Part 1: The data
TL;DR: October was a round trip for interest rates. The 10-year Treasury yield hit 5.31% on October 5, its highest since 2002, then fell to 4.94% by month-end, and almost all of that drop came from the after-inflation part of the yield. The Fed held rates steady on October 28, but both dissenters wanted another increase. September’s inflation reports came in hot: consumer prices were up 3.72% and the Fed’s preferred core measure 3.12%. Investors now expect a bit more inflation over the next five years. The economy grew at a strong 3.4% pace in the third quarter, but hiring was slow and consumers are spending faster than their incomes are rising, so the saving rate fell to 3.8%. Slower dollar growth has also thinned the cushion we described last month, and federal debt is now edging up relative to the economy. Fuel is still the pinch point. Diesel’s margin over crude narrowed as crude rallied, but pump diesel is 71% above a year ago and jet fuel nearly double. US diesel and heating oil stocks are the lowest for this time of year since records began in 1982, which matters for anyone heating with oil this winter. Credit spreads widened into month-end, with the most strain on the weakest borrowers. Big tech can still borrow: Microsoft sold $25 billion of bonds on October 29, though spreads on the AA-rated bonds of Alphabet, Amazon and Meta ended the month about a third above their twelve-month average. Amazon, Alphabet, Microsoft and Meta spent $202 billion on data centers and other projects in the third quarter, more than double a year earlier, and more of that is now borrowed.
Interest rates
Indicator | Latest read | Four weeks prior | Change | As of | Read |
Broad US dollar index | 121.67 | 121.79 (10/1) | -0.1% | Oct 30 | near 12-month high (122.00 on 10/14) |
2-year Treasury yield | 4.56% | 4.78% (10/1) | -22 bp | Oct 30 | low of the month |
10-year Treasury yield | 4.94% | 5.24% (10/1) | -30 bp | Oct 30 | 5.31% on 10/5, highest since 2002 |
30-year Treasury yield | 5.31% | 5.61% (10/1) | -30 bp | Oct 30 | 5.67% on 10/7 |
2-year to 10-year spread | +38 bp | +46 bp (10/1) | -8 bp | Oct 30 | ranged +29 to +51 bp |
10-year after-inflation (real) yield | 2.60% | 2.88% (10/1) | -28 bp | Oct 30 | 2.95% on 10/5, highest since 2008 |
10-year market-implied inflation | 2.34% | 2.36% (10/1) | -2 bp | Oct 30 | little changed |
5-year market-implied inflation | 2.46% | 2.36% (10/1) | +10 bp | Oct 30 | up after hot inflation data |
High-yield corporate spread | 3.44% | 3.24% (10/1) | +20 bp | Oct 30 | widest since March |
Investment-grade corporate spread | 0.94% | 0.86% (10/1) | +8 bp | Oct 30 | matches its 12-month high |
VIX (expected stock volatility) | 19.42 | 16.39 (10/1) | +3.03 | Oct 30 | highest month-end since March |
Weekly jobless claims | 194k | 199k (wk 9/26) | -5k | wk end 10/24 | near the low end of its range |
Source: US Treasury daily par yield curve and par real yield curve rates (2-, 10- and 30-year yields; 10-year after-inflation yield; 2-year to 10-year spread); Federal Reserve Bank of St. Louis FRED database (Federal Reserve Board broad dollar index; 10-year and 5-year breakeven inflation; ICE BofA US High Yield and US Corporate index option-adjusted spreads; Cboe VIX; US Department of Labor initial jobless claims). Latest values October 30, 2026, except jobless claims (week ending October 24, 2026). Prior values October 1, 2026, and jobless claims for the week ending September 26, 2026. Twelve-month and since-2002 comparisons are Bergenn Financial Group calculations from the same sources.
The Fed paused. The Fed held its policy rate at 3.75% to 4.00% on October 28, after raising it in September. The vote was 10 to 2, and both dissenters, Lisa Cook and Beth Hammack, wanted another quarter-point increase. The statement described economic activity as expanding “at a strong pace” and said inflation “remains elevated.”
October was a round trip for long-term rates. The 10-year yield reached 5.31% on October 5, its highest close since May 2002. The after-inflation 10-year yield reached 2.95% the same day, its highest since November 2008. From there yields fell for most of the month. By October 30 the 10-year was 4.94%, 30 basis points lower than four weeks earlier, and the 2-year had fallen 22 basis points to 4.56%. The largest single move came on October 22, when the 10-year fell 17 basis points and the 2-year 16 in one session.
Last month we pointed out that the after-inflation part of the 10-year yield had done all of the rising. In October it did the falling: 28 of the 10-year’s 30 basis point decline. Expected inflation over ten years barely moved.
