On 16 November 2026 Fed Chairman Kevin Warsh described three “illustrative scenarios, not forecasts” for AI, which he calls “super intelligence” (the term President Trump prefers, according to Reuters): a shared productivity boom, a boom with job displacement, and an investment disappointment in which equity valuations fall. He attached no numbers and said nothing about interest rates or inflation. For finance teams the scenarios are useful as named planning cases, provided you supply your own assumptions. Other Fed officials filled in parts of the picture the same week: Vice Chair Jefferson said faster productivity growth could raise interest rates over time, and Governor Barr set out how much of the AI build-out is financed with debt and off-balance-sheet commitments.
Central bankers rarely name a scenario in which the stock market falls. Kevin Warsh did so on Monday, in remarks opening the Productivity and Jobs Task Force conference at the Federal Reserve Bank of San Francisco. He also named two scenarios in which growth runs above its long-run average. Finance teams now have three cases to set against their own plans.
Our Notes briefing on the remarks covers the announcement. This analysis goes back to the speech text, adds what other Fed officials said in the same week, and sets out how to turn the three scenarios into planning cases with rate and capex assumptions you can defend.
This is not financial or investment advice. Nothing here predicts a Federal Reserve decision; statements about rates are attributed to officials or to market-pricing sources. Check your own position with your advisers.
What happened, and what the coverage adds
The Federal Reserve Board published the remarks on 16 November. The conference was hosted by the Board and the San Francisco Fed. Its task force is one of five the Fed announced in June and is led by Marc Andreessen, Stanford economist Charles Jones (on leave at Anthropic) and Microsoft’s Asha Sharma.
The passage that matters for planning is short. Warsh said the Board’s staff “has begun to evaluate various contingencies, applying stylized, model-based frameworks to generate scenarios”, and added: “These are illustrative scenarios, not forecasts.” He said many other outcomes could be drawn, “probably including one that will prove most accurate”.
| Scenario | Growth, in Warsh’s words | Jobs and markets |
|---|---|---|
| 1. Shared gains | Real GDP growth “rises persistently above its long-run average” | The US “stays close to full employment”; gains shared across labor and owners of capital |
| 2. Growth with displacement | GDP growth “would still rise above its long-term average” | Automation and labor market frictions “could push the jobless rate above levels consistent with full employment” |
| 3. Disappointment | “An economic slowdown”; later a recovery “without the productivity gains” | Equity valuations “could fall”, with “a more notable rise in unemployment” |
The coverage adds several points that the text alone does not show:
- No rates, no inflation. The Associated Press reported that Warsh did not comment on the Fed’s next steps on interest rates and did not mention inflation, although several colleagues link the data center build-out to faster price increases.
- The panels went further than the speech. CNBC’s analysis reports that Warsh asked panelists whether AI’s effect on services should be expected to bring disinflation over the medium term, and that he compared the possible effect to the “China shock” in goods.
- Executives played down near-term displacement. Yahoo Finance quoted Palo Alto Networks chief executive Nikesh Arora as saying it is “still early to give it arms and legs”, meaning independent authority to act.
- A report is coming. According to CNBC, the task force members will write a report for Warsh within the next few months, and the wider Fed would then debate whether and how to use it.
We read the speech on federalreserve.gov. We could not open the Bloomberg or Quartz reports, and we could read only the opening of the Washington Examiner article. The speech gives no figures for growth, unemployment or rates under any scenario.
What Google Trends shows
We compared four search terms worldwide over the past 90 days. Values are relative: 100 is the highest point in the comparison, which “Kevin Warsh” reached on 28 August. All four terms were lower in the last seven days than in the first 60. “AI productivity” averaged 11.8 early in the window and 2.0 in the last week. “Fed rate hike” fell from 7.9 to 0.7; its peak of 76 came on 16 September, the second day of that month’s FOMC meeting.
The speech is too recent to show in the averages. The last five daily values for “Kevin Warsh” are 1, 0, 0, 7 and 12, so interest picked up on the day of the remarks and the day after, but from a low base and far below the August peak. “Superintelligence” never rose above 11 and sat at 1 on each of the last five days. The Fed writes the term as two words, so this one-word query may understate interest.
Search interest in a December hike is low even though, according to CME FedWatch data cited by CNBC, markets have largely priced in a quarter-point increase. This is a specialist story, and board members are more likely to have seen a headline than the caveats.
Reading the three scenarios as a planner
Three features of the speech matter when you translate it into a model.
Two of the three scenarios are growth scenarios. Only the third has a slowdown. The difference between the first and second is distribution: whether jobs are “displaced, rather than complemented, by technology”. A plan with one AI upside and one AI downside misses the case in which revenue holds up while the labor market weakens.
The good scenario is not a low-rate scenario. Warsh did not address rates. Vice Chair Philip Jefferson did, in a speech on 13 November. If the technology leads to permanently higher productivity growth, he said, “it may increase firms’ desire to invest and, hence, their demand for funding, raising the level of interest rates”. He added that predicting the neutral rate is difficult. Jefferson also said AI-related spending “is putting upward pressure on resource utilization and the general price level”.
