Transcript
Use with ChatGPT / Claude Auto-scroll A
Eisman introduces the week’s main themes and explains how the midterm results triggered a sharp market reaction. He lays out the negative implications for AI data center construction, related sectors, and the broader economy.
The implications of the midterms are potentially quite consequential on the AI narrative and the economy.The initial reaction of investors on Thursday was to sell in a panic and the market suffered a massive correction.The yield on core's debt jumped to 15% which is a prohibitive rate.Investors began to think that maybe not everyone will benefit from the AI story.Fiserve and FIS have been horrible stocks.Best to stay away.AMD also reported Tuesday night, AMD did deliver.This tech correction is going to hurt Anthropic's valuation for its IPO.Growth has slowed.AI capex accounts for at least 50% of GDP growth this year.A slowdown in AI capex means a slowdown in the economy.Let's get started.
Hi, this is Steve Eisman and this is the weekly rap.This is for the week ending Friday, November 6, but recorded Thursday night, November 5th.For Substack Premium this past Wednesday, November 4th, we posted an interview with Substack newsletter writer Herb Greenberg.We had a wide ranging conversation about the market and faulty narratives.This coming Wednesday, November 11th, we will post an interview with Melody Wright of Unicus Research, who is an expert in housing.We will discuss the horrific state of US housing and why Melody thinks housing prices are going to decline in most parts of the country.We hope you tune in.If you like the work we are doing, please visit us at steveisman.com and subscribe to Substack and our YouTube channel.And on to the rap.On this week's rap, we will discuss number one, and this is a big one, implications of the midterms on the AI narrative and the economy.
Number two, the change in the AI narrative because of coreweave.Number three, Anthropic amended its S1 with data that is quite revealing.Number four, a slew of earnings reports.And number five, a list of my favorite sci-fi novels.One thing I have learned about investing in equity markets is that the narrative trumps all.It trumps valuation.As long as a narrative is intact, the stocks related to that narrative work.One of the major problems with the AI narrative is that it keeps changing and also growing in complexity.It's hard to keep up.During the summer, the AI narrative was in full force.It was going to conquer the world.And then a few weeks ago, the AI narrative shifted.The yield on Cororeweed's debt jumped to 15% which is a prohibitive rate.Investors began to think that maybe, just maybe, not everyone will benefit from the AI story.
And just as we were all digesting that, the narrative shifted again this week because of the Democrats sweeping both houses of Congress.In the first part of this wrap, I'm going to discuss the change in the AI narrative from the midterms.Afterwards, I'll discuss how Core Weave changed the AI narrative prior to the midterms.You need to understand both to get the full story.Now, the implications of the midterms are potentially quite consequential, and it's going to take weeks to think through them all.At first, the market did not know quite what to do with this new information.So, on Wednesday, most stocks were up.However, on Thursday, investors began to see the negative implications and the market suffered a massive correction with the S&P down 4.4% and NASDAQ down an incredible 5 1.5%.If, and it's an if, Democrats go through with plans to slow AI data center construction, then the implications are quite large.
Many of those implications are negative, but there are a few that are potentially positive.Let's start with the negative.If AI data center construction slows, a whole slew of companies that are involved in that construction will see their growth rates slow obviously.And that is mostly why the market was down so much on Thursday.I'm going to list a bunch of companies and sectors that are negatively impacted.Believe me, the list is not comprehensive, but it does give you a flavor of how many sectors this touches.Those sectors and companies include one power suppliers like GE, Vernova and Quanta.Two semiconductors like Nvidia, AMD and Micron.Three data center equipment providers.This is a very big group and includes companies like Dell, Super Micro, Arista and Cisco.Four, industrial companies that are involved in AI data center construction like Eaton, Verdive and Envent.Five, Neocloud companies, Cororeweave, Iron, and Nebus.
