Earnings, Cashflows and Stock Prices: The AI Storm blows through!

0:00 Hi, welcome back. About a month ago, I did a session on rising interest rates in the US and across the world and how equities have been surprisingly resilient that they've held their own and actually gone up even as rates have increased. And one reason I noted was that in the aggregate at least US companies are reporting much higher earnings. That led me to think about a an issue that I think we all need to consider more seriously. We know that companies are spending trillions of dollars on AI on AI capex and we hear the stories about how this capex is good and bad and also stories about the accounting consequences. In fact, one story I've heard is it's the AI capex that is causing the increase in earnings at US companies. And I'm befuddled because when companies usually invest in capex, it lowers earnings. It doesn't increase them. How much it lowers earnings will depend on how you amatize the capex. So clearly I'm not looking at the complete picture. So in this session, here's what I plan to do. I plan to set the table by first updating what I did in the last session by bringing in what happened in September to interest rates. And if you thought interest rates were going up in the previous eight months of the year, you should take a look at what happened in September. So, I'm going to update the rates in September and what stock prices did and also estimate a measure of the price of risk and equity markets at the end of September.

1:31 Then I want to turn my attention to the nuts and bolts of how AI capex is playing its way through the financial state of the companies, both of the companies making the capex. And here I think is the missing piece as well as the companies that are benefiting because they supply the ingredients for the capex. So let's get the show on the road. As I as I noted at the start, this has been a year of rising rates. The Treasury bond rate during the course of the year has gone from 4.18% to more than 5%. At the same time, stocks are reporting much higher earnings. So you got higher rates and much higher earnings. And at least up till now, the higher earnings effect is dominating the higher rate effect. So it behooves us all if we're investing in equities to take a look at why earnings are rising, where is the increase in earnings coming from and is it sustainable. So as I said, this session is about focus on focusing on those nuts and bolts. So let's start with the interest rate story. last session that when I talked about interest rates, it was the end of August and I talked about how rates had risen already during the course of the year and I might have used the word slowly because the rate rise had been gradual. Well, in September that changed rates continued to rise but this time they climbed much more quickly. In fact, during the course of September the 10-year T- bond rate, I'm going to keep that as my focus increased from 4.75% to 5.29% 29% a 0.54% increase and that increase was across the board you're saying so what.54% is not a big deal is it in fact to see how big a deal 0.54% is here's what I did I went back all the way to 1962 start of ' 62 and every month I computed the change in the T- bond rate so this is a histogram nothing know nothing special about it but I wanted to see how unusual is a 0.54% increase in rates. As you look at the distribution of changes in rates, you can see that the 0.54% would put you in the top 10% of months.

3:41 I mean, there have been months where rates have gone up more than 0.54%, but this was an unusually unusual month in terms of the increase in rates. Now, normally when rates increase by that much, you'd expect stocks to suffer, but at least in September, they did not. So, I wanted to focus in on equity resilience in 2026. And I want to do it in a little more detail than focusing on an index. So, here's what I did. I took all of the companies that were publicly listed on December 31st of 2025 in the US. And I'll explain in a minute why I started with that sample. And then I looked at the aggregate market cap of these companies by month, every month of these nine months of the year. So as you can see over the 9 months in the aggregate US equities have added about 12 to 13% in overall market cap but you can also see the un the that it's been pretty unbalanced. Technology clearly is is done much better than some of the other sectors. But one of the things that's interesting about this graph is it kind of maps out how how desparent this market has been in terms of how it's reacted to rates. Now, it's kind of tough to read that effect by looking at monthly rates. So, here's what I'm going to do. I'm going to break down what's happened in 2026 by quarter. And each quarter, I'm going to look at the percentage ch the the dollar change in that quarter, as well as the percentage change in 2026 by sector. So, let's look at the percentage change by sector in 2026. First, this column, let's look at the best performing sectors. Technology is up there, but you also see energy and no surprise there. Energy is up collectively by 39%, technology is up 24 and a half%. Materials up 21%. But there are sectors that have almost done nothing during the course of the year.

5:36 Communication services, consumer discretionary, utilities. Now let me take a pause and explain why when I did these aggregate market caps I focused only on the companies that were listed on December 31st to 2025. If you look at the total market cap without that constraint the market cap for communication services has actually increased about 2 trillion in 2026 which I seem to be ignoring. You're saying that's not fair. Well here's where the two trillion came from. It came from SpaceX going public during the course of the year. Now to treat that as an increase in market cap is not quite fair because it's not like you could have bought SpaceX at zero at the start of the year and it rose to 2.2 trillion.

