AI's CapEx Driven Endgame - A Shareholder Crash, Not an Economic Crisis

AI's CapEx Driven Endgame - A Shareholder Crash, Not an Economic Crisis

Ian Harnett (Chief Investment Advisor of Absolute Strategy Research) argues that the massive surge in AI infrastructure spending—the buying of servers, chips, and the building of data centres —is one of the final, classic signals of a major technological bubble nearing its peak. This echoes the late-1990s Telecommunications, Media, and Technology (TMT) bubble, which also saw an enormous and often wasteful build-out of physical capacity like fibre optic networks.

However, Harnett maintains a crucial, nuanced distinction that is key to understanding both the risks to investors and the ultimate benefits to society: the role of equity-driven capex versus debt-driven capex.

Historical Bubbles and the Funding Timeline

• 1986–1991: Japanese Asset Price Bubble: Driven by Debt. Defaults crippled the banking system, leading to a severe, prolonged recession (the "Lost Decade").
• 1997–2000: Dot-com (TMT) Bubble: Driven by Equity. Losses were confined to shareholders, resulting in a short, mild recession (a contained market event).
• 2002–2008: US Housing Bubble: Driven by Debt (High Leverage). Widespread defaults on debt caused a systemic cascade, collapsing the banking sector and triggering the Global Financial Crisis (GFC) (a devastating economic crisis).
• Today: AI Bubble: Driven by Equity. The equity-funded capex suggests the consequences will be a market correction, not a systemic credit crisis.

AI Bubble Endgame - 2025

The Critical Difference: Equity vs. Debt Funding

Equity-Driven Capex (AI & Dot-com Era): The capex has been largely funded by equity (stock sales or retained earnings of highly valued companies).

• The Pain: When the bubble bursts, the losses are confined primarily to the shareholders as stock values plummet. Harnett warns of potential stock declines of 70%–80%—a painful fate for late investors.
• The Economic Benefit: Because there is no systemic cascade of defaults on debt, the broader economy is insulated. The bust is an equity event, not a banking or credit crisis.

Debt-Driven Capex (e.g., 2008 GFC): A bust driven by debt-funded capex is far more dangerous.

• The Pain: When debt cannot be repaid, it triggers defaults, bank failures, credit freezes, and a systemic economic crisis. The fallout spreads across the entire economy, leading to a much more severe and prolonged recession.

Cash Flow Impact: Equity vs. Debt

The choice of funding mechanism has critical implications for corporate cash flows

Debt is Cheaper, but Fixed: Debt is generally less expensive than equity, but it creates a mandatory, fixed obligation to pay interest and principal. If end-user cash flows decline, these fixed payments can quickly lead to default and bankruptcy, posing a systemic risk.

Equity is More Expensive, but Flexible: Equity, being capital provided by shareholders, is the most expensive form of capital, as investors expect the highest return. However, it offers cash flow flexibility because there is no mandatory repayment. When the bust comes, the companies dont face mandatory insolvency; they just stop investing, and shareholders bear the losses, keeping the economic consequences contained.

The Paradox of "Schumpeterian Waste"

The key difference with the AI bubble is that the over-investment, or "Schumpeterian Waste," is a necessary step toward the new technology's success. The bubble's hype lowers the cost of capital, allowing for the rapid deployment of massive AI compute power that would otherwise be impossible. When the bubble inevitably bursts, the cost of this new capacity is effectively written down. This cheap, abundant infrastructure becomes the foundation for AI's widespread and ultimate adoption across society. The good news for society is the guaranteed ubiquity of AI down the road; the bad news for late equity investors is that they are financially underwriting this progress and face steep losses when the cycle concludes.