And six, all of Elon Musk's companies.The Democrats hate Musk and may try to throw roadblocks up against anything he wants to do.The initial reaction of investors on Thursday was to sell in a panic.I'm guessing that a bunch of hedge funds were liquidated just like the situational awareness hedge fund run by Leopold Ashen Brener months ago.Now admittedly it is very early to think through all the implications of this correction.However, one of my preliminary thoughts is that a slowdown in AI data center construction could eventually be a positive a pretty big positive actually for the hyperscalers Microsoft, Amazon and Google.Why do I say that?The torid pace of construction has caused these companies to raise massive amounts of capital and investors are frankly growing sick of it.A slower pace could mean that these companies could return to positive cash flow, perhaps even in late 2027, thereby enabling them to fund AI data center construction purely out of cash flow and at a more sustainable pace.
It could also mean that the AI construction story grows slower but for longer.In my view, that would be a major positive.Now, I really don't like how much the AI story has become dependent on the capital markets.For me, the less dependent, the better.Now, before we turn to how Core Weave changed the AI narrative weeks ago, which frankly seems almost like ancient history at this point, let's see how all the stocks I mentioned did on Thursday, one at a time.GEV down 7 12%.Quanta down 12.3%.Nvidia down 9.3% AMD down 17.3% Micron down 17 1.5% Dell down 15% Super Micro down 8 12% Arista down 10.8% Cisco down 7.6% Eaton down 7 12% Verdive down 14.1% Envent down 12.8 Coreweave down 9.8% 8%.Nebus down 14.9%.Iron down 17%.Ouch.SpaceX down 9.1%, Tesla down 10.8, Microsoft down 3.6, Amazon down 2.2, and Google down 1.9.
Clearly, the hyperscalers held up the best.One more point, AI capex accounts for at least 50% of GDP growth this year.A slowdown in AI capex means a slowdown in the economy.Not necessarily a recession, but at the very least a slowdown.And because of those fears, the banks, investment banks, and alternative asset managers were all down on Thursday.Now, let's turn to how the AI narrative changed prior to this week because of coreweave.
This chapter examines how CoreWeave exposed the fragility of the neocloud model and how hyperscalers are also becoming dependent on capital markets. Eisman then analyzes Google’s search slowdown and Anthropic’s amended S-1, arguing that growth has cooled and valuation risk is rising.
Prior to the midterms, there was a major shift in the AI narrative that is still very important to understand.It wasn't so much that the narrative itself changed, but who is controlling that narrative did change.We saw it weeks ago with coreweave debt being repriced from a yield of 9% to 15%.At a cost of debt of 15%, building AI data centers becomes prohibitively expensive.This movement in coreweave debt made clear something that should have been clear to everyone months ago which is the neoclouds are the weak players in the game of AI data center construction.What do I mean by that?The neoclouds cororeweave, Nebius and iron don't have other businesses.All they do is raise debt and take the proceeds to build AI data centers that they then rent out.They are competing with the hyperscalers Amazon, Google and Microsoft that have massive businesses outside of data centers that can support AI data center construction.
The Neoclouds are completely dependent upon the goodwill of the fixed income markets.And the fixed income markets don't really care about the romance of the AI story.They care about facts.They care about cash flow.They care about math.As Sergeant Joe Friday said in Draget, quote, "All we care about are the facts, ma'am."
Just the facts, if you don't mind, ma'am.
Now, while it's true that the hyperscalers have other businesses that support AI capex, the fact remains that AI capex has dramatically altered their businesses from capital light to capital intense and from cash flow generation to negative cash flow.This quarter the free cash flow of Google and Amazon was negative9 billion and -33 billion.All the hyperscalers have been raising capital to fund AI capex and in some cases they are raising equity capital during the last week of October.Meta for example began a process of raising 90 billion in equity because even the hyperscalers have to raise capital.Here too the AI story has become partially dependent on the whims of the capital markets.Now, that could change if the change in the political environment causes AI data center construction to slow and the hyperscalers then return to positive cash flow.We shall see.Before we turn to anthropic, I want to review last week's earnings report from Google because it had potentially large implications for Google's future.
