Liz Thomas says the first rate hike is not what breaks a market. On her reading of every US hiking cycle back more than fifty years, the damage lands somewhere between the third and the fifth.
Most of the pushback she has had is that 25 basis points in September would change nothing. She agrees with that, and says it misses the point.
"So, you've got a market that likely sniffs it out ahead of time, and I agree that the first hike wouldn't be the problem, but it's usually somewhere between the third and the fifth hike that would be the problem."
Thomas is head of investment strategy at SoFi, writes the firm's weekly market note, and was publicly calling a Treasury-versus-Fed standoff before Kevin Warsh's hawkish Jackson Hole speech made it one.
I listened to the full episode so you can skip it. 42 minutes of audio, 18 minutes of reading.
Here are the 11 calls that matter.
👤 Guest: Liz Thomas, Head of Investment Strategy at SoFi, who writes the firm's weekly market note and appears regularly on CNBC
🎙️ Hosts: Guy Adami and Dan Nathan, co-founders of Risk Reversal Media and both former traders
📰 Published: 10 September 2026 on YouTube (MRKT Call)
🔴 YouTube | ⏱️ 42 min | ✅ Time saved: 24 min
Key Takeaways
The market has bottomed only after the Fed stopped hiking, in every cycle she charted, and the break usually comes between the third and fifth hike
In 2000 the roll-over began after the fourth hike; in 1999 the peak came after four, with a 50 basis point hike landing on a market that had already turned
She would buy the 10-year Treasury here, on the view that the Treasury wins its fight with the Fed and Warsh has to back down
Scott Bessent's buyback announcement failed to pull yields down but she says he still has about a trillion dollars of unused capacity
The Fed should neither hike nor cut, and the election calendar pushes the first plausible hike out to December
Split the economy in two — everything that is AI, and everything else — and the second half is already rolling over
Industrials have turned and capital goods orders flatlined in August, which she reads as the market knowing before the data does
The AI capex race is so competitive that she expects companies to cut workers before they cut the build-out, reversing the usual order
She started buying a gold ETF last week, and is long a software index rather than any single software name
"Inflation" was said on only 204 S&P 500 earnings calls last quarter, which she puts down to companies being too busy talking about AI
1. Yields North of 4.9%
The episode opened with the 10-year Treasury at 492 and Adami noting he has been calling 5% for a long time. Thomas said the immediate driver is crude oil but that it is not the whole story.
She does not accept the single-cause explanation. The conversation around yields stems from crude prices, and she granted that a Fed rate hike cannot fix supply imbalances in oil. "There are other factors at work here."
What worries her is how orderly the move has been. "The trajectory has been somewhat sort of methodical. I mean it hasn't had this huge spike today notwithstanding which is one of the reasons I don't think the market has reacted in kind."
She thinks the level is now high enough to matter. "Certainly the stock market is not built for higher rates and we're getting to levels where the market seems to care."
The hosts put the market-implied odds of a September hike at somewhere around 70% — one of them read 71.8% off his own screen during the exchange, against roughly 60/40 a few weeks earlier
2. She'd Buy the 10-Year Here
Thomas had already said publicly that she would buy the 10-year once the yield went through 480, and she repeated it on air. The reason is not a forecast that yields fall; it is a view about who wins a policy argument.
The trade is on the table now. "The other day I said on another CNBC program that above 480 I would actually buy the 10-year." With the yield well through that level, she said she would buy it the same day
She frames it as a yield, not a rally. "So, I believe that the 10-year Treasury will be contained and I think that it is a buy here as a high yield option in investment portfolios. So, I would purchase it. I would purchase it today."
The Treasury's first attempt did not work, and she says it has plenty left. "We've got the Treasury clearly wanting yields to come down. It didn't work. His buyback announcement didn't work, but he's still got plenty of dry powder. Apparently a trillion dollars of dry powder to use if he wants to."
Her definition of the Treasury winning is about rhetoric, not the level of yields. "I think the win is that Warsh is going to have to back down and the Treasury is going to be the one that whose rhetoric wins the day."
She called the standoff before it happened, and says she was mocked for it. Speaking after Bessent announced the buyback program, she said: "Well, this is going to be interesting if Warsh comes out at Jackson Hole and gives a hawkish speech because it's going to be a tug-of-war then between the Treasury and the Fed." Her account of the reaction: "And then it turned out to be the case."
Adami's own framing is that the bond market has taken the decision out of the Fed's hands. He said he does not believe the Treasury and the Fed are on the same page whatever they say, and that Bessent's rhetoric has boxed the Fed in
3. Why She Says No Hike
Asked directly what the Fed should do next week, Thomas gave the same answer twice and then walked through the political calendar that follows from it.
