobile, and cloud — and by the part of the technology stack they represent between applications, hardware, and platforms.

"As an example, Microsoft was a large participant in the PC platform and application markets in the 1990s as well as in the current cloud platform and application markets," the analysts said. "We apply their revenue breakdown between platform and application and allocate those percentages of their market cap to the specific waves and tech layers."

The team measured the impact on total market capitalization of each wave compared with the previous one — and each time, the wave was roughly three times the size of the one before. The internet era generated 3.5 times as much value as the PC wave. Mobile technology generated 3.4 times as much value as the internet wave. Cloud computing generated three times as much value as mobile.

That "rule of three" theory puts the value of the AI wave at about $20 trillion.

"The thing that I found shocking is the growth rates are fairly consistent," Miller said.

Who gets the $20 trillion?

It's somewhat more difficult to answer exactly which companies will get the largest slice of the $20 trillion pie.

Hardware companies tend to get the earliest piece, but when the Fundrise analysts broke down the studied companies' returns by wave and their place within the stack, they found that the application layer at the top almost always got the biggest share.

Amazon, for example, would be classified as within the application layer of the internet stack, Instagram at the top of the mobile stack, and Salesforce at the top of the cloud stack.

Eventually, the players closer to the consumer get the most benefit out of each new wave. That's why, Miller said, this moment in AI feels so uncertain.

Miller likened the AI wave to the mobile wave in about 2009, when there were few applications for the iPhone. Miller argued that it's shortsighted to assume that the return on investment in AI isn't coming because it's not here yet.

"The lag is super consistent," he said.

Nvidia's share is consistent too. Fundrise estimated the GPU giant's 2030 market capitalization at $4 trillion and, working backward, found that Nvidia's total share of the AI boom would also value the entire wave at $20 trillion.

History says patience

We'll have to wait a decade or so to find out if Miller's team got it right, but they're not alone in the $20 trillion ballpark.

Analysts at the research firm IDC set out on a similar mission but with a very different approach. They used an input-output model, incorporating data from both the supply chain and the end-user side of AI across different geographies. They integrated that data into a single multiplier — an amount multiplied by every dollar spent building AI to estimate the total economic input. The method was different, and the figure they were aiming for was broader than total market capitalization, but they landed in the same neighborhood: $19.9 trillion in global economic impact.

Miller said his team's estimates were conservative if anything, since one of the biggest X factors is the computing power Nvidia and its frantic competitors are able to generate.

"Nvidia or GPUs are producing more compute faster than Moore's law ," Miller said. "So the question is if compute is the primary underlying driver of economic activity, economic growth, economic opportunity, then there's an argument that it's actually bigger."