6:18 You'd have bought it at 2.2 trillion stayed around 2.2 trillion. So I've focused in only on the companies that started the year as publicly listed in coming up with these aggregates. Clearly some sectors did much better than others. But I also looked at another statistic because I kept reading about how how topheavy this market was and how unbalanced this market was. And to look at that balance, I looked at the percentage of stocks in each sector that went up by quarter in 2026 and over the co course of the year. So let's take over the course of the 9 months.

6:55 Technology clearly has had a very good 9 months up 24 and a.5%. But only 52% of technology stocks have been up in 2026. Put differently, the sector is up 24 1.5%. But 48% of technology stocks are actually down for the year. In fact, across the market, it's almost 50/50. Half of stocks are up, half of stocks are down. So even though the aggregate market is 12.25%. It is entirely possible that if you look at your portfolio, you're seeing none of the gain because you happen to hold a disproportionate amount of your money in the 50% that went down. So that percentage of stocks up. Now, I'm not going to read too much into this. Is it good? Is it bad? It is what it is. You can see that the markets have done reasonably well over the course of the year, but it's been the effects have been dispared across sectors and probably across market caps. So, how we were to make sense of all of these different movements in equity markets?

7:53 One way I keep my sanity, and it's barely, is by computing an implied equity risk premium for the S&P 500 at the start of every month. You think, what's an implied equity risk premium? It's analogous yield to m to a yield to maturity for a bond. If you're familiar with the yield to maturity calculations in a bond, you take the bond price, you take the expected cash flows on a bond, which are coupons and a face value. And the yield to maturity is an internal rate of return. What you can expect to make on that bond if you pay that price today. If you apply that to stocks and you take the S&P 500 at the start of 2026, 7,651.54 is what you'd pay for the index. And you think about the cash flows you expect to get on stocks. It's not coupons, right?

8:37 its dividends and buybacks. I can't tell you what they will be in the future, but I can tell you what they were in the last 12 months. In this case, they're about 198.17 more from buybacks than dividends in the US. And as a final piece of the puzzle, to get expected cash flows, I bring in analyst forecast. I know, I know you don't trust analysts, but in this case, they're estimating aggregated earnings for the S&P 500, and they don't do badly. They're not that biased when it comes to know index level earnings and they're they've become increasingly optimistic over the course of 2025 or 2026. The start of 2020 start of October 2026 they were forecasting a growth of 14.59%.

9:20 On a compounded basis actually much more in the near year 2026 2027. We'll talk about why. But after year five, I bring that growth rate down to the growth rate of the economy. Now I one of the tricks I learned to use to make that forecast is rather than sitting there and wrestling with forecasting the entire economy, I use the risk-free rate as my proxy 5.29%. So I've got expected cash flows and then I look for that discount rate that makes the present value of those cash flows equal to the level of the index today. That's it. That's basically what I'm solving for. What I get as my solved discount rate is 8.99%.

10:03 You're saying, "What does that tell me?" At the start of October 2026, if you bought the S&P 500, you can expect to earn 8.99%. You might not like it. You might say it's too low, too high. This is what you can expect to earn given what you paid for the index and expected earnings in the future. And to convert that into equity risk premium, you subtract out the risk-free rate. That used to be easy until a couple of years ago, you just subtracted out the T-bond rate, which would give you 3.7% as your equity risk premium.

10:36 But to the extent that you view the US Treasury itself as risky, because after all, the US has lost its AAA rating, you'd get an equity risk premium of 3.92%. Now, part of you saying, "What do I do with that?" I'm not a market timer, but if you're a market timer, a low value for the equity risk bould you. You're demanding too low a price for risk. The end of 1999, peak of the dot boom, this number was 2%. 3.7% or if you prefer 3.92%.

11:05 Is a lower number than normal, but not by much. In fact, over the course of this year and going into 2025, you can see that the equity risk premium has kind of moved around within a range between 3.7%, the low that we have now, and maybe 4 4.2%. In fact, the total expected return on stocks has actually gone up during the course of the year. It just hasn't gone up as much as the T-bond rate has.

11:34 Markets are repricing stocks to deliver a higher return. But T-Bond rates rising more has lowered the implied equity risk premium. Make of that whatever you want because now I want to turn my attention to what I came to do this session for which is to look at the effect of AI capex in the accounting statements and in terms of corporate finance consequences. Now you heard the numbers right the hyperscalers I you know in my in my post on AI and my session on AI the six largest hyperscalers in the space which include Meta Alphabet Amazon Microsoft Coreweave and Oracle have collectively already invested 2 trillion. If you bring in the other players in the game and you update that number you're already at close to 3 trillion and that number could get as high as 8 to 10 trillion if you look at the commitments they made. So let's start with that as a fact. Companies are investing massive amounts in capex.