The bond market’s inflation expectations, October 30, with October 1 for comparison:
Maturity | Nominal yield | After-inflation yield | Implied inflation | Implied, Oct 1 | Change |
5 years | 4.71% | 2.25% | 2.46% | 2.36% | +10 bp |
10 years | 4.94% | 2.60% | 2.34% | 2.36% | -2 bp |
30 years | 5.31% | 3.08% | 2.23% | 2.30% | -7 bp |
Source: US Treasury daily par yield curve and par real yield curve rates, October 1 and October 30, 2026. Implied inflation is the nominal yield less the after-inflation yield, calculated by Bergenn Financial Group.
The five-year is the exception. Its after-inflation yield fell 40 basis points while its implied inflation rose 10, to 2.46%. Investors now expect somewhat more inflation over the next five years and less over thirty. That fits the data that arrived during the month: inflation reports came in hot and the jobs report came in soft. Last month expected inflation fell at every maturity. This month the near term moved up.
Interest rates and risk, twelve-month trajectory. The after-inflation yield (gold) did the rising into September and most of the falling in October, while implied inflation (steel blue) has barely moved all year. Credit spreads and stock market volatility rose into month-end.
Source: US Treasury daily par yield curve and par real yield curve rates (10-year yield, after-inflation yield and implied inflation; 2-year to 10-year gap); Federal Reserve Bank of St. Louis FRED database (Federal Reserve Board broad dollar index; ICE BofA US High Yield Index option-adjusted spread; Cboe VIX; US Department of Labor initial jobless claims). Month-end values, October 2025 to October 2026; the October 2026 point is October 30, 2026 (jobless claims, week ending October 24, 2026). Implied inflation is the nominal yield less the after-inflation yield, calculated by Bergenn Financial Group.
Inflation, jobs and growth
Indicator | Latest read | Prior reading | Change | Read |
Consumer prices (CPI), past 12 months | 3.72% (Sep) | 3.35% (Aug) | +0.37 | gasoline pushing it up |
Core CPI, excluding food and energy | 2.57% (Sep) | 2.45% (Aug) | +0.12 | up from a 12-month low |
PCE prices, past 12 months | 3.65% (Sep) | 3.44% (Aug, revised) | +0.21 | |
Core PCE (the Fed’s preferred measure) | 3.12% (Sep) | 3.03% (Aug, revised) | +0.09 | Fed target is 2% |
Jobs added | +29k (Sep) | +133k (Aug) | -104k | Jul and Aug revised down 60k |
Unemployment rate | 4.2% (Sep) | 4.1% (Aug) | +0.1 | |
Average hourly earnings, past 12 months | 3.03% (Sep) | 3.11% (Aug) | -0.08 | slowest since 2021 |
Real GDP growth, annual rate | 3.4% (Q3) | 2.2% (Q2) | +1.2 | advance estimate |
GDP price index, annual rate | 2.2% (Q3) | 6.1% (Q2) | -3.9 | inflation share shrank |
Consumer spending after inflation, past 12 months | 2.54% (Sep) | 2.69% (Aug, revised) | -0.15 | |
After-tax income after inflation, past 12 months | 0.95% (Sep) | 1.33% (Aug, revised) | -0.38 | below 1% |
Personal saving rate | 3.8% (Sep) | 4.0% (Aug) | -0.2 | 4.5% in July |
Source: US Bureau of Labor Statistics, Consumer Price Index (September 2026, released October 14, 2026) and Employment Situation (September 2026, released October 2, 2026); US Bureau of Economic Analysis, Gross Domestic Product, third quarter 2026 advance estimate (October 29, 2026) and Personal Income and Outlays, September 2026 (October 30, 2026); all via FRED. Changes are in percentage points except jobs. “Slowest since 2021” for hourly earnings is as reported in the press on October 2, 2026.
Inflation firmed. September’s reports were the first to include September’s oil spike. Headline consumer prices rose 3.72% from a year earlier, up from 3.35%. Core consumer prices rose 2.57%, up from 2.45%. Core PCE, the Fed’s preferred measure, rose to 3.12%. Retail gasoline was still 41% above a year earlier in late October, so the October report, due in mid-November, will carry a large energy contribution again.
A correction. Last month we reported core PCE at 3.34% and stuck. On September 30 the government’s annual revision lowered it: July is now 2.98% and August 3.03%. Most of the change came from new methods for measuring software prices and investment-management fees. The gap between core PCE and core CPI, which we put at 0.88 percentage points last month, is now 0.55.
Growth was strong; incomes were not. The economy grew at a 3.4% annual pace after inflation in the third quarter, up from 2.2% in the second, and the inflation part of growth shrank. Consumers did much of the lifting, and they are stretching to do it. After inflation, spending is up 2.5% from a year ago while after-tax income is up less than 1%. The difference is coming out of savings: the saving rate fell to 3.8% in September from 4.5% in July.
Hiring stayed slow. The economy added 29,000 jobs in September, and July and August were revised down by a combined 60,000. Unemployment rose to 4.2%. Layoffs remain low: weekly jobless claims ended the month at 194,000. The October jobs report, released November 6, was much stronger; we cover it in Part 3.
Last month’s debt arithmetic, updated. In September we said the debt was roughly stable because the economy was growing faster in dollar terms than the average rate on the debt, by enough to offset the deficit before interest. We also said the cushion would shrink if inflation cooled. It has. The fiscal year closed on September 30 with a deficit of $1.99 trillion, of which $1.05 trillion was net interest, leaving a deficit before interest of about 2.85% of GDP. The average rate on the debt edged up to 3.52%. Dollar growth slowed to 5.9% over the past year, from 6.3%, and the Atlanta Fed’s running estimate for the fourth quarter points to about 5.4%. On those numbers the debt is edging up relative to the economy, by roughly 1 percentage point of GDP a year. Holding it flat would take dollar growth of about 6.7% (our estimates).
Source: US Treasury, Monthly Treasury Statement (fiscal year through September 30, 2026) and average interest rates on Treasury securities (September 30, 2026); Federal Reserve Bank of Atlanta, GDPNow (October 29, 2026); US Bureau of Economic Analysis (GDP and PCE prices). The fourth-quarter estimate adds GDPNow’s real growth to third-quarter PCE inflation. Calculations by Bergenn Financial Group; estimates are opinions, and actual results will vary.
Credit
Spreads, the extra yield companies pay over the government to borrow, widened over the month and ended it at their widest since spring. The high-yield spread rose from 3.24% to 3.44%, its widest since March. Investment grade widened from 0.86% to 0.94%, matching its 12-month high, and emerging-market corporate debt from 1.61% to 1.81%. Much of the move came on October 30 alone: high yield was 3.26% the day before.
The weakest borrowers are under the most strain. The spread on CCC-rated debt, the lowest tier of high yield, reached 13.56% on October 23, its highest in twelve months, and ended the month at 13.27%. The gap between CCC and BB, the highest tier, reached 11.50 percentage points the same day. Last month we called a 9.28-point gap the widest of this cycle. It has kept widening, so lenders are drawing a sharper line between strong and weak borrowers.
AI borrowers. The largest AI builders can still borrow on good terms. On October 29 Microsoft (MSFT) sold $25 billion of bonds in ten parts. It drew about $115 billion of orders and priced up to 0.30 percentage points inside its first guidance. Spreads on AA-rated bonds, the band that includes Alphabet (GOOGL), Amazon (AMZN) and Meta Platforms (META), still drifted wider, from 0.61% to 0.70% over the month, about a third above their twelve-month average. The pressure is greater further down. Shares of CoreWeave (CRWV), which rents out AI chips and has borrowed heavily to buy them, fell 31% in the four weeks to October 29, and its 2031 bonds were reported trading above a 16% yield in mid-October.
Source: ICE BofA US High Yield, US Corporate, AA US Corporate, CCC & Lower US High Yield, BB US High Yield and Emerging Markets Corporate Plus index option-adjusted spreads, via FRED, October 1 to October 30, 2026 (the AA band’s twelve-month average from October 31, 2025); Yahoo Finance (CoreWeave share price, October 1 to October 29, 2026); press reports on the Microsoft bond sale (October 29, 2026) and CoreWeave bond prices (mid-October 2026).
Energy and commodities
Indicator | Latest read | Four weeks prior | Change | As of | Read |
Diesel margin over crude, $ per barrel (Jan 2027 futures) | $79.01 | $85.69 (10/1) | -$6.68 | Oct 30 | $93.51 on 10/13; about twice April’s level |
Diesel futures, $ per gallon (Jan 2027) | $4.582 | $4.384 (10/1) | +4.5% | Oct 30 | rose a third as much as crude |
Jet fuel, US Gulf Coast, $ per gallon | $4.444 | $4.355 (10/1) | +2.0% | Oct 27 | +98% from a year ago |
Retail diesel, $ per gallon | $6.352 | $6.382 (wk 9/28) | -0.5% | wk of 10/26 | +71% from a year ago |
Retail gasoline, $ per gallon | $4.269 | $4.465 (wk 9/28) | -4.4% | wk of 10/26 | +41% from a year ago |
US diesel and heating oil inventories, million barrels | 102.8 | 105.2 (wk 9/25) | -2.4 | wk end 10/23 | lowest for the season since records began in 1982 |
Brent crude futures (Jan 2027) | $113.43 | $98.43 (10/1) | +15.2% | Oct 30 | crude outran diesel |
Brent crude, physical cargoes | $123.10 | $114.82 (10/1) | +7.2% | Oct 27 | about $10 above futures |
Gold, per ounce | $3,898.90 | $4,202.30 (10/1) | -7.2% | Oct 30 | 27% below its January 29 peak |
US manufacturing construction spending | $170.7bn | $170.3bn (Jul) | +0.2% | Aug 2026 | first rise after seven declines |
Emerging-market corporate spread | 1.81% | 1.61% (10/1) | +20 bp | Oct 30 | widest since March |
Manufacturing new orders proxy | 13.8 | 15.6 (Sep) | -1.8 | Oct 2026 | still growing, more slowly |
Japan core inflation, vs. year ago | 2.1% | 1.7% (Aug) | +0.4 | Sep 2026 | above the 2% target |
Source: Yahoo Finance (NYMEX NY Harbor ULSD and ICE Brent futures, January 2027 contracts; COMEX gold futures); Federal Reserve Bank of St. Louis FRED database (US Energy Information Administration US Gulf Coast kerosene-type jet fuel spot price, weekly US on-highway diesel and regular gasoline retail prices, and Europe Brent spot price; US Census Bureau manufacturing construction spending; ICE BofA Emerging Markets Corporate Plus Index spread; Federal Reserve Bank of New York and Federal Reserve Bank of Philadelphia manufacturing survey new orders, averaged for the proxy); US Energy Information Administration, Weekly Petroleum Status Report (distillate fuel oil inventories, weekly records from 1982); Statistics Bureau of Japan (consumer prices excluding fresh food, September 2026, as reported by InvestingLive, October 23, 2026). The diesel margin is the ULSD futures price times 42 gallons per barrel, less the Brent futures price for the same month, calculated by Bergenn Financial Group. Data as of the dates shown, latest October 30, 2026.