Google missed on most metrics, but that was not the most important takeaway.Google's search revenue was up only 12%.Which was a disappointment.By way of comparison, in the first quarter of 2026, search revenue grew 19%.Now, why is that important?The fear surrounding Google for some time has been that AI search would replace Google search.Now, that sentence doesn't sound so bad until you think it through.The Google search business is a capital light high business.It's an amazing business.so amazing that most of us could not until recently even contemplate not using Google search.It's its own paradigm.But it is probably being replaced by a highly capital inensive low business.The capital intensity is the spending on AI data centers and the spending to create LLM and agentic AI models.I call it a low business because switching costs between LLMs looks inexpensive.
The Google search business is just a much better business.And the results indicate that Google search is being replaced.Yes, slowly, but replaced.Now, on Monday of this week, Anthropic filed an amended S1.The original S1 had numbers only through June 30th, 2026.I pointed out that Enthropic had benefited from token maxing, which ended this summer.The Financial Times published some revenue and expense numbers that confirmed unofficially that Anthropic's revenue had slowed post token maxing in the third quarter.The amended S1's revenue figures were in line with the Financial Times.So, let's quickly reanalyze the numbers.I'm putting up a table that shows Enthropic's quarterly revenue from the third quarter of 2025 through the third quarter of 2026.Prior to the amended filing, all the commentary in the traditional media was about how great the numbers look through the second quarter, but it's the third quarter that I think really counts.
I also show the company's compute and infrastructure expense during the same period.I then calculate Anthropic's efficiency ratio defined as its revenue divided by the compute expense.And I also show the sequential percentage change in quarterly revenue.For those of you who are listening only, a clear picture emerges and it is what I have feared.During the first half of this year, I thought that Enthropic and Open AI were benefiting from token maxing.That was where enterprises told their employees to use AI and to hell with the cost.That ended this summer when costs got out of control and companies now had the option to switch to cheaper openweight models.I assumed that once token maxing ended, growth would slow.Still growth but slower.And that is what these numbers seem to show.If we go back to the fourth quarter of last year, just before token maxing, Anthropic's quarterly revenue grew sequentially by 43%.
Its revenue divided by computer expense was 0.69 times.Then token maxing took off and so did Anthropic.Q126 and Q226 sequential revenue growth jumped to 135% and 147%.Wow.By the second quarter of 2026, the revenue divided by compute expense ratio had jumped to over two times from 0.69 times in the fourth quarter of last year.So this all looks great, but now we get to the third quarter.Token maxing is over.Anthropic is still growing but slower.Q326 sequential revenue growth was 47% versus 147% in the second quarter.More alarming, the revenue divided by compute expense ratio in Q3 declined to 1.72 times versus 2.08 times in the second quarter.Growth has slowed.It's also important to point out that Anthropic cut its token pricing by as much as 50% in October so as to spark demand.According to Open Router, Anthropic's token demand has increased, but I doubt enough to offset the price cuts.
It looks like OpenAI is following and Google is also following in its pricing of its new Gemini LLM.It looks like a full-blown price war is happening as we speak.One more point on entropy.The amended S1 reveals total leasing and all other AI related commitments of 895 billion.That seems to imply that sometime next year, Anthropic's AI compute run rate will run to 100 billion per year in expenses.It also implies that Anthropic will continue to bleed cash.Finally, just to state the obvious, this tech correction is going to hurt Anthropic's valuation for its IPO, which will happen in a few weeks.Moving on.Glean is the context platform for enterprise AI.AI needs more than a capable model to do useful work.It needs context about your business.Glean connects enterprise knowledge and people across systems, giving employees and AI agents the context they need to move their most important work forward while preserving your existing permissions and security boundaries.
