A Demand-Bottleneck Model Raises Questions About Bitcoin’s AI Bull Case

Bitwise Chief Investment Officer Matt Hougan's bullish Bitcoin argument links AI productivity gains to scarce-asset demand, while a demand-bottleneck model projects that investment-heavy growth can shrink consumption's share and eventually slow innovation.

In an Oct. 14 memo, Hougan presents AI-driven abundance, a bust after heavy AI investment followed by policy intervention, and moderate growth as paths that all support Bitcoin (BTC). His rationale links productivity gains with demand for scarce assets, a downturn with monetary expansion, and steady growth with continued adoption.

The demand-bottleneck scenario describes automation requiring more capital and equipment. As output shifts toward investment, consumption becomes a smaller share of the economy. The model ties returns on innovation to consumer markets, so a smaller market can eventually make new AI research unprofitable.

The scenario starts with consumption at 69% of gross domestic product (GDP) and sets 62.1% as the threshold for new AI research entry. These figures describe consumption's share of output; the scenario does not require total consumer spending in dollars to fall.

Trend growth settles at about 1.6%, below the 2% pre-AI baseline. Output growth peaks just below 2.4% about 15 years after AI activation, then falls below the baseline after about 35 years. These are model projections measured from AI activation, not calendar-year forecasts.

Real wages per worker settle about 15% below the no-AI path in the demand-bottleneck scenario. The estimate assumes an unchanged workforce, no newly emerging tasks for workers, and compensation linked to each worker’s marginal productivity. Across the paper's automation scenarios, labor's share of tasks falls from 60% to between 15% and 50% over three decades. A separate transformative-AI scenario reaches 5.5% growth after 20 years and continues rising, illustrating the range of modeled outcomes.

U.S. labor-market observations offer an early but inconclusive comparison. Contacts in five of 12 Federal Reserve districts described firms reducing staffing through attrition, delaying hiring or expecting fewer jobs after AI-related productivity investment. Most districts reported little or no net change in employment; labor-market conditions overall were little changed. More than half of districts described higher-income households as driving consumer spending. The observations are qualitative and cover information collected on or before Oct. 5.

The model does not study Bitcoin or assess how AI-related wages and consumer-market changes affect BTC demand, ownership or price. One possible connection is that wealth concentrated among investors could support demand for Bitcoin while weaker worker wages limit broader purchasing power. That is an inference, not a result measured by the model.