Diesel is still the pinch point, but its margin narrowed. The gap between the price of diesel and the price of the crude it is made from, the refining margin, fell from $86 to $79 a barrel over the month on January futures, after reaching $93.51 on October 13. Crude did the moving: Brent rose 15% while diesel futures rose 4.5%. Measured over thirteen weeks, diesel rose less than crude in percentage terms but more in dollars: the margin is still well above the $66 a barrel it stood at on July 30. The narrowing only began after the October 13 peak, so we would not overread one month. The margin is still about twice its April level and far above where it ran from 2021 to 2025.
Fuel prices stay high. At the pump, diesel averaged $6.35 a gallon in the last week of October, 71% more than a year ago and $0.18 below its September 21 record. Jet fuel on the Gulf Coast cost $4.44 a gallon on October 27, nearly double a year earlier. Gasoline eased to $4.27, still 41% above a year ago. US stocks of diesel and heating oil fell to 102.8 million barrels, the lowest for this time of year in weekly government records going back to 1982 and about 10% below the five-year average. Diesel and heating oil are essentially the same product, and many Connecticut households heat with oil, so these stocks matter going into winter.
Crude tightened again. Physical cargoes of Brent sold for $123.10 a barrel on October 27, about $10 above futures. As we wrote last month, when physical barrels trade well above futures, buyers need oil now and cannot wait.
Diesel and refining. Left: the margin between diesel and Brent crude futures for January 2027 delivery, with October shaded. Right: US distillate inventories (diesel and heating oil) in 2026 against the range and average of the previous five years.
Source: Yahoo Finance (NYMEX NY Harbor ULSD and ICE Brent futures, January 2027 contracts, daily settlements, April 1 to October 30, 2026); US Energy Information Administration, Weekly Petroleum Status Report (US distillate fuel oil inventories, weekly, 2021 to the week ending October 23, 2026). The margin is the ULSD price times 42 gallons per barrel, less the Brent price; the 4-week average is the average of the last 20 trading days. Calculated by Bergenn Financial Group.
The state of the wars, as of October 30. About 4 million barrels a day of refining capacity remains out of action across the two wars: nearly 2 million in the Middle East, by the International Energy Agency’s count, and roughly 2 million in Russia, by our tally of press reports. No damaged refinery in either region had been confirmed back in service. Ukraine’s strikes on Russian refineries appear to have resumed: Rosneft’s Novokuibyshevsk plant caught fire on the night of October 29, a day after the Wall Street Journal reported a pause in such strikes. Reuters reported that Russia plans to extend its ban on diesel exports by producers while allowing larger export quotas. In the Gulf the damage so far is to shipping and export sites, not refining units: a fire at Saudi Arabia’s Ras Tanura export terminal on October 27 did not affect operations, according to Saudi Aramco, and two tankers were hit near the Strait of Hormuz on October 29 and 30. Qatar is mediating between Washington and Tehran. This changes week to week.
Gold and Japan. Gold fell 7.2% over the month to $3,899 an ounce and is 27% below its January 29 peak of $5,318. It fell even as after-inflation yields eased, so lower rates gave it no lift this month. Japan’s core inflation rose to 2.1% in September from 1.7%, which supports the case for further Bank of Japan rate increases.
Source: International Energy Agency (Middle East refining outages); Wall Street Journal (October 29, 2026); Reuters (Russian diesel export policy, October 29 to 30, 2026); Saudi Aramco statement on Ras Tanura (October 27, 2026); press reports on tanker attacks (October 29 to 30, 2026). The Russian capacity figure is Bergenn Financial Group’s estimate from press reports of individual refinery outages.