Your enterprise knowledge stays independent of any single model provider.with access to 40 plus open and frontier models.Glean routes tasks based on performance and cost so you can take advantage of better models as they emerge without locking your business into one provider.Two additional points.Glean cuts AI cost by 75% per task compared to Clawude Co-work.Glean uses auto routing of multiple models to make sure you get all the AI power you need with all the cost savings your CFO wants.Also, you won't find the AI model that's always perfect for your next task.The good news is you don't have to.Glean uses context and signals across your business to automatically route every task to the right AI model.I've got a very good friend and former colleague whose company uses Glean and he loves it.Glean, the context platform for enterprise AI.
Visit glean.com/isman to unlock an exclusive offer just for listeners of the real Eisman playbook.That's glean.com/isman.Terms and conditions apply.
Eisman reviews major earnings reports and explains why many results matter less than the broader AI-capex debate. He covers Palantir, Apple, Eaton, payment companies, Apollo, Booking, AMD, Arista, Qualcomm, and FICO, highlighting which businesses are strong, which are under pressure, and how AI exposure is reshaping valuations.
In the continuing saga known as Oracle, and it is now a very sad saga, the company began the process of raising 10 billion in debt and equity as well to help fund its AI data center construction programs.The market has become petrified of Oracle's balance sheet and its need for capital.And the stock was down 9% on Monday when that news was announced.One piece of news in the world of the telep.I created that word.I'm very proud of it.On Wednesday, SpaceX announced a deal with Qualcomm to provide key network technology for Starlink's mobile service.Looks like Elon Musk is going fullin to challenge the three mobile providers, which are T-Mobile, AT&T, and Verizon.And while this is going on, I think the big three are just uninvestable.Now, let's turn to earnings.This week's earnings, in my view, are just less important than last week.
The major issues in the AI debate are one, capital intensity and negative cash flow.Two, the fixed income markets turning on the NeoClouds.Three, investors are scared of Oracle's balance sheet.Four, potential delays in AI data center construction because of a lack of power.Five, a lack of moes with respect to LLM and Agentic AI.Six, an LLM price war.And seven, how much will a change in the political landscape impact AI data center construction?That's quite a list.And these are the key issues.And while this week's earnings are informative, they have little to do with these major issues.That's why results at AMD or Arista, for example, only show the state of current AI capex, which we all know is strong.Let's get started, though.First up is Palanteer.company reported Monday night and the results were strong.Revenue of 2.3 billion grew 96%.
It's impressive.EPS beat as well and there is nothing to really criticize here.Palunteer's role in the AI world at least at this point seems to be helping enterprises get their databases in order so that the enterprises can then use AI.Palanteer is also trying to position itself to sit on top of enterprise databases so as to protect their data from the LLM providers.These projects should take years.Now, despite these good numbers, Palanteer was down 8% after hours Monday night.Why?In the end, Palunteer is a partial AI play and the market has been freaking out about all AI plays.Good results or no?Apple also reported Monday night.I'd call it a mixed quarter.An okay quarter, but mixed.Revenue of 114.5 billion was up a nice 12% versus last year and a beat.Earnings per share of 210 was up 14% versus last year and also a beat.
So far so good.However, the allimportant service revenue slowed to 10%.The stock was down after hours Monday night because Apple guided service revenue growth for the next quarter below 10% and margin guidance was light as the company continues to deal with memory chip price inflation.I'd note that Apple really is not part of the AI debate.It is not building its own LLM models and that is why it still has very very strong free cash flow.It looks like Apple is going to use its iPhone position, its insurmountable iPhone position, to make other LLM entities rent space on its platform.It's a potentially lowrisk, highreward strategy, and that is also why Apple's results tell us nothing about what's going on in AI.On Tuesday, the industrial company Eaton reported, "Full disclosure, I have owned Eaton for quite a while.Eaton is an electrification company with about 20% of its revenue coming from AI related construction.
