Tech Brew Ride Home: Could AI Spending Blow Up The Economy? With Paul Kudrosky
Release Date: August 9, 2025
Host: Brian McCullough
Guest: Paul Kudrosky, Seed Stage Investor and Former Blogger at Infectious Greed
Introduction
In this insightful episode of Tech Brew Ride Home, host Brian McCullough welcomes Paul Kudrosky, a seasoned seed stage investor known for his influential blog, Infectious Greed. The discussion centers around Kudrosky's provocative analysis on AI capital expenditures (CapEx) and their potentially explosive impact on the U.S. economy.
The Impact of AI CapEx on GDP Growth
[00:05] Brian McCullough:
Brian introduces Paul Kudrosky, highlighting his significant but overlooked analysis on AI spending and its economic ramifications.
[00:45] Paul Kudrosky:
Paul acknowledges the introduction and begins to delve into his seminal piece, "Honey, AI CapEx is Eating the Economy." He explains his curiosity sparked by unusually high AI CapEx figures contrasting with a contracting GDP in Q1 of the year.
[01:58] Paul Kudrosky:
Paul breaks down the data, estimating that AI-related CapEx by tech giants like Meta, Google, Amazon, and Microsoft could account for approximately 0.7% of the U.S. GDP growth in the first quarter, potentially rising to 1.5% when considering ancillary expenditures. He emphasizes that these numbers suggest AI CapEx might be driving nearly half of the GDP growth during this period.
Notable Quote:
"We have decent numbers on something mysterious. This is what's unusual here..." ([01:58])
[07:58] Brian McCullough:
Brian reflects on the scale of AI CapEx, noting its dramatic increase—from a tenth of 1% of GDP a decade ago to nearly 1% today, doubling in the last four years.
[09:31] Paul Kudrosky:
Paul underscores the unprecedented surge in AI CapEx, describing it as a "fire hose of money" unleashed by major tech players and new entrants like private equity firms aiming to capitalize on AI-driven data center expansions.
The Sustainability and Risks of AI-Driven Data Centers
[12:15] Paul Kudrosky:
Drawing historical parallels, Paul compares current AI CapEx to the railroad and telecom booms. However, he highlights a critical difference: data centers are not enduring assets. The mean lifespan of GPU installations in these centers is roughly three years, necessitating frequent and costly upgrades.
Notable Quote:
"The mean life of a data center's GPU install is about three years. So this is very different in terms of a perishable good." ([13:18])
[16:20] Paul Kudrosky:
Paul discusses the declining returns on investment for data centers. Initially, returns were substantial (50% return on a 14% cost of capital), but they have since plummeted to margins barely above the cost of capital, making future profitability questionable.
Notable Quote:
"We're rapidly hitting the point where we're selling it... only marginally above the cost of goods sold." ([17:09])
Capital Sources Fueling the AI Boom
[18:16] Brian McCullough:
Brian shifts the conversation to the diverse capital sources driving AI CapEx, including internal cash flows from tech giants, venture capital, private equity, and notably, a significant rise in debt financing.
[19:14] Paul Kudrosky:
Paul elaborates on the role of private credit and Special Purpose Vehicles (SPVs) in masking the true extent of debt being funneled into AI data centers. He warns of the risks associated with obscured debt structures, drawing parallels to the financial crisis's CDOs.
Notable Quote:
"Private credit institutions are the fastest movers in this area because they don't have the same obligations with respect to meeting some of the financial security things." ([19:14])
[21:35] Brian McCullough:
Brian connects this capital influx to the broader shift from equity to debt markets, referencing a 70% year-over-year increase in debt issuance by tech firms.
Potential Economic Fallout
[22:01] Paul Kudrosky:
Paul outlines four primary risks if AI-driven CapEx were to collapse:
- Economic Contraction: Removal of AI CapEx could halve GDP growth, exacerbating fiscal challenges.
- Obfuscated Debt Risks: Similar to the 2008 financial crisis, complex debt structures could lead to a systemic crisis.
- Price Collapse: Excessive GPU rental supply could crash prices, undermining data center viability.
- Sectoral Disruption: Rapid AI adoption could lead to significant job displacement across various sectors.
Notable Quote:
"If you take away this component of GDP growth, that's calamitous." ([22:01])
[26:04] Paul Kudrosky:
Highlighting Goldman Sachs' optimistic job replacement models, Paul criticizes their oversight of the impending GPU rental price crash, arguing it would render such models naive.
Diversion of Capital and Its Consequences
[34:25] Paul Kudrosky:
Paul warns that massive capital allocation towards AI CapEx diverts resources from other critical sectors like manufacturing. He draws historical parallels to the decline of U.S. manufacturing due to similar capital misallocations during past tech booms.
Notable Quote:
"The money comes from somewhere. Who's not getting money? Capital drying up for manufacturing because it was diverted into being massive capex on fiber and other things." ([34:27])
Investment Perspectives on AI
[35:33] Paul Kudrosky:
Discussing his investment strategy, Paul reveals a cautious yet selective approach to AI. He favors investments where AI adoption becomes legally or operationally necessary, such as in medicine, while expressing indifference towards sectors where AI adds little tangible value.
Notable Quote:
"Where it's obvious that not using AI will be a lot of liability... that's where we see the most promising investments." ([35:20])
[37:03] Paul Kudrosky:
On a personal note, Paul shares his limited use of AI, mainly leveraging it for technical troubleshooting, emphasizing his preference for AI applications with definitive, reliable outcomes.
Closing Remarks
In the final segments, Brian and Paul discuss the relentless capital flow into AI despite the mounting risks. Paul cautions that without a macroeconomic trigger, the current AI-driven CapEx boom may continue until systemic vulnerabilities manifest. He underscores the importance of recognizing and understanding these economic drivers to prevent misguided policy decisions.
Notable Quote:
"If you don't understand the levers of growth, you make terrible decisions and you need to understand how your economy works." ([07:58])
Conclusion
Paul Kudrosky's analysis offers a stark warning about the unsustainable surge in AI CapEx and its hidden risks. As AI continues to reshape the economic landscape, understanding the intricate dynamics of capital allocation and its broader implications becomes crucial for investors, policymakers, and stakeholders alike.
For more insights from Paul Kudrosky, visit paulkedrosky.com.