Her position is a hold, not a cut. "Yes, I don't think the Fed should hike. I certainly don't think they should cut. I think they should stay the course for the foreseeable future."
She is confident enough to predict as well as prescribe. "I absolutely do not think that they should hike rates next week. I don't think they will hike rates next week."
The calendar then does the rest. "And then realistically, this is turning into a political battle because if they don't do it next week, they're not going to do it a week before the midterm election. So that means they don't do it in October. So then the earliest chance that they do it is December."
Her case is that the data do not describe an overheating economy. "And then we can come to our senses and realize that one and a half to two and a half% GDP growth and jobs of less than 100,000 a month is not an overheating economy and there is no reason to hike rates."
She also argues Warsh has boxed himself in on the target. The economy, in her account, has done fine with inflation above 2% for five years, and starting to hike in September would signal he intends to keep going until it reaches 2% — which she says neither the economy nor the market can withstand
Nathan's counter is that holding is itself a message. He put the scenario of a 10-year above 5%, Brent at 110, WTI at 107 and inflation still climbing into November: "If it's November 1st and we have gas at the pump at $4.5, you know what I mean? We have grocery prices going higher. We have mortgage prices or mortgage rates at all-time highs." His conclusion was that the Fed chair keeps the Fed's credibility by staying the course
4. The Late-'90s Hiking Chart
This is the section Thomas brought her own charts for, from that week's SoFi note. The argument is that the first hike is a red herring and the cumulative path is what ends cycles.
She concedes the pushback and then reframes it. "The push back that I've gotten so far is well even if the Fed does raise rates 25 basis points in September, it's not going to make any difference. That is absolutely true and I agree. It's the messaging that it would continue and the effect of cumulative rate hikes is what ends up being the problem."
The pattern in the chart is that the market never bottoms while the Fed is still going. "So what you can see from this chart, if you just look at the times when the Fed is raising rates, so the dark line when it's going up, the market bottoms after that process is done in every single case. And usually the Fed continues to raise rates into a market that has already rolled over."
The 2000 cycle gives her the count. "So if you take averages, if you look at ... back in 2000, it was after the fourth rate hike that the market started to really roll over and it hit its peak right before the Fed hiked rates again."
Which produces the claim the episode turns on. "So, you've got a market that likely sniffs it out ahead of time, and I agree that the first hike wouldn't be the problem, but it's usually somewhere between the third and the fifth hike that would be the problem."
She expects the early hikes to look harmless, and says that is the trap. A market would keep rising for the first few, people would tell her she was wrong, "And eventually the market would roll over in my opinion. And I think that hiking cycle would mark the beginning of the end."
The 1999 analogue is the one she lined up event for event. Taking the pause after Long-Term Capital Management's collapse as the parallel to today, she said the market was choppy through the 1999 hikes and then "peaks after the Fed has hiked rates four times. And then they hiked again by 50 basis points after that, the market had already rolled over."
She also pointed at the recession bars on the chart. Every shaded recession sits in the aftermath of one of these hiking cycles, on a pattern she said runs back more than fifty years
5. Two Economies, AI and Rest
Thomas's second chart split the economy rather than the market, and her point was that the non-AI half is already softening while the index hides it.
The framing is deliberately crude. "So this is basically let's think about the economy in two groups. There's everything that's AI and then there's everything else."
Industrials have turned, and she is careful about how much to read into it. "And right now you're looking at industrials the industrial sector which is the blue line here has rolled over and it's not concerning. Everything's still fine. It's still in an uptrend. It hasn't made a new low or anything like that. But it's rolled over and it's a little bit out of character with the rest of the market and some of the cyclicality signals that we're getting."
The orders data point the same way. "We've got ... capital goods orders that kind of flatlined in August." Her question was whether the market is picking up a slowdown before the economic data does
A hike would land on the part of the economy that did not cause the inflation. "So, if you hike rates, you hurt companies like this. You hurt companies in the economy like the industrial sector that's trying to engage in capex for other reasons than AI." High oil prices, she added, hit the same companies
Meanwhile the index can keep going, which is the part she thinks is genuinely different. "The interesting situation that we're in the market right now and with AI is that if we return to that really concentrated market leadership of the Mag 7 and the hyperscalers, then you could have a situation where you've got an economy under pressure, the manufacturing side of the economy, the consumer side of the economy, jobs, everything under pressure because of inflation and war and sluggish growth. But a market that on the index level is whistling past the graveyard and continues to rise for a while."
6. Layoffs Before Capex Cuts
Nathan put a chain of consequences to her: hike into this, get a pullback in AI demand, add real competition from a 5% bond yield, and the economy suffers before the stock market does. Thomas agreed and then named the mechanism she thinks does the damage.