Because of that, the stock has done well over the last few years with the stock up 15% year-to- date, although that 15% was prior to the Thursday correction.Eaton reported earnings per share of $3.68, up a nice 20% and a beat.Sales were 8.95 billion, up 28%.Impressive and also a beat.For 2026, the company raised guidance.These are very good numbers.And on these numbers, the stock was up 6% on Tuesday.Moving on, FServe, a financial services nonAI name, reported on Tuesday.And for the first time in a long time, the results were not a disaster.Fireserve is in the payment space and it's one of the older companies that has been losing share to the newcomers.It's had a revolving door of CEOs and there has not been any good news in a while.This quarter, the company reported earnings per share of $1.75 versus the consensus of $1.72.
So, a beat.It's an EPS for the first time in quite a while.Now, revenue was 4.85 billion versus the estimate of 4.88.So, a miss, but not calamitous.Organic revenue growth was 0% and Fiserve gave decent guidance.On these numbers, the stock was up 12% on Tuesday.I'd call this a relief rally and that the results were not an utter disaster.But to me, this is not a long-term investment.In my view, the payment space is just too tough.Outside of Visa and Mastercard, the payment space has few barriers to entry and the companies in the space are all entering each other's markets.And just to re-emphasize how difficult the space can be, FIS Five serves chief competitor reported on Wednesday, one day later, and the actual results were not terrible.Earnings per share was $161, which was up 6.6% and revenue was up 3.7%.
Not great, but not horrendous.That was not the news.The news was that the company announced it was pausing stock buybacks and some deal making to focus on deleveraging the company.The company will only resume stock purchases once it gets its gross leverage down to 2.8 times.Mathematically, that probably means that there will be no stock buybacks until sometime in 2028.Also, the company announced that the CFO is stepping down immediately and nobody likes to see that.FIS also cut its EPS guidance for 2026.Not good.And the stock was down 24% on Wednesday.FSER and FIS have been horrible stocks.Fiserve reached a peak of 235 in March 2025 and it is now 62.FIS has also been a disaster reaching a peak of 122 in February 2022 and is now around 30.Best to stay away.Apollo reported and the market was not crazy about its results with the stock down 5% on Tuesday.
Apollo has two major sources of income fee related earnings FRE and spread related earnings SR which is interest income.Apollo showed 23% FRE growth, 53 billion of inflows, and 18% assets under management growth.Good.More importantly though, because Apollo owns a large insurance company, Athen spread related income, S sur plays a very important role in how people analyze the company.And here Apollo missed expectations with S up just 7% year-over-year as the company probably took a very conservative credit approach for new investments.Overall, this was a mixed quarter.And it's also important to note that Apollo is now one of the largest financiers of AI data centers.So it is now a partial AI play.And that helps explain why the stock is down 19% this year.Bookings reported Tuesday night.Bookings is one of those companies that people worry about because of Agentic AI.
Investors worry that the brand name of companies like Bookings will deteriorate because all Agentic AI cares about is price.It's too early, however, to see any real impact from Aentic AI.However, it is not too early to see the impact of a weakening consumer.Bookings reported earnings per share of 466 up 17% versus last year and a beat.That sounds good.However, revenue of 8.89 89 billion was down versus 9.01 billion a year ago and it was also miss versus expectations and the company is guiding fourth quarter revenue growth of only 0% to 2%.Moving on, AMD also reported Tuesday night given AI capex one would expect strong results and AMD did deliver.EPS was $29, up 74% versus last year and a beat.Revenue of 13.38 billion was up 45% versus last year and also a beat, but only a slight beat.For the fourth quarter, AMD expects revenue to be approximately 17 billion, which is quite a jump versus the third quarter.
However, the fact that a chip company had good results, it should not be a surprise.Arista also reported Tuesday night.Now, Arista is a chief competitor of Cisco and like AMD, given current AI cap X levels, Arisa should report a good quarter and it did.Earnings per share was $120, up 60% versus last year and a beat.Revenue of 3.7 billion was up 58% also a beat.The company provided fourth quarter guidance that was higher than estimates.But like AMD, Arisa's results are not going to settle the AI debate.They merely provide a window into the high level of current capex.Wednesday night, Qualcomm reported and the results were okay, but the guidance was not.Earnings per share was 217, which was in line, but down 28% versus last year.Revenue beat expectations, but was down 8% versus last year.The guidance for the December quarter was a touch light.
Why?Qualcomm's biggest market for its chips is cell phones.And because of the dramatic increase in memory chip prices, cell phone sales have been hurt.And Qualcomm has raised prices on its chips.But the dimmunition in cell phone chip volume has hurt more.That's why Qualcomm's handset chip sales were down 26% this quarter.FICO reported Wednesday night and the report was fascinating.Not the results, those were fine, but the guidance was not.Total revenue grew 27%.That was good.Earnings per share was 1159, up 43% versus last year.Also good.The guidance not so good.FICO's fiscal year ends September 30.So when it reports that quarter, it always provides revenue and earnings guidance for the next fiscal year.Not this time.FICO refrained from giving any guidance at all.Why?Given the tumultuous changes in mortgage scoring, management intimated that in light of competitive pressures from Vantage Score, it would not provide guidance.
FICO management said is working on multiple pricing programs and looking to retain as much volume as possible.Prior to yesterday, there were only two times in the company's history that it refrained from giving guidance.during the great financial crisis and during COVID.In my view, this is a company that has lost its precious monopoly in less than a year.That is embarrassing and FICO needs new management desperately.
The episode closes with Eisman’s science fiction reading list and a brief mention of upcoming interviews and a special AI master class. He ends with subscription reminders and the standard investment disclaimer.
And now for the mailbag.On a recent appearance on CNBC, I said that growing up, I had read a ton of science fiction novels.I said this in the context of also saying that I thought that many techies have read a lot of sci-fi and view it as a road map for the future.One viewer asked for my sci-fi list.Before I do that, I just want to say that I began reading sci-fi when I was in high school and continued reading it well into my 30s.Thereafter, I mostly switched to the fantasy genre.So, this week, I'll provide my sci-fi list, and at some point in the future, I'll give you my fantasy list.It's not every sci-fi novel I've ever read, but the list is composed of many of my favorites.So, here it goes.One, 1984 by George Orwell.Two, Fahrenheit 451 by Ray Bradbury.
Three, Brave New World by Aldois Huxley.Four, The Dune series by Frank Herbert.Five, Hitchhiker's Guide to the Galaxy series by Douglas Adams.Six, The Foundation Trilogy by Isaac Asimov.Seven, The Enders Games series by Orson Scott Card.I Robot by Isaac Asimov.Flowers for Algenon by Daniel Keys.Animal Farm by George Orwell.Stranger in a Strange Land by Robert Heinland.Footnote: The Origin of the Word Grock comes from that novel.Childhood's End by Arthur C.Clark.The Forever War by Joe Holdman.The Stand by Stephen King.It's my favorite horror novel of all time.Red Mars series by Kim Stanley Robinson.Riverworld, which is a series by Philip Jose Farmer.One of my favorites, Red Rising Series by Pierce Brown, The Three Body Problem Series by Leu Sixen, I hope I pronounced that correct, The Time Machine by HG Wells, and Watchmen by Alan Moore, which is a graphic novel.
This week on Monday, November 2nd, on our free service, we release an interview with Aiden McCon Ewing of the popular podcast called The Lemonade Stand.We had a wide- ranging interview about the affordability of housing, especially for young Americans, the AI story, and many other themes.On Monday, November 9, we will release a special episode.It's a master class in which I delve into all aspects of AI.Think of it as everything you wanted to know about AI, but were afraid to ask.The best way to support the real Eisman playbook is to subscribe to Substack through Steveman.com and to YouTube.and we greatly appreciate your support.And that's the wrap.This podcast is forformational purposes only and does not constitute investment advice.The hosts and guests may hold positions in stocks discussed.Opinions expressed on their own and not recommendations.Please do your own due diligence and consult a licensed financial adviser before making any investment decisions.