What the Fed fears most is not tariffs but unanchored inflation expectations
Chicago Fed President Goolsbee identifies the biggest risk as inflation: after five and a half years above target, the AI narrative may push rates up in the short term rather than down; his key advice for rate-sensitive industries is to watch the data, not the stock market.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
19 chairs are not for consensus but for argument
He explains why the FOMC should keep 12 regional Fed presidents plus 7 Board members, totaling 19 chairs: not for consensus, but to let different worldviews collide under high uncertainty. Goolsbee says colleagues can shout at him or change his interpretation of the data. This structure insulates rate decisions from the political cycle as much as possible, because staggered 14-year terms prevent anyone from replacing the whole table at once. The direct lesson for founders: when making major judgments, surround yourself with people who dare to challenge you, not an echo chamber.
— Austan GoolsbeeThe Fed's real fear is not price hikes but unanchored expectations
The real danger of tariff and oil shocks is not a one-time price increase but convincing people that high inflation will persist. Once such expectations form, workers demand 6% wages and firms price at 7%, forcing the Fed to use a deep recession to break them. Goolsbee calls this 'unanchoring of inflation expectations' and says it is the self-fulfilling mechanism he watches most. For entrepreneurs, this logic means: whether a cost increase is one-off matters more than how much costs rise.
— Austan GoolsbeeFive and a half years above target, the Fed's own record is not pretty
Goolsbee admits the Fed's performance has not been pretty: nearly six years, at least five and a half above the 2% target, with a recent reversal in the past year. He cites on-the-ground observations from the Seventh District (Midwest): businesses report sharply higher input costs, farmers say prices are not rising but costs are crushing, and everyone talks about affordability. His conclusion: with the public highly sensitive to prices, any new supply shock—tariffs, Middle East oil prices, chip shortages, AI data centers—is more likely to be interpreted as long-term inflation.
— Austan GoolsbeeThe new chair is reining in the Fed's forward guidance
New Chair Kevin Walsh is changing the Fed's communication style: he dislikes 'forward guidance'—explicit promises like 'if X happens, we will raise or cut rates.' Goolsbee observes that FOMC statements have become noticeably shorter, and Walsh has set up five working groups covering inflation, AI and productivity, the balance sheet, and other areas, staffed with prominent people he trusts. Reports are not out yet, but this could become Walsh's signature reform. Markets should adapt: the Fed will make fewer promises, and future expectation volatility may be higher.
— Austan GoolsbeeReplacing people to influence rates is the red line for independence
Goolsbee divides today's pressures into two types: personal attacks on Powell, Lisa Cook, and others, including criminal investigations and attempts to remove them; and direct demands to 'cut rates.' He says the Fed's independence has traditionally been narrowly defined—setting rates without political interference. Once a sitting government tries to influence rates by replacing people, 'any economist would see that as a problem.' He does not go all the way, but this is his clearest statement to the White House.
— Austan GoolsbeeAI is a reason to hike rates in the short term, not cut
He first cools the AI narrative with district facts: manufacturing has the highest share in the entire Fed system, but firms are still looking for use cases and complaining that tokens are too expensive; productivity growth, after a two-and-a-half-year pickup, has shown about six months of weakness. Then he offers a counterintuitive judgment: if everyone simultaneously pours money into data centers, betting with IPOs or expected future earnings, the short-term economy could overheat. The central bank's daily job is to prevent overheating, not to judge whether AI will revolutionize things in 20 years, so AI's short-term policy implication may be to push rates up, not down.
— Austan GoolsbeeAI taking all jobs is the lump of labor fallacy
He attributes the narrative that 'AI will make everyone unemployed' to the lump of labor fallacy: treating the total amount of labor as a fixed stone that, once AI surpasses it, leaves no new jobs for people. He says this claim has been proven wrong every time in the past; he admits 'maybe this time is different,' but he is willing to bet the opposite: in 20 years, there will not be 95% unemployment, nor will only six people own all AI companies. He positions himself as a 'grim optimist'—not shying away from short-term pain but optimistic about long-term income growth.
— Austan GoolsbeeThe terminal rate is roughly 3%, provided inflation turns
Goolsbee offers a rare quantitative anchor: his personal loose view is that if inflation returns to a 2% path, rates will converge to the neutral rate (what he calls star), roughly 3% nominal = 2% inflation + 1% real. But the precondition is that inflation must turn downward; the past five and a half years have been above target, and the last year has been moving in the wrong direction. For rate-sensitive industries, his advice is to watch data and inflation, not the stock market—the people around the FOMC table discuss the real economy; the stock market is only a secondary reference.
— Austan GoolsbeeIn their own words · checked verbatim
I think that 19 people sitting around the table. Each one with a different worldview is pretty important.
Austan Goolsbee4:15
If.People.Become convinced.That inflation is going to be with them for an extended period.The job of the Fed becomes a hundred times harder.
Austan Goolsbee9:25
the chairman doesn't like forward guidance let's do less explicit promises let's not tie your hands
Austan Goolsbee14:44
if a sitting administration is trying to remove Fed officials to get interest rates down or to influence interest rate decisions, that's a problematic circumstance
Austan Goolsbee20:34
The more hype there is, the more chance there is that it overheats things today
Austan Goolsbee25:00
I guess I'm a grim optimist.
Austan Goolsbee29:35
Low hiring, low firing is extremely unusual environment.
Austan Goolsbee31:09
We've been above the target for five and a half years.
Austan Goolsbee35:00
Figures
| Number of seats at the Fed's rate-setting table | 19 (12 regional Fed presidents + 7 Board appointees) | 4:15 |
| Term length for FOMC political appointees | 14 years, staggered | 4:15 |
| Duration above the 2% inflation target | Nearly 6 years, of which nearly 5.5 years | 10:31 |
| Number of working groups set up by Walsh | 5 (inflation, AI and productivity, balance sheet, etc.) | 14:44 |
| Chicago Fed district's manufacturing status | Highest share among all Fed districts | 23:58 |
| Observation window for productivity weakness after pickup | After 2.5 years of pickup, about 6 months of weakness | 25:00 |
| Loose target for the neutral rate | 3% nominal = 2% inflation + 1% real | 35:00 |
Glossary
- Forward guidance
- A central bank's explicit commitment about the future path of interest rates; new Chair Walsh dislikes it and advocates fewer promises.
- Core inflation
- Inflation excluding food and energy; the Fed considers it a better reflection of underlying true inflation.
- Unanchoring of inflation expectations
- When inflation expectations lose their anchor, wages and prices chase each other in a self-fulfilling spiral.
- Stagflation
- A situation where inflation and employment both deteriorate, putting the two goals in conflict.
- Lump of labor fallacy
- The mistaken assumption that the total number of jobs is fixed, so technological substitution will not create new jobs.
- FOMC
- The Federal Open Market Committee, the Fed's rate-setting body, with 19 people around the table voting.
How to listen
Founders, CFOs, and investors concerned about funding costs and rate trends; those planning AI capital expenditures or hiring decisions.
The opening ad and Jackson Hole travel pleasantries (about the first 4 minutes) can be fast-forwarded; the core content is in the middle and later sections.