Invest1 publisher2 min readPublished
Ackman and Wood are measuring different ends of AI spending in their Fed rate dispute
Cathie Wood cites 99.99% annual falls in AI inference costs against Bill Ackman's warning that the Fed's September hike won't slow AI spending. Her figures price AI's output and his worry is demand for its inputs, so both can hold at once.
The Investor · Invest desk
Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction
What happened
- Bill Ackman posted on X on September 25, 2026 that the Fed's 25 basis point hike may have been a mistake.
- Wood replied on September 29 that rising rates reflect genuine real yields and stronger-than-expected growth, not inflation pressure.
- At ARK's October "In The Know" session, Wood said the 10-year Treasury yield sits at its median dating back to 1790.
- Wood cited OpenAI's revenue run rate climbing from $20 billion to $70 billion as evidence that cheaper AI is unlocking much bigger usage.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- contradiction Wood's demand evidence is AI spending that keeps rising, the same condition Ackman's inflation case starts from, so her numbers leave his premise intact.
- cost If AI spending shrugs off the hike, other borrowers absorb the tightening while inflation stays intact, so sectors outside AI pay for a policy aimed at prices.
- exposure With 90% of global data center financing directed to the US, Fed rate decisions set the borrowing cost for almost the whole AI build-out.
Ackman and Wood are talking about two different prices. His worry, as Crypto Briefing summarizes it, is demand for AI resources [2]. If buyers keep paying regardless of cost, higher borrowing costs might not slow them, and in his framing the spending could feed a self-perpetuating inflationary spiral [2]. Her sharpest figure is about the output: inference costs falling 99.99% a year at constant performance, she said [5]. Taken at face value, the same work costs one ten-thousandth of what it did a year earlier [2].
Her demand evidence is OpenAI's revenue run rate, up from $20 billion to $70 billion [6], a 3.5-fold rise [1]. The account does not say over what period. If the climb and the cost decline covered the same year, and OpenAI's prices fell in line with Wood's cost figure, the quantity of constant-performance inference sold would have grown about 35,000 times [3]. Spending that rises while unit prices collapse is elastic demand for the product. It is also an extra $50 billion a year going into AI [4], and Ackman's argument is that a 25 basis point hike leaves spending of that kind untouched [1].
If AI building slows in the quarters after the September hike [1], Ackman's premise fails and the Fed's brake works on AI as it does on other capital spending. If inflation stays contained while policymakers keep tightening, Wood's thesis gains ground, by Crypto Briefing's reading [10]. In the third case both are right. Output prices fall as Wood describes, while demand for the inputs ignores the rate as Ackman describes [2] [5].
I think the third case is the likeliest, because none of Wood's figures measures how AI buyers respond to the cost of debt. A clear slowdown in data center financing after September would prove that wrong. The counter-thesis is hers, and it has data. US money supply is growing about 5.7% without producing higher inflation, according to Crypto Briefing [8]. Wood said the likely result of cheaper AI is "benign deflation" [7], with prices falling because things get cheaper to produce.
What to watch
- OpenAI's next disclosed run rate and the window behind the climb from $20 billion, to check usage growth against Wood's 99.99% cost figure.
- Whether falling AI inference costs appear in broad US price data, the condition Wood's "benign deflation" call needs.