A Former CIA Advisor Says This Tech Boom Is Already Bigger Than the Internet Ever Was
Washington, D.C., June 28, 2026 (GLOBE NEWSWIRE) -- During the dot-com boom, telecom companies spent nearly half a trillion dollars laying 80 million miles of fiber-optic cable across the country, enough to circle the globe more than 3,200 times. An estimated 85% of it would eventually go unused.
Today, it's estimated that nearly $5 trillion will be spent on AI data centers in the U.S. alone.
According to financial researcher Jim Rickards, that comparison is just the beginning. In a new free presentation, Rickards explores why some analysts believe the AI buildout may already be consuming more capital and energy than the internet boom that preceded it, and why he believes the differences between the two booms may matter just as much as the similarities.
The Largest Buildout in Decades?
The modern AI economy requires enormous physical infrastructure.
Massive data centers.
Specialized semiconductor facilities.
Energy generation.
Cooling systems.
Transmission networks.
Cloud infrastructure.
Unlike many previous software revolutions, AI's expansion requires significant investment in the physical world, which means it also runs into physical limits the internet boom never had to contend with.
According to George Noble, a former Peter Lynch protégé who ran the number one Fidelity fund in the U.S., the low-hanging fruit in AI development is already gone: every incremental improvement now requires exponentially more compute, more data centers, and more power, to the point where it may cost five times the energy and money to make these models only twice as good.
Tim Dettmers, a professor at Carnegie Mellon University and AI researcher, has gone further, arguing that rack-level hardware optimization will likely hit a physical wall as soon as 2026 or 2027. Venture capitalist Marc Andreessen has made a similar observation, noting that GPU capacity continues to increase while the resulting intelligence gains have started to flatten out.
Rickards argues that this distinction, a technology boom that is bound by physical and energy constraints rather than just code, may become one of the defining financial stories of the decade.
The Similarities—and Differences
The internet ultimately succeeded beyond almost anyone's expectations.
Rickards believes AI may follow a similar path.
However, he notes that successful technologies and successful investments are not always the same thing.
The late-1990s internet boom produced extraordinary innovation while also creating one of the largest stock-market bubbles in history, one built in part on a financing pattern Rickards has studied closely. Companies like Lucent Technologies, Nortel, and Cisco lent billions of dollars to cash-strapped customers, who then used that borrowed money to purchase the lenders' own equipment.
The sales were booked as revenue, creating what amounted to a feedback loop between borrowing and reported growth. At its peak, Lucent had nearly five million shareholders and was the most widely held stock in America. When the bubble burst, Lucent fell from $75 to $0.76 a share, Nortel fell from over $8,000 a share to around $50, and Cisco collapsed from $50 to $8.
Rickards believes a similar dynamic may be forming in AI today, with companies investing in start-ups that then purchase their chips and cloud capacity, creating circular revenue streams. Michael Burry, the investor who famously predicted the 2008 subprime mortgage collapse, has gone so far as to call one leading AI chipmaker the "Cisco of the AI boom."
Author and former Institute for Public Policy Research fellow Grace Blakeley has noted that the last time markets saw this level of circularity in the tech sector was the dot-com bubble, with one key difference: the amounts now being pumped into AI infrastructure dwarf what was spent laying fiber-optic cable in the 1990s.
The question facing investors today, according to Rickards, is whether current valuations remain realistic given both the scale of capital being deployed and the financing structures supporting it.
Why July 29th Could Matter
Rickards points to July 29th as a date investors may want to watch closely.
Major AI company, Meta is expected to provide updated earnings guidance around that time.
For Rickards, this report may offer valuable insight into whether growth, spending, and demand remain aligned with the assumptions supporting today's valuations. He points to a similar moment during the dot-com era: in March 2000, a single Barron's article highlighting how quickly internet companies were burning through cash helped trigger a decline that eventually sent the Nasdaq down nearly 80%, a slide that took 15 years to fully recover from.
Rickards believes a comparable string of disappointing AI earnings this summer could mark a similar turning point.
About the Presentation
Jim Rickards examines why some analysts are comparing the scale of today's AI spending to the internet boom and what that could mean for investors, in a free presentation available now. Click HERE to watch.
About Jim Rickards and Paradigm Press
Before becoming one of America's best-known financial authors, Jim Rickards advised government agencies on financial threats and crisis preparedness. His work included building financial threat-detection models for the CIA and consulting on market-risk scenarios for the Pentagon.
Paradigm Press is one of the most widely read independent financial research publishers in the United States, rated 4.8 stars on Google across more than 1,900 reviews. Free from advertiser influence, Paradigm Press is committed to helping everyday Americans understand the forces shaping their wealth.