The competitive pressure decides which cost gets cut. "One of the risks of those AI companies is exactly what you just outlined in that the pressure for them to keep spending on capex and keep spending on the AI buildout in order to maintain competitive is so high that companies would perhaps rather lay off workers to save costs rather than stop spending on AI."
That would keep the GDP figures looking fine while the labor data quietly deteriorates, because the private investment still shows up even as headcount comes down
And it inverts the order companies normally cut in. "Usually that's the thing that they want to do last. They try to save costs everywhere else. Marketing, advertising, R&D, capex, all of it before they have to lay people off. This cycle could see a reverse in that."
Nathan's evidence is already on the tape. He pointed to the debt funding the last leg of the build-out getting more expensive, and to Oracle as the test case: "we saw with Oracle, Oracle's like exhibit A for this, what did they start doing? They started firing tens of thousands of people"
Thomas would not date it. She said the reversal has not happened across the board yet and would take a while to show up in the labor data
7. Still Buying Gold
Adami asked whether the gold trade survives a day when higher yields are pushing the metal down, and brought his own supporting evidence from central banks.
Her answer was yes, with a position to back it. "I believe that it is. I believe that it is too. In fact, I started buying a gold ETF last week."
She is not underwriting a return to the January highs. "I don't think that's going to happen again, but I would buy it here because of everything else that's going on" — the "everything else" being the war and how long it is lasting
Adami's case is official-sector buying and repatriation. "I'm sure you read about the Netherlands repatriating somewhere between 75 and 81 tons of their gold, which comes on the heels of France doing it earlier this year."
8. Long the Software ETF
Adami raised the Salesforce quarter as evidence that the market's fear of software being destroyed by AI had gone too far. Thomas said she owns the sector through an index fund, and explained why she stopped owning individual names in it.
She is long the software index and staying long. "So ... you are correct that I have been long IGV and I still am long IGV."
The single name did not survive. "I was also long one software name before the whole SaaS apocalypse thing came out and obviously got killed in the SaaS apocalypse. I decided to exit the singular name because to your point just now, some of them are not going to win. And I think it's too early to know which ones those are."
The index is a bet against the extinction narrative, not on a winner. "So I feel better about buying an ETF and owning the software complex broadly because I don't think everybody dies and it was trading as if everybody was going to go out of business. And that's just not going to be the case."
Cybersecurity is the part of the basket she is most confident in. "Also knowing that we've got cyber security that frankly in this environment the demand for cyber security is only going up." She said she had just recorded a podcast interview with a chief security officer and that some of what he described was frightening
It is a hold, not a trade. "So I would be long software as an industry." And: "I don't think that's a trading game right now."
The hosts' own argument for software was switching costs. They had been telling listeners to buy the dips on valuation and sentiment, and made the case that a large employer cannot rip out a Salesforce deployment that thousands of staff and customers depend on. One host added that the Salesforce chief executive appearing alongside Anthropic's Dario Amodei after the results was itself the message — that the AI labs are not planning to put software companies out of business
9. Memory Has Put In a Top
The hosts walked charts of the Magnificent Seven and then of a DRAM-focused ETF holding Micron and SK Hynix. This call is the programme's, not the guest's.
The technical read is that the memory trade topped months ago. The host pointed to an island reversal at the peak of the memory rally, and a second one in August, as a signal that has not been invalidated since
The conclusion is to sell rallies rather than short. "I think you're going to see bounces along the way. I think the trajectory is lower for these names. I'm not saying you go out and short them, but you have to take advantage of bounces to pair down positions in memory stocks because I think in my opinion, for the stocks at least, we've put in a pretty significant top that's going to last for quite some time."
On the Magnificent Seven, the dispersion was the observation. Meta and Microsoft had run hard, Google and Amazon had been weaker, and Nvidia had come back close to its prior highs
10. Only 204 Inflation Calls
The show ran FactSet's earnings-call data, from John Butters' Earnings Insight, and both hosts found the headline number implausibly low.
The count has been falling for four years. "The term inflation has been cited on 204 earnings calls conducted by the S&P 500 companies for Q2 2026. That is below the previous quarter's number of 217 and below the recent peak number of 410 in Q2 2022."
Adami did not believe it. "Liz, how can it only be 204 companies in the S&P 500 on their recent Q2 calls have spoken the word inflation? It seems like that number I would have thought double that to be honest with you."
Thomas agreed and gave a one-line explanation. "Yeah, that does feel a little bit low. I think they're too busy talking about AI."
She also thinks the subject has moved off the corporate agenda and onto the household one. "Also, I think inflation right now is more of a consumer story than it is a business story and consumers are talking about it a lot. We get University of Michigan consumer sentiment on Friday and I expect that to be bad because it's usually skewed more towards inflation."