The displacement scenario has no clear policy response. Jefferson noted that monetary policy “cannot address the structural factors that determine long-run rates of employment”. A rise in unemployment caused by skills mismatch would not necessarily bring lower rates.
Jefferson was cautious on the evidence. Labor productivity growth averaged 2.2% between 2020 and 2025, against 1.5% in the previous business cycle, but the pickup began before popular AI chatbots appeared in 2022. One survey of senior managers that he cited estimated that AI had raised sales per employee by 0.5% over three years.
Where the disappointment scenario would bite
Warsh’s third scenario is brief. Governor Michael Barr’s speech in Milan on the same day, drawing on the Fed’s Financial Stability Report of 13 November, describes the financing behind it:
- Debt. On-balance-sheet debt of the seven largest AI firms is “expected to exceed $800 billion before year’s end”.
- Commitments. The report estimates that the potential maximum nominal value of long-term contracts, lease commitments and residual value guarantees “now exceeds $3 trillion”.
- Circular financing. Hyperscalers and GPU producers hold more than $100 billion of cumulative equity investments in model labs and cloud start-ups that buy or rent their products.
Barr named two channels from a repricing to the wider economy: a direct drop in investment, and lower consumer and business spending as wealth falls. CFO Dive reported that 87% of the market contacts surveyed for the report cited AI as a top risk, up from 50% in the first half of the year. Our note on the Financial Stability Report has more.
A company does not need to be a technology firm to be exposed. Ask whether your revenue depends on customers’ AI capital spending, and whether your lenders, pension assets or share-based pay depend on current equity levels.
A playbook for turning the scenarios into planning cases
These steps are our suggestions; the numbers must come from your own business.
- Name three cases after the speech. Use the same labels in board papers, treasury policy and the long-range plan.
- Separate the near-term rate path from the scenarios. The December meeting is on 8 and 9 December and includes a Summary of Economic Projections. The target range is 3-3/4 to 4 percent. Run that sensitivity as its own exercise; our December treasury playbook covers it.
- Do not pair faster growth with cheaper money by default. In the two growth cases, test flat and higher long-term rates. Jefferson’s reasoning is the basis for the higher path.
- Model unemployment separately from revenue in case two. Decide what a weaker labor market means for customer demand, your wage bill and credit losses.
- Build case three from the financing channel. Assume lower equity values, wider credit spreads and slower customer capex, then test covenant headroom, refinancing dates and counterparty concentration.
- Stress AI business cases against the evidence. If a project assumes productivity gains well above the 0.5% sales-per-employee figure Jefferson cited, record the reason.
- Stage capex with decision points. Separate committed from optional AI-related spending and define what would trigger a pause.
- List off-balance-sheet commitments. Long-term compute, lease and energy contracts are the items Barr highlighted. Know your own.
- Assign indicators and a review date. Choose a few measures for each case, such as productivity data, unemployment and equity valuations.
What to watch next
- 18 November: minutes of the October FOMC meeting, according to CNBC. The Fed’s calendar did not yet list a release date when we checked.
- 25 November: the PCE price index, according to The Motley Fool.
- 8 and 9 December: the FOMC meeting. Barchart reported on 16 November that markets priced a 100% chance of a 25 basis point increase; a Motley Fool article of 14 November put futures pricing at 87% for a quarter point and 13% for a half point. Governor Christopher Waller said on 10 November that a 50 basis point hike “should be considered” if core CPI came in around 0.3% or higher. Cleveland Fed President Beth Hammack said “I like moving in smaller increments”, Reuters reported. See our note on Waller’s speech.
- The task force report: expected within the next few months, according to CNBC.
Frequently asked questions
Did Warsh say which scenario is most likely?
No. He said he was not offering predictions and that the most accurate outcome may be one he did not describe.
Did the speech change the outlook for the December meeting?
The prepared remarks do not mention interest rates, inflation or the December meeting. Market pricing for December, as reported by CNBC and Barchart, reflects inflation data and other officials’ comments.
Are there official numbers for the three scenarios?
Not in the published remarks. Warsh gave directions only: growth above or below its long-run average, unemployment near or above full-employment levels.
Does a productivity boom mean lower borrowing costs?
Not necessarily. The Associated Press notes that Warsh has previously argued that AI could let the economy grow without worsening inflation. Jefferson said last week that permanently higher productivity growth could raise the level of interest rates. Plan for both.
Sources and further reading
- Federal Reserve: Chairman Warsh, Opening Remarks (16 November 2026); San Francisco Fed conference page and agenda; Fed task force page.
- Other Fed speeches: Jefferson (13 November), Barr (16 November), Waller (10 November).
- Fed documents: FOMC statement, 28 October 2026; FOMC calendar; Financial Stability Report.
- Reuters on the remarks (via Yahoo Finance); Associated Press (via Yahoo Finance); Yahoo Finance.
- CNBC: analysis, news report, Treasury market report.
- CFO Dive, American Banker, Axios and Decrypt (all via Yahoo Finance); Washington Examiner (opening paragraphs only).
- December pricing and officials: Barchart, The Motley Fool, Reuters on Hammack (all via Yahoo Finance).
- Google Trends: four-term comparison, past 90 days.
Discover more from Kurums | Business Intelligence
Subscribe to get the latest posts sent to your email.