TLDR:
- Cathie Wood challenges Bill Ackman on inflation, arguing that stronger real growth and falling technology costs can explain higher rates.
- Bill Ackman questions whether higher borrowing costs can slow AI infrastructure investment when companies expect exceptionally large returns.
- Wood cites a 99.99% annual decline in AI inference costs at constant performance, presenting the figure as support for benign deflation.
- Wood links cheaper AI to expanding usage, citing OpenAI revenue run rate growth from $20 billion to $70 billion in her economic argument.
The ARK Invest CEO, Cathie Wood, has challenged Bill Ackman over whether artificial intelligence will fuel lasting inflation. She argues falling AI inference costs could support stronger economic growth while limiting pressure on consumer prices. Her response follows concerns that Federal Reserve rate increases could fail to restrain continued investment in computing infrastructure.
Wood also places the 10-year Treasury yield near its historical median, citing records dating to 1790. The disagreement highlights how cheaper technology and heavy construction spending could push inflation in different directions. Both investors question conventional assumptions, but they offer sharply different readings of the economic consequences.
Cathie Wood Challenges the Inflation Case Behind Rate Hikes
The Federal Reserve raised its benchmark rate by 25 basis points on September 16, 2026. That increased the target range to 3.75%–4.00%, following a unanimous vote. Officials said inflation remained elevated and reaffirmed their commitment to a 2% target.
Bill Ackman questioned that decision in a September 25 post on X. He suggested demand for computing capacity and energy could remain strong despite more expensive borrowing. Companies pursuing major AI breakthroughs may keep investing because they expect unusually large returns.
His concern centers on financing costs becoming embedded in goods and services. If AI investment stays resilient, tighter policy could raise costs without reducing that spending sufficiently. He warned that this could create a cycle of rising costs and further rate increases.
Cathie Wood offered a different interpretation on September 29, pointing to real yields and growth exceeding expectations. Real yields represent returns after adjusting for inflation. Under her argument, higher rates can reflect stronger economic prospects while productivity improvements help contain prices.
The Fed also reported solid economic expansion, strong productivity growth, and robust capital investment in September. Its statement still identified elevated inflation as a continuing problem. This combination suggests efficiency gains can coexist with price pressures while companies build infrastructure and expand their operations.
For Cathie Wood, the historical yield comparison provides context for evaluating those increases. However, a historical median alone cannot establish whether current policy is restrictive. Inflation expectations, borrowing conditions, and productivity trends also matter when assessing the economic effects of higher rates.
Falling AI Inference Costs Reshape the Inflation Debate
In the October edition of ARK’s In The Know, Cathie Wood highlighted sharply falling technology costs. She cited a 99.99% annual decline in AI inference costs at a constant performance level. Inference refers to running a trained model to generate outputs, including answers and predictions.
That estimate compares the expense of delivering similar capabilities as models and computing systems improve. Its broader economic impact depends on how widely businesses adopt those efficiencies. Potential savings could support automation, lower service costs, and expand access to tools previously considered expensive.
Cathie Wood linked those changes to OpenAI’s revenue run rate rising from $20 billion to $70 billion. She presented that increase as evidence that cheaper AI can encourage substantially greater usage. A run rate expresses revenue on an annualized basis, rather than a completed year of reported sales.
She describes the potential result as benign deflation, with productivity gains supporting output as production costs fall. That differs from falling prices caused by weakening demand. Yet cheaper inference can coexist with expensive electricity, land, and construction during a rapid infrastructure buildout.
Bill Ackman’s concern involves demand for data centers and other physical infrastructure that support AI services. Cheaper inference could increase usage, adding pressure to electricity supplies and computing capacity. That makes the speed of new supply relevant alongside the pace of technological improvement.
Cathie Wood also cited US money supply growth near 5.7%, arguing it had not triggered renewed inflation. ARK’s briefing placed 90% of global data center financing in the United States.