A hike would put the word back on the calls. "I think if the Fed hikes rates and cites inflation as a problem, we're going to hear a lot more of companies talking about this and perhaps using it as an excuse for lowered guidance."
The sector split supported her industrials worry. The data had industrials citing inflation most often, at 48 calls, with health care showing the largest quarter-over-quarter decline. Thomas: "So, that's a sector that I think is vulnerable right now and has been one of the sectors in the market that we've hung our hat on, showing strength this year."
What would change her mind on the economy. "If industrial stocks continue to roll over and we get manufacturing data that weakens, then you have to ... sort of perk your ears up about is the economy really all that strong."
One host's counter-argument was that the fall in mentions is rational. Inflation is still rising, but rising more slowly than at the 2022 peak, so chief executives and finance chiefs have adjusted to it rather than stopped noticing it
11. What a Hot CPI Would Do
With the CPI print due the next morning, Adami asked what markets do with a hot number. Thomas's answer made it conditional on oil rather than on the print.
The pairing is what matters. "Well, if a hot number in CPI is paired with a drop in oil prices, I think the markets are totally fine because CPI is backward-looking and it's a little bit lagged and really the market is worried about oil prices."
She expects a warm print. "I would expect CPI to come in a little warm."
And she thinks it should be looked through. "But if you get a fall in oil prices in the subsequent days, I think it sort of gets ignored" — which she gave as another reason the Fed should not hike the following week
Adami closed where he started. His point was that inflation was a problem long before crude oil became the story, that other forces are at work in bond markets globally, and that "I still think yields head higher here" whatever CPI prints
Bonus Insights
Apex Fintech Solutions' strategist Mike Tracy supplied charts the hosts walked through, including a log chart of the 10-year yield starting on 10 September 2001 — the note was written for the 25th anniversary, and the hosts' reaction was that the yield looks ready to break out
The same note carried Nvidia's 2001 chart, from when it was a gaming-chip company that had just won the Xbox GPU contract from Microsoft, at a split-adjusted price the host put at around a dime
Adami's caution on the level of yields is that history is not a comfort here: "what we haven't had is the debt levels we currently have and it's a rate of change thing as well." His argument is that a generation of investors was trained on zero interest rates and the move off them has been fast
The hosts recalled Rick Santelli coming on the programme in October 2023 with a chart board and talking about yields eventually reaching 10% or 12% — which, they noted, marked the short-term top in yields at the time
Thomas said her own podcast's most recent episode was with the financial journalist Zach Guzman on blockchain, crypto and tokenization, a subject she said she learned a lot about; Nathan said tokenization of stocks and other assets now looks like one of the few pillars of the crypto case people are coalescing around
The episode was recorded on the 25th anniversary of the 11 September attacks, and closed on it: Adami had met Fred Guttenberg, who lost his daughter at Parkland and whose brother, a first responder, died of cancer, in a green room that morning, and both hosts named the calls they took on trading desks that day
The hosts also previewed an interview with Deepwater Asset Management's Gene Munster on the Apple iPhone Duo, dropping the next morning, and spent a stretch on the Milwaukee Brewers' 91-and-56 record
Thomas's bottom line is that the Fed should not hike, probably will not hike next week, and that if it starts in the fall the first move will look harmless — with the real damage arriving three to five hikes later, in the half of the economy that has nothing to do with artificial intelligence.
Products, Companies & Tools Mentioned
iShares Expanded Tech-Software ETF (The software index she is long, held instead of single names because she says some of them will not survive and it is too early to know which)
Salesforce (The hosts' evidence that the market's software-extinction fear went too far; they argue the cost of ripping out a deployment is what protects it)
Anthropic (Its chief executive appearing with Salesforce's after the results was read as the AI labs signaling they are not trying to kill software companies)
Oracle (Nathan's example of a company cutting tens of thousands of jobs while funding an AI build-out with debt)
Micron and SK Hynix (The memory makers behind the DRAM ETF the hosts charted, and the names behind their call that the memory trade has put in a lasting top)
Nvidia (Back near its prior highs in the current chart, and the subject of a 2001 chart from when the Xbox contract made it a gaming-chip story)
Roundhill Magnificent Seven ETF (The vehicle the hosts used to show how uneven the recovery in the largest technology names has been)
SoFi (Thomas's firm; the weekly market note behind the late-'90s and industrials charts is hers)
FactSet (Source of the earnings-call data, through John Butters' Earnings Insight, that produced the 204 figure)
Apex Fintech Solutions (Mike Tracy's market note supplied the 25-year yield chart and the 2001 Nvidia chart)
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: