Bitwise CIO Matt Hougan's bullish cases assume AI abundance or policy easing ultimately supports Bitcoin. A BIS working paper models a different risk: productivity rises while wages and consumer demand lose ground.
In a Bank for International Settlements model, consumer spending falls from about 69% to 62.1% of output as AI investment expands. The shift eventually makes further innovation less profitable, even as businesses keep adding machinery and computing capacity.
That scenario exposes a gap in the bullish case for Bitcoin: output can rise while the purchasing power that supports demand weakens. The model does not forecast Bitcoin prices. Its results depend on assumptions about labor and wages.
Demand bottleneck
Published on Oct. 12, 2026, BIS Working Paper No. 1381, Transformative AI and demand bottlenecks, examines how weaker consumer demand could constrain AI progress. BIS economist Phurichai Rungcharoenkitkul describes a feedback loop: AI automates more tasks as investment takes a larger share of output. When consumption's share shrinks, the market for new products narrows. That makes further AI innovation less profitable. Rungcharoenkitkul's BIS working paper lays out scenarios ranging from transformative AI to a demand bottleneck, where thinning consumer demand stalls progress.
In the paper's calibrated scenario, long-run sustainable growth falls from a 2% pre-AI baseline to about 1.6%. Actual output growth peaks just under 2.4% around 15 years into the AI era. It remains above 2% for about 35 years. Those figures reflect a prolonged investment boom, not a guarantee that household demand keeps pace.
The paper also models a high-growth case. Output grows about 5.5% a year 20 years into the transition, then continues accelerating. Total consumption can still rise. But if output increases 10% while consumption's share drops from 69% to 62.1%, the consumer-market index slips from 69 to 68.31. That is a 1% contraction inside a larger economy.
Wages and rates
The model estimates labor's share of tasks falling from 60% to a range of 15% to 50% over three decades. Under the demand-bottleneck calibration, real wages settle about 15% below the path they would have followed without that degree of automation. The estimate assumes fixed labor supply. It also uses a static set of worker tasks, with wages tied to marginal productivity. Different assumptions could change the result materially.
The International Monetary Fund estimates that AI will affect about 40% of jobs worldwide and up to 60% in advanced economies. IMF Managing Director Kristalina Georgieva said the effects range from productivity support for workers to downgraded or eliminated jobs. The estimate measures exposure to AI; it does not predict that every affected job will disappear.
Interest rates move in different directions across the modeled transition. The natural rate may rise during the investment boom as demand for capital strengthens, then fall as the demand bottleneck takes hold. The paper's pre-AI reference is 1.65%. Its median estimate 20 years out is 1.2%, with a 90% model band from -0.6% to 4.4%.
That distinction matters to Bitcoin. Lower rates can support demand for risk assets during a healthy productivity transition. But falling rates alongside weak consumption can signal a different environment, especially when financial stress rises and credit contracts. Investors may then cut leverage and sell assets. The rate direction alone does not show which conditions are taking hold.
Signals in the US
The Fed's Oct. 14 Beige Book offers mixed early evidence. It does not confirm the BIS feedback loop. The qualitative survey gathered observations through Oct. 5. Businesses in five of 12 districts reported cutting headcount through attrition, delaying hiring or expecting fewer jobs once AI-related productivity gains arrived. Consumer spending was broadly flat. More than half of districts said higher-income households drove it.
Four districts described employment conditions as flat. The remaining eight split between modest increases in three districts and slight increases in five. The survey points to uneven purchasing power, but it does not establish that AI caused the pattern or that a long-term demand bottleneck has begun.
Trade and market data add context. The US trade deficit widened to $105.6 billion in August, driven by strong AI-hardware imports. Economist Dean Maki estimates that about 70% of AI-related capital spending went to imports. On Oct. 14, after a hot CPI print, CoreWeave fell about 16%. Nebius and Dell each dropped about 11%, while Oracle fell about 9%. AI infrastructure debt also faced elevated borrowing yields. The equity moves reflect expensive capital, a nearer-term concern than the paper's multi-decade demand mechanism.
Hougan's Oct. 14 memo lays out three favorable routes for Bitcoin. One depends on AI abundance making intelligence-based services cheaper, shifting value toward scarce assets. Another sees an AI investment bubble burst, bringing recession, policy intervention and monetary expansion that could increase Bitcoin demand. The third rests on modest productivity gains, continued ETF demand and higher institutional allocations. Hougan acknowledges interim drawdowns and risks beyond macro conditions. Each route ultimately assumes that abundance, instability or monetary expansion adds demand.
Bitcoin has a fixed supply, but its market price still depends on investor demand. Purchasing power and liquidity matter too. Weaker wage growth could leave households with less room to save and accumulate the asset. If gains accrue mainly to wealthy investors and institutions, demand may rely more heavily on that narrower group. Weaker consumption could eventually weigh on corporate earnings and market liquidity. These are transmission hypotheses based on the BIS findings; the paper itself examines growth and rates, not Bitcoin.
Concentrated investor wealth could also sustain Bitcoin demand if those receiving the gains choose to hold it. The net effect depends on whether that capital outweighs reduced purchasing power among households. Previous market episodes have shown how collateral constraints can amplify stress. A collateral squeeze account describes how slow transfers and split collateral can expose Bitcoin hedges to liquidation even when the strategy works.
The contrast to watch is whether productivity and real wages improve together, or investment expands as hiring and worker income weaken. Hougan's scarcity case has stronger support if consumption holds its share and ETF subscriptions continue across conditions. An investment boom accompanied by a falling consumption share and wider AI credit spreads would point to a thinner demand base. Stalled ETF flows during continued AI spending would be an early market signal.
Bitcoin's limited supply does not determine who can afford to buy it. Broad household income can support recurring savings. Concentrated wealth can sustain demand among a smaller investor base. Those channels may respond differently to interest rates and credit availability. The Oct. 27-28 Fed meeting can shape the near-term rate path.
