Sep 26, 2026 · 1h 11m
High bond yields expose weakness beneath resilient indexes
Market Starting To Crack Under High Bond Yields? | WMR 9 24 26
The episode examines whether restrictive real rates, concentrated AI spending, and a frozen labor market are signaling broader economic fragility.
- 1Narrow technology leadership may be masking deterioration across economically sensitive sectors as borrowing costs rise.
- 2AI infrastructure spending could support headline growth while creating payoff risks if demand, pricing, or technology changes quickly.
- 3Five-year Treasuries offer income and defensive liquidity as investors reassess portfolios amid higher yields and geopolitical inflation risks.
Don't miss
Mike Leibowitz explains why five-year Treasuries can provide income, potential capital gains, and liquidity for buying stocks after a major decline.
The brief
Mike Leibowitz and Adam Taggart argue that calm major indexes conceal stress in economically sensitive sectors, where high interest rates are already affecting housing, consumers, and businesses.
A small group of technology stocks is carrying the market, while AI infrastructure spending may be inflating growth even as the broader economy stagnates and hiring freezes.
The AI buildout faces a telecom-era payoff test: hyperscalers must generate enough revenue to justify data centers, while cheaper models could compress usage, prices, and earnings.
Rising bond yields are making real rates more restrictive, complicating the Federal Reserve’s choices and increasing pressure on indebted companies, employment, and valuations.
Leibowitz describes bonds as defensive optionality, favoring five-year Treasuries for income, potential capital gains, and liquidity that could fund stock purchases after a major decline.
The discussion closes by weighing reduced mega-cap exposure, greater value and utility positions, and geopolitical risks involving Iran, oil, shipping, inflation, and yields.
What was said on this episode
14 statements · 2 positive · 9 negative · 3 mixed
Rising rates will hurt most companies not yet benefiting significantly from AI, causing layoffs.
“ultimately rates went out, right? If rates just keep going up, ultimately they went out because most companies that are not involved, that are not reaping the benefits of AI, which is almost every company, right? I mean, there are very few that are really profiting right now, generating significant revenue from AI are going to suffer. And that's where layoffs occur.”
Listen at 14:45
AI data-center construction has limited employment impact but significantly boosts reported GDP.
“A lot of it is just due to AI data center build out, which is very small. And it's not really impacting that many people, but it is showing up in the bottom line pretty significantly.”
Listen at 15:27
Telecom fiber-optic investments eventually generated revenue but bankrupted many investing companies.
“It's the telecom fiber optic build out all over again that these companies spent a ton of money. And the revenue from that ultimately came. And those investments by those companies, many of which are bankrupt today, laid the groundwork for what we have today for this whole Internet network and ultimately for AI.”
Listen at 18:58
Oracle is vulnerable because its substantial debt trades at junk levels.
“Oracle is potentially the one, you know, the one kind of problem because they do have a lot of debt. Their debt is trading at junk levels.”
Listen at 20:36
Hyperscalers may reduce capital expenditure and data-center construction if revenue disappoints.
“they may just say, OK, we're getting ahead of ourselves. We're spending too much. We're not seeing a revenue. Let's hold off. Let's just build five data centers instead of seven. Right. Let's reduce our CapEx plans.”
Listen at 21:28
Reduced hyperscaler spending is a major market risk over the next year or two.
“that I think is the biggest risk. for one of the big risks for the market in the coming, call it, year or two.”
Listen at 21:46
A 20–30% slowdown in AI capital expenditure could trigger a 20–25% stock-market correction.
“a 20 or 30 slowdown in that that's going to be really material i mean that that could easily send the stock market into a 20 25 correction i think right”
Listen at 22:50
Computing technology has repeatedly become smaller and more efficient under Moore’s law.
“We have proven over time, especially with computing, that we're able to reduce the size. This is Moore's law. We've been reducing the size and adding to the efficiency.”
Listen at 23:03
More efficient computing could reduce data-center spending and leave existing facilities obsolete.
“all that money going into data centers just starts shrinking over time. It's less of an economic impact. It's better for those companies. But they may now have very outdated data centers that are just don't make sense.”
Listen at 23:38
Greater use of efficient or free AI models may substantially reduce compute demand.
“the demand for compute may actually go down pretty substantially as companies kind of balance shift which type of AI model they're going to use.”
Listen at 25:02
Core inflation excluding food and energy is the most useful inflation measure.
“I think you and I hate to say this because I always laugh at the Fed for saying it, but you really need to look at core inflation, exclude food and energy.”
Listen at 42:09
Core inflation remains above the Federal Reserve’s 2% target.
“It's still above the 2% target.”
Listen at 43:40
The Federal Reserve cannot reduce oil-driven inflation without significantly harming the economy.
“oil is having a big impact. And that's a problem because it's a problem that the Fed can't fix without really harming the economy.”
Listen at 43:45
Stable inflation expectations combined with rising bond yields make real interest rates increasingly restrictive.
“if inflation expectations and I know this is contrary to how a lot of people feel, but if they're relatively stable, right, they're not really changing very dramatically, but bond yields are going up. It just means that the real rate is becoming more and more restrictive.”
Listen at 44:11
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.