Big tech spending
Amazon, Alphabet, Microsoft and Meta reported in the last ten days of October. Together they spent $201.9 billion on data centers, chips and other capital projects in the third quarter, up 22% from the second quarter and more than double a year earlier. Amazon alone spent $77.3 billion.
Spending is now running ahead of the cash these businesses generate. After capital spending, free cash flow was negative at Amazon (-$34.8 billion), Alphabet (-$9.1 billion) and Meta (-$1.2 billion); only Microsoft was positive ($18.6 billion). More of the difference is being borrowed. Amazon added a net $48.9 billion of debt in the quarter and Alphabet $25.4 billion.
Meanwhile the price of using AI keeps falling. Silicon Data’s broad index of AI token prices, the unit AI models are billed in, set a new low of $0.94 per million tokens in late October, down about 55% since late May. In September we wrote that token prices had stabilized at the end of the summer. They have since resumed their decline, while spending on the computing behind them keeps climbing.
Capital spending by the four largest cloud and AI spenders, by calendar quarter.
Source: Company quarterly cash flow statements (capital expenditures, free cash flow and net debt issuance) for Amazon (AMZN), Alphabet (GOOGL), Microsoft (MSFT) and Meta Platforms (META), via Yahoo Finance and Financial Modeling Prep; calendar quarters, third quarter 2025 to third quarter 2026 (Microsoft’s fiscal quarters end on the same dates). Silicon Data, SDLLMTK LLM token price index, as of October 24, 2026.
Part 2: WTF just happened?
I had this part written and then the election and the two days after happened and it was back to work on a rewrite. Everything changed and nothing changed and the space between those two is where we must dig in and see what the path forward is likely going to be.
As you saw in part 1, the macro signals are all over the place while netting out to not much. There’s been incredible churn in the markets under the surface but we’re only about 6% off the record high from October 6th. Between Monday Nov. 2 and today (Friday) there was quite a bit of sound and fury signifying very little net change in terms of market position. We had a political tsunami followed by gains followed by one of the biggest market drops since Covid. And for all of that, the 5-day change was only -2.68%. The change since June 1st? -2.94% and up 7.76% since the start of the year. SMH since June 1st? -23.67%. But SMH is still up 28.83% for the year. The hyperscalers (Microsoft, Alphabet, Amazon, Meta and Oracle)? -7.69% since June 1st and -0.52% since Jan 1st. Honestly kind of mild numbers for the folks ushering in either a golden age or the end of humanity depending on what you read. Especially with all the money being spent on building data centers (roughly $700 billion this year from the four biggest spenders alone) and promises from all the big players to spend even more next year if you believe their guidance from the last earnings calls. That sets the stage for this week.
While not much has changed in terms of nuts and bolts, perception has shifted massively after Tuesday. Why is that? Texas and other traditionally red states flipped blue. And partly because of candidates campaigning on slowing the growth of data centers in their states. So now we have conflict between the hyperscalers who are viewing compute and AI as existential to their future and citizens who have approached data centers with NIMBY-ist skepticism. I think this will be the central tension until the governing agendas of newly elected officials become clear. And I think this was the root cause of much of the sell-off on Thursday. At least it was the catalyst. I think the initial drop triggered mechanical liquidations of leveraged calls and algorithmic traders that deepened what could have been a simple risk-off day into one of the worst days since Covid.
So what does this new landscape look like? I think the big winners of the AI trade up till now were the “picks and shovels” of the AI build out. So semiconductor stocks, memory, optical interfaces, contractors and specialty construction firms. Anything that went into a data center was in short supply and was doing well. I think moving forward this will be less broadly true. While I don’t think America has built its last data center, I think the pace of new announcements and deals will be slower and more considered. With increased scrutiny and possibly legislation, I think they will also be more expensive. So I think that brings up some new classes of winners moving forward. Anyone with compute already up and running will have pricing power when it rents out that capacity, and looks like it will be doing well. The chips themselves wear out and lose value fast, so the lasting edge is having the power, the grid connections and the buildings already in place. The demand for compute is strong and only going up. Don’t confuse token price with token demand. The price for a given level of AI capability keeps collapsing, even if the newest models can still charge a premium, but the overall demand for tokens is on the rise. And this is with AI agents like Meta’s Muse still in their infancy. So expect the cloud segments of the hyperscalers like Azure and Google Cloud to have eye-popping revenue growth. Even more so than the already eye-popping numbers. This will help with their FCF. It will still probably be negative for everyone but Microsoft, but a little less so. Or maybe not since the build out may be getting more expensive. But it might be slowing down so that will help. But they may still see it as existential and look to increase buildout even more so. It’s going to be a delicate balance to see if the increased revenue and slower buildout lessens debt burden for them or the mindset that this is still an existential competition pushes things into overdrive. Or overseas. As you can tell, it’s too soon to tell which direction things are going. But before the election, we knew hyperscalers were only going to be limited on capex spending by the availability of equipment and power, and how much lenders would give them. Now the concern is that, to some extent, it’s been taken out of their hands. The money being spent and expected to be spent is a major contributor to the US economic growth story in the past few years, though less than the headline numbers suggest, since a lot of the chips and servers are imported. The big question is can the rest of the market pick up the slack an AI slowdown creates?
Source: Yahoo Finance daily closes through November 6, 2026 (S&P 500 index; VanEck Semiconductor ETF, SMH; Microsoft, Alphabet, Amazon, Meta Platforms and Oracle, weighted by market value; the S&P 500’s record close of October 6, 2026); company capital-spending guidance for 2026 from third-quarter earnings reports. Calculations by Bergenn Financial Group.
Part 3: Can anyone else pick up the slack?
So let’s look at the rest of the market. Are the companies outside of AI making more money? And is there a legitimate rotation happening or does it only seem that way because semis got hammered?
Start with the headline: with 88% of the S&P 500 reported, third quarter earnings are up 31.4% from a year ago according to FactSet. That’s the third straight quarter above 25%. Net profit margins came in at 14.9%, the second highest FactSet has recorded since it started tracking in 2009. And since June 1st, the stocks of companies outside of AI are up 5.5% while AI stocks are down 16.4%. So has market leadership already passed and no one noticed?
To answer that, I split the index in two: the companies building, selling or renting out AI, and everything else.[1] Then I looked at every company that has reported a quarter ending in August, September or October (364 of them), using operating profit. I also looked at how the stocks have done since June 1st.
AI vs. non-AI: third quarter results and stock prices since June 1st
AI | Non-AI | Non-AI excluding energy | |
Companies reported | 28 | 336 | 318 |
Sales growth vs. a year ago | +31.9% | +9.9% | +6.7% |
Operating profit growth vs. a year ago | +64.5% | +19.4% | +10.6% |
Operating margin (a year ago) | 31.0% (24.8%) | 17.0% (15.6%) | 16.8% (16.2%) |
Share of total operating profit | 28.5% | 71.5% | 61.6% |
Share of the increase in operating profit | 49.1% | 50.9% | 25.9% |
Stock price change since June 1, weighted by size | -16.4% | +5.5% | +5.4% |
Share of stocks up since June 1 | 10% | 66% | 67% |
Source: Company results for the latest quarter ending August through October 2026, compared with the same quarter a year earlier, for 364 S&P 500 companies, via Yahoo Finance as of November 8, 2026; financial companies use pre-tax profit. Stock prices: S&P 500 members, closes June 1 to November 6, 2026, Yahoo Finance, weighted by current market value. AI and non-AI groups as defined in footnote 1; 28 of the 42 AI companies had results for a quarter ending August through October (NVIDIA and Broadcom have not reported one yet). Calculations by Bergenn Financial Group.
On earnings, AI is still leading by a mile. AI operating profit grew 65%, more than three times as fast as non-AI. FactSet had Micron as the single biggest contributor to S&P 500 earnings growth for the quarter, ahead of even NVIDIA, and semiconductor earnings were up 133%. The AI side earns a bit under 30% of the profits and produced about half of the growth. Non-AI earns the other 70% and produced the other half.
But a big piece of the non-AI half is energy. Energy earnings were up 149%, per FactSet, and most of that came from refining. Oil averaged 32% more than a year ago, but refiners, the companies that turn crude into diesel, jet fuel and gasoline, saw earnings jump 542%. Great for refiners. Not so great for anyone using diesel as an input. Energy alone produced a quarter of the index’s profit growth this quarter. Take it out and the rest of the non-AI world, about 60% of the profits, produced about a quarter of the growth.
Here’s the thing, though: the rest of the non-AI world is doing fine. Sales are up about 7% and operating profit about 11%, and the median company was up 11% too, so it isn’t one or two big names carrying the group. Margins ticked up, from 16.2% to 16.8%, and six out of ten of these companies widened their margins. They did that with diesel at $6.29 a gallon at the pump, up 68% from a year ago, and gasoline up 38%, according to the EIA. That’s not nothing. Honestly, it’s a healthy and normal set of numbers. It shows that companies overall are doing well, and it fits with the strong jobs report that came out on Friday: 165,000 jobs added in October against about 65,000 expected, unemployment down to 4.1%, and August and September revised up by a combined 84,000. The weak spot was consumer discretionary companies (think restaurants, autos, travel and apparel), where sales grew less than 3%. But none of it looks anything like the AI numbers: sales up 32%, operating profit up 65% and margins up six points to 31%.
Even inside AI there’s a split. The cloud giants grew operating profit 20%, with margins about flat. The supply chain (the chips, memory, gear and construction) saw operating profit more than quadruple, with margins going from 19% to 43%. That’s what a shortage looks like, and it’s the part with the most to lose if the build-out slows.
So is there a rotation? In stock prices, absolutely, and it started long before Tuesday. It goes back to June, and Tuesday just sped it up. The typical AI stock is down 27% since June 1st, and only one in ten is higher. Two-thirds of non-AI stocks are higher. AI’s share of the S&P 500’s total value has dropped from about 40% to 34%. That’s how the index can be down 2.9% since June 1st while the typical stock in it is up. Those are the marks of a broadening market. The one thing missing is small caps, down 5.3% since June 1st. But small companies are the most sensitive to interest rates and credit, and with the 10-year Treasury yield still near 4.9%, I think that’s more about rates than about the rotation.
At the AI vs. non-AI level, the rotation runs opposite to the earnings. AI profits grew 65% and AI stocks are down 16%. Non-AI profits outside of energy grew 11% and those stocks are up 5%.
Non-AI companies by sector
Sector (non-AI companies only) | Price change since June 1 | Share of stocks up since June 1 | Q3 operating profit growth, median company | Q3 operating margin (a year ago) |
Health Care | +25.5% | 93% | +14.2% | 11.1% (10.2%) |
Consumer Staples | +11.3% | 78% | +14.0% | 12.9% (12.1%) |
Financials | +11.2% | 83% | +5.9% | 27.6% (28.3%) |
Energy | +8.3% | 48% | +35.2% | 18.1% (10.4%) |
Materials | +2.6% | 63% | +4.5% | 16.0% (14.4%) |
Utilities | +2.5% | 71% | +15.1% | 27.2% (26.6%) |
Real Estate | +2.2% | 37% | +3.4% | 20.7% (22.7%) |
Industrials | -1.8% | 59% | +14.5% | 14.4% (12.3%) |
Information Technology | -1.8% | 47% | +17.2% | 24.7% (24.3%) |
Consumer Discretionary | -2.7% | 70% | +6.9% | 11.0% (10.7%) |
Communication Services | -14.4% | 18% | +37.3% | 24.4% (21.1%) |
All non-AI | +5.5% | 66% | +11.5% | 17.0% (15.6%) |
Source: S&P 500 members outside the AI group, by sector. Price change weighted by current market value, closes June 1 to November 6, 2026, Yahoo Finance. Operating profit: latest quarter ending August through October 2026 against the same quarter a year earlier, company results via Yahoo Finance as of November 8, 2026 (pre-tax profit for financial companies); median of the companies reporting. The financials margin excludes Fiserv, which reported a one-time operating loss of $5.8 billion. Calculations by Bergenn Financial Group.
Health care is up 25.5%, and 93% of health care stocks are higher. Staples are up 11.3% and financials 11.2%. Staples and financials are the steady, defensive corners of the market. Health care is part defensive and part growth (biotech is increasingly seen as an AI beneficiary), and it has the earnings to back it up: the typical company grew operating profit 14% with margins rising, and FactSet has analysts expecting 22% earnings growth for the sector next year. Staples are growing at a decent clip. Financials much less so, with the typical company up 6% and margins slipping. Meanwhile the non-AI parts of tech and communication services, mostly software and media companies, have lagged even though their profits grew faster than most.
That tells me the market is paying up for steadiness. You can see it in valuations. Consumer staples now trade at 23.7 times next year’s expected earnings. Tech, home to most of the AI chip and hardware companies, trades at 17.3, the lowest for the sector since March 2020, according to FactSet. People are paying more for a dollar of toothpaste and cereal profits than for a dollar of chip profits. Some of that is fair. Chip earnings are cyclical, the market is betting this is close to as good as it gets, and after Tuesday there’s a real reason to think the build-out gets slower and more expensive. So the rotation is at least as much about fear of the AI side as it is about excitement for everything else. As for the market as a whole, the S&P 500 trades at 17.7 times next year’s expected earnings, its lowest in three years, because earnings estimates are up 33.6% this year while the index is up 7.4%. That only looks cheap if those estimates hold. Measured against the last ten years of inflation-adjusted earnings, the market trades at about 39 times, a level topped only at the dot-com peak and earlier in this run, and last quarter’s record profit margin was flattered by big one-time investment gains at Alphabet and Amazon.
Election week was a good example of both. Wednesday looked like a textbook rotation day: non-AI stocks rose 2.3% while AI stocks barely moved, with homebuilders up 7.3%, airlines 6.7% and retailers 5.5%. On Thursday AI stocks fell 7.2% and non-AI stocks 2.8%. From Tuesday’s close through Friday, AI was down 5.3% and non-AI down 1.0%. Inside AI the split was even sharper. The cloud giants (Microsoft, Alphabet, Amazon, Meta and Oracle) were up 1.4% for the week, while the AI supply chain fell 11.9%. That lines up with what I said in Part 2: the picks and shovels take the hit and the companies that already own compute hold up.
The part that gives me pause is what analysts are doing with their estimates. Since September 30th, they’ve raised their 2027 earnings forecasts for tech, where the AI chipmakers sit (to 42.2% growth from 40.3%), and trimmed them for most of the other sectors: industrials to 14.8% from 16.3%, financials to 8.3% from 9.2%, utilities to 8.3% from 9.0%, staples to 6.5% from 7.0%, real estate to 6.0% from 6.5%. Health care is the one real exception. The biggest cuts to fourth quarter estimates over the past month include Boeing, United Airlines (-36.5%), D.R. Horton (-36.5%), International Paper and Mosaic. I read these as higher jet fuel costs and still-high mortgage rates doing damage. And most of those estimates were set before Tuesday, so they don’t reflect anything the election may have changed.
Here’s the rough math on why that matters for the slack question. Analysts expect S&P 500 earnings to grow 16.1% in 2027. If AI profits simply stopped growing next year, the non-AI side would need to grow a bit over 22% to hit that number. It did 19% this quarter, which sounds close, but energy did a lot of that, and analysts expect energy earnings to fall 8% next year. Take energy out and the rest of the non-AI world would need to grow about 27%, against the 11% it just did. And that assumes the non-AI side doesn’t take a hit when AI spending slows. A lot of non-AI profit comes from that spending: it’s revenue for many of the utilities, industrial suppliers, copper miners and builders that sit in the non-AI column. If the build-out slows, their profits slow too, unless consumers or the government spend more to make up the difference, and with the saving rate down to 3.8%, consumers are already stretched. A rotation can move money from one part of the market to another. It can’t, by itself, replace earnings growth.
So where does that leave us? Is a sector rotation under way? In prices, yes, and it has been since June. In earnings, not yet. The non-AI companies are in good shape, with margins edging up and profits growing at a healthy clip, but outside of energy they’re growing like a normal economy, and the people paid to forecast them are nudging their numbers down. I think the thing to watch over the next couple of quarters is whether that flips. If non-AI earnings start accelerating, first in analysts’ estimates and then in reported results, the rotation has legs of its own. If they don’t, a lot of this is investors hiding out in the steady stuff until the AI picture clears up, and that money can go right back the other way. I’d also keep an eye on credit. More of the build-out is being paid for with borrowed money, and spreads on the AA-rated bonds of Alphabet, Amazon and Meta are already about 30% above their 12-month average. If lenders pull back, the build-out slows no matter what the politicians decide. NVIDIA reports on the 17th, so we won’t wait long for the next clue on the AI side. Beyond that, it’s going to take a few more months to know what the lasting effects on policy will be from the blue tsunami and what that means for AI. Until we get some clarity, I’d expect this to be a headline-driven, volatile market.
Source: FactSet Earnings Insight (November 6 and October 16, 2026); Yahoo Finance daily closes through November 6, 2026, for S&P 500 members, the S&P 500 index and the iShares Russell 2000 (IWM), iShares US Home Construction (ITB), U.S. Global Jets (JETS) and SPDR S&P Retail (XRT) ETFs; US Energy Information Administration weekly retail diesel and regular gasoline prices, November 2, 2026 against November 3, 2025, via FRED; US Bureau of Labor Statistics, Employment Situation, October 2026 (released November 6, 2026), with the payroll forecast from FactSet; company quarterly results via Yahoo Finance; ICE BofA AA US Corporate index option-adjusted spread via FRED (November 5, 2026, against its twelve-month average); Shiller cyclically adjusted price-to-earnings ratio via multpl.com (November 2026); 10-year Treasury yield via Yahoo Finance (November 6, 2026); US Bureau of Economic Analysis personal saving rate (September 2026). The AI and non-AI figures, the shares of market value and the roughly 22% and 27% figures are Bergenn Financial Group calculations; estimates are opinions, and actual results will vary.
This note is for information only. It reflects our views as of the date shown, is based on data believed reliable but not guaranteed, and is not a recommendation to buy or sell any security or a substitute for advice specific to your circumstances. Estimates are opinions based on stated assumptions; actual results will vary. Past performance is not indicative of future results.
Securities offered through Kestra Investment Services, LLC (Kestra IS), Member FINRA/SIPC. Investment advisory services offered through Kestra Advisory Services, LLC (Kestra AS), an affiliate of Kestra IS. Kestra IS or Kestra AS are not affiliated with Bergenn Financial Group, LLC. Bergenn Financial Group, LLC does not offer tax or legal advice. Investor Disclosures: https://www.kestrafinancial.com/disclosures
[1] AI is 42 companies: the cloud giants (Microsoft, Alphabet, Amazon, Meta and Oracle), NVIDIA and the other AI chipmakers and chip-equipment makers, Palantir, and the companies selling data centers their servers, networking, storage, power and construction. Everything else is non-AI. That includes Apple and Tesla, since their profits come from phones and cars, and the chipmakers that sell mostly into cars, factories and phones. Moving Apple and Tesla to the AI side would barely change the numbers: non-AI operating profit growth would go from 19.4% to 20.5%. Some utilities, industrial suppliers and materials companies in the non-AI group also sell into the build-out, so this split understates how much of the market depends on AI spending.