12:34 Let's call these pe these companies the builders. They're investing. Now when you think about investing in capex, capex goes into your balance sheet as an asset, right? Not it's not like I'm giving them credit. It show it's just the accounting shows up as an asset. And when your balance sheet assets go up, something has to change on the other side, either debt or equity. And that effect will depend on how the AI capex is being funded.

13:02 Now if you look at the big hyperscalers especially the ones in the max 7 it's about a third of the financing is coming from debt 2/3 from equity. You're saying where are they getting the equity? They have cash businesses. Alphabet is 175 billion coming from the advertising business that's covering a big chunk. There are other investors in AI capex who are using far more debt but your debt and equity will reflect how the AI capex gets funded. So that's where it's going to show up on the balance sheets of the AI builders. You're saying what about the income statement? There is no way the year that you do capex that your earnings are going to go up. They can only go down. And how much will they go down? They go down by how much you're amortizing that AI capex at. I've seen a lot of articles talking about the scandal in this space. They argue that companies are spreading out their amotization over too many years. That's true. That'll lower the amatization this year, but it's not going to increase earnings. It's going to make the hit on earnings lower. It is true that there can be earnings from the AI capex, but it's still early in the process. We're not seeing much of those earnings.

14:08 Now, finally, if you look at the cash flows, you've got more book value because of the AI capex. You got lower earnings. So you're saying what about cash flows? The cash flows have to go down, right? Cuz no matter how much you downplay the amotization effect on earnings, capex is capex. You're spending the money. So if you look at the builders, that's what's happening in their financials. But remember, there's another side to the transaction. If you think about what this AI capex is going into, and I've tried to get different statistics on this, but a big chunk of it is going into chips. Somebody's selling them the chips, right? Who are the biggest players in the game? Nvidia obviously is the biggest beneficiary. Taiwan semiconductors followed by equipment networking equipment and there you get companies like Broadcom that are playing a role.

14:57 My point is capex to the builders becomes revenues to the suppliers of architecture which include the chip companies, the networking equipment companies, the power companies, not the power the price of the power they sell but actually building power plants and the real estate business data centers. So what's showing up as capex for some companies becomes revenues for other companies in the space. And you know how that's going to play out in those companies. Those revenues become net income. How much net income will depend on the net margins. Company like Nvidia which has astonishingly high net margins. 50 60% we'll see much of it flow into net income.

15:41 And what about the cash flows? That high net income will also translate into much higher free cash flow equity. When you look at markets in the aggregate you're including both the builders and the suppliers. Right? So if you look at the aggregated effect, the capex is the capex. But you look at earnings, the earnings effect will depend on what the net margins are of the suppliers relative to the amortization. And if the amotization is low and the net margins are high, which is the case now, the earnings collectively for companies will increase as a consequence of AI. That's why AI capex will increase earnings.

16:16 It's not at the hyperscalers, but it's collectively across companies. But collectively across companies, cash flows have to decrease. And it's just common sense. If you're investing 2, three, four trillion. That's got to come out of the collective cash flows. So you'll have lower cash flows. The hyperscalers will see the big effect. That effect will dominate whatever free cash flow equity increases you see at the suppliers. So if you are looking at aggregate numbers, I'd expect aggregate invested capital to rise, aggregate net income to drop and aggregate cash flows to be lower. So let's look at the numbers and see if that's what's happening. Let's start with the capex part. So here I focused on the capex and statement of the statement of cash flows. Before I start though, two caveats. One is I'm looking at only publicly traded companies because that's what those are the companies for in which the data is publicly accessible.

17:11 You're saying so what? There are private businesses in the AI capex space, suppliers and builders which are pretty big, right? I mean, I'm not going to be mysterious. Anthropic and Open AI are not in my sample because they're private companies. SpaceX in my is my sample only a portion of the period. So to the extent that those private companies have capex and revenue effects and earnings effects, I'm missing that in this table. The second is some or a big part of this AI capex depending who you talk to is off balance sheet in the form of joint ventures and offbalance sheet entities that might not be percolating all the way through the accounting statement. So what you're going to see is an incomplete effect but it's still the effect I'm going to focus on because that's a number you're going to see out there as the public number for companies.

18:01 So here I've gone back to my sector breakdown and I've looked at the capex in the last calendar two last two calendar years the last 12 months and in the first two quarters in in the first in the second quarter of 2026 relative to the second quarter of 2025 just to see how much numbers have changed. Let's take a few examples. Let's take technology. The second quarter of 2026 is is 80% higher than the second quarter of 2025. shows you the rate of change even with my incomplete information.

18:34 Communication services and consumer discretion are also seeing big jumps. But again, you can see the variation across sectors. Some sectors there's hardly any change. Some sectors huge change. Collectively, capex across companies is up 36% on a quarter-to- quarter basis if you focus just in the second quarter and 26% if you look at the last 12 months versus the last calendar year. Capex is growing at US companies. How is this showing up in in in in the balance sheet? It's showing up as higher book equity, higher debt, and higher invested capital. Book equity is up 13% in the aggregate across all companies, almost 1.8 trillion.

19:15 Debt is up $1.9 trillion, which is about 8%. And invested capital is about up about 3.7 trillion, 13%. That capex is showing up as higher book values. Again, I'm not rewarding or punishing companies, but this invested capital is going to carry forward. And when you compute things like return in equity and return invested capital for the entire market in the aggregate, that number will be affected by this rise in invested capital, rise in book equity, rise in total debt. So notice the debt did go up a little bit more than the equity. At least in the aggregate, more of the funding for this capex is coming from debt than equity.

19:58 What about the earnings effect? Again, let's focus in the aggregate. In the aggregate, and I've done three comparisons here, which first is the last 12 months versus the calendar year 2025. And in the aggregate, earnings are up 49% on that dimension. I've also compared the second quarter of 2026 to the second quarter of 2025, and their earnings are up again 49%. First quarter 20. So clearly across the board, you're seeing this big jump in earnings. What does that tell you? At the moment at least the earnings at the suppliers of AI capex the boom at those companies is offsetting the amotization.

20:36 Now the people who are arguing that amotization lives are being set artificially high are right. Maybe we're overstating the earnings effect but my guess is even if you adjust to the amotization life you're going to see a pretty big jump in earnings. Looking across sectors again you can see that the effects vary across sectors. I mean technology has seen earnings go from 531 billion to 727 bill. So from calendar year 2025 to the last 12 months you're seeing an increase of almost $200 billion in earnings. So on every comparison you can see variations across. But here again there's something that is worth watching. Even if you take a sector like technology that in the aggregate is seeing a big jump in earnings and you look at the percentage of companies within technology that are reporting higher earnings less than half of all companies. So the story continues to what you saw in the market cap collectively the market is reporting higher earnings but individually those higher earnings are not spreading across companies it's in a subset of companies.

21:48 So now let's move on to the final effect which is you've got higher capex higher book values perhaps de higher earnings at least in the aggregate what happens to cash flows I think the cash flow effect is much clearer collectively cash flows at companies will have to decrease when you reinvest more because that reinvestment whether it comes from the buyers or the sellers of that capex ultimately comes out of aggregate earnings. So even if you report higher earnings, you have higher capex offsetting those higher earnings. Now rather than track out the direct cash flow effect, I'm going to look at what companies are doing in terms of returning cash flows to shareholders, the form of dividends and stock buybacks. Again, in this table, you have the aggregated dividends and buybacks across all US companies broken down by sector. And I've looked at the last calendar year, the first two quarters of 2026, as well as the last 12 months. So let's take dividends. Collectively, US companies returned about $857 billion in dividends in the last 12 months. That's up about 5% 6% from where it was in the last calendar year. Not bad, right? But not not clearly not the 45 50% you're seeing jump in earnings. You see, what about buybacks? In the last 12 months, buybacks have been about 1.3 trillion.

23:09 That's about up about 8% from the 1.2 trillion that it was in the last calendar year. And in fact the second quarter of 2026 saw slight dip from the second quarter of 2025 in terms of buybacks. Buybacks are growing but again at a rate far less than earnings. In fact interestingly if you bring in stock issuances net them out against buybacks the the growth in net buybacks buybacks net stock issuances has been about 5%. Again, big differences across sectors, but you can see that dividends and buybacks are growing at rates far lower than the growth anatomics. You know where this shows up? If you track how much companies are returning to their shareholders, which I do for at least for the S&P 500, and measure it up against earnings for much of the last 25 years, starting around 2004, US companies have been returning higher and higher percentage of their earnings to their shareholders.

24:03 80 90 even more than 100%. In the last three or four years, you've seen the divergence. Earnings have continued to surge at much higher rates, but the cash flows have not increased at the same rate. The percentage of earnings