Wharton Just Revealed AI Is the Biggest Waste of Money in History

Summary

The video argues that the AI investment boom poses a risk to the wider economy because spending and market gains are highly concentrated among a few large technology companies. It emphasizes that a correction is not necessarily imminent: historical industry booms have often ended in sharp declines, but not always. The central concern is that debt-financed investment and concentrated exposure could spread losses well beyond the AI sector if expected productivity gains fail to materialize.

AI Spending and the Economic Bet

• The video cites a Wharton/Point72 study by Jessica Wachter and Jonathan Wachter examining how much AI investment would need to deliver to justify its cost.
• Five U.S. technology firms reportedly spent $381 billion on AI infrastructure in 2025, with spending projected to be about twice as high in 2026. The study also considers nearly $1.1 trillion in spending by 2027.
• The video compares this with roughly $121 billion invested in internet fiber during the dot-com boom, arguing that AI spending is two to three times larger after accounting for the modern economy. It says AI investment could approach one-fifth of total U.S. investment next year.
• Morgan Stanley is cited as estimating that more than half of $2.9 trillion in spending through 2028 will come from external capital. The video warns that greater reliance on debt could make a downturn more damaging.
• It reports that Alphabet recorded its first cash deficit since Google went public in 2004.
• The study’s break-even case, as described in the video, requires roughly 2.7% productivity growth. The speaker says this would mean achieving in about three years gains comparable to those the U.S. received over a decade during the IT boom. The subtitles also describe this as a “2.7-fold increase,” an apparent inconsistency; the stated figure elsewhere is 2.7% growth.

Concentration and Market Risk

The video highlights several indicators of concentrated exposure:

• Data-center spending accounted for 92% of U.S. GDP growth in the first half of 2025, according to the figures presented.
• Stocks make up 48% of household financial assets, reportedly the highest share since 1945.
• Five stocks accounted for 93% of the S&P 500’s gains since July; the top 10 make up nearly 40% of the index, the highest share since the mid-1960s.
• The top 10% of households reportedly own 88% of shares and account for 45% of spending.

The speaker cautions that when a small group of companies and stocks drive a large share of investment and market performance, trouble in that group can affect the broader economy. In a downturn, assets that normally move independently may fall together.

Anthropic IPO and Valuation Concerns

• The video says leaked IPO-prospectus details put Anthropic’s prospective valuation at about $2 trillion.
• Based on the prior year’s revenue, the speaker says that valuation would take 435 years of revenue—not profit—to recoup.
• About 47% of Anthropic’s revenue reportedly came through Amazon and Google, which are also investors in the company and providers of its computing capacity.
• At a $2 trillion valuation, the video estimates Anthropic would trade at roughly 100 times revenue. It also cites Jay Ritter’s research that IPO companies with annualized revenue multiples above 40 have historically underperformed the market by 58.5% over the following three years.
• The speaker interprets the reported IPO timing as a possible opportunity for early investors and insiders to reduce risk, rather than simply raise development capital. This is presented as an interpretation, not an established motive.
• The video compares the current IPO environment with 2007 and warns that investors who buy late in a cycle may wait years for a recovery.

Bubble Framework and Historical Context

The video presents two historical lenses:

1. Harvard economists’ bubble-industry analysis: They examined industries whose stocks doubled within two years and substantially outperformed the overall market. In about half of those cases, the industry subsequently fell 40% or more within two years. The video stresses that a boom does not make a crash certain or immediate.
2. Carlota Perez’s four-stage technology-cycle model:
   • Eruption: New technology emerges and early investors enter.
   • Hype: Large amounts of capital flow in and asset prices move ahead of the technology’s demonstrated returns.
   • Turning point: Hype fades and the market reprices.
   • Synergy and maturity: The technology spreads through the economy and delivers broader benefits.

The speaker says AI appears to be in the hype phase and that the gap between hype and broad economic adoption is where short-term financial damage can occur.

Scenarios, Triggers, and Risks

• The Wharton/Point72 study considers scenarios in which AI makes no further major leap, makes one, or makes two; the two-leap case is described as a “singularity” scenario.
• The video says industry leaders’ expectations are at or above the study’s most optimistic scenarios, while the speaker argues there is not yet evidence to support such rapid gains.
• Possible triggers for a market repricing include a delayed Anthropic IPO, a loan default, or large customers abandoning advanced AI models. Meta and Microsoft are said to be reducing some cloud-service commitments.
• The video invokes the historical pattern that bubbles can be set off by a triggering event that synchronizes investor decisions and prompts simultaneous selling.
• It links the potential risk more to debt and concentration—drawing comparisons with 1929 and 2008—than to the 2000 dot-com crash. The video says a correction could create opportunities, while noting that advanced language models may remain important even if today’s leading companies do not.

Investing Implications and Cautions

• The speaker says he believes a correction will occur but does not predict when. He notes that the S&P 500 is at a record high and that the broader market may not be pricing in the risks he describes.
• He raises—but does not prescribe—moving money into Treasury bills or gold, and notes the difficulty of finding investments with uncorrelated returns.
• Citing Ray Dalio, the video says a portfolio with 15–20 good streams of uncorrelated returns could substantially reduce risk without lowering expected returns. This is presented as a demanding diversification approach, not a specific allocation recommendation.
• The practical question the speaker leaves viewers with is whether they can remain patient asset holders or might be forced to sell during a downturn. He cautions against being forced to act on someone else’s deadline.
• The video does not give a specific portfolio allocation or explicit buy/sell instruction. It also contains no explicit “not financial advice” disclaimer.

Assets, Companies, Sectors, and Instruments Mentioned

• Index: S&P 500
• Potential diversifiers/instruments: Treasury bills, gold
• Companies: Anthropic, Amazon, Google/Alphabet, OpenAI, Meta, Microsoft
• Sector/theme: AI infrastructure, data centers, cloud computing, chipmakers/AI hardware, technology IPOs
• Other references: stocks, debt, external capital, IPOs, cloud-service commitments

Disclosure

The presenter says the video was created with support from The Sovereign Professional and includes a promotional segment for the product. The presenter claims it can offer a return on investment exceeding its cost and describes a 14-day trial period.

Presenters and Sources

Presenter: Brendan Dell.

Sources and people cited or discussed: Jessica Wachter and Jonathan Wachter (Wharton/Point72 study); Jay Ritter; Paul Kedrosky; Daron Acemoglu; Dario Amodei; Ilya Sutskever; Elon Musk; Carlota Perez; Ray Dalio; Charles Kindleberger; Harvard economists; Princeton economists; MIT Technology Review.