The AI infrastructure boom offers a timely way to ask a deceptively simple finance question: What is an asset worth when its future payoff is uncertain, but everyone is watching everyone else? The Bank for International Settlements (BIS) reports that the five largest hyperscalers are set to spend more than $1 trillion on AI-related capital expenditure from 2025 through 2026. The same report says these commitments are outpacing earnings and free cash flow, with some firms issuing debt to finance the race. [1]
That combination does not prove that AI assets are in a bubble. It does create a useful classroom setting for separating two ideas that are often blended together: value based on expected future cash flows and price based on beliefs about future resale. The BIS’s January 2026 bulletin makes the tension explicit. AI investment is surging, firms will need to shift some financing from operating cash flow toward debt, and the sustainability of the boom depends on companies meeting high earnings expectations. It also notes that equity prices have run far ahead of debt-market pricing. [2]
A controlled market for speculative demand
MobLab’s Asset Market Game gives students a clean laboratory for this distinction. Players begin with cash and shares in a company. Each share pays a random per-period dividend, and students buy or sell shares in a competitive double auction during each round. [3] In the instructor guide’s default configuration, there are ten periods, the dividend is either $1.00 or $0.40 with equal probability, cash earns no interest, and the asset has no terminal value. The expected dividend is $0.70, so a risk-neutral trader values the asset at $7.00 at the start of the first period. If the high dividend occurs every time, the maximum dividend payoff is $10.00.
A transaction above that $10.00 maximum cannot be justified by the asset’s own dividends. It requires a belief that another trader will later pay even more. That is the game’s precise definition of speculative demand: a price supported by expected resale rather than by the asset’s maximum payoff.
Bring the 2026 case into the debrief
After students play, display the transaction-price graph and ask which trades can be explained by expected dividends alone, when a buyer needs a resale story, and what would make that resale story fail.
The BIS Annual Economic Report warns that competitive pressure may lead AI firms to over-commit resources to projects with uncertain returns. If returns disappoint, financing could pull back and turn an investment boom into a prolonged investment bust.[1] The IMF’s October 2025 Global Financial Stability Report similarly says valuations had returned to stretched levels, valuation models placed risk-asset prices well above fundamentals, and a sharp decline in asset prices could strain banks and open-ended funds.[4]
Close by stressing the model’s boundary. The Asset Market Game does not forecast AI prices or reproduce hyperscaler balance sheets. It makes one mechanism observable: even when payoff information is common knowledge, beliefs about other traders can move prices beyond fundamental benchmarks.
Ready to turn valuation theory into an observable market? Try MobLab’s Asset Market Game in your macroeconomics or finance course.
References
[1] BIS Annual Economic Report 2026: Progress and peril. [2] BIS Bulletin 120: Financing the AI boom: from cash flows to debt. [3] MobLab Asset Market (Bubbles & Crashes) Instructor Guide. [4] IMF Global Financial Stability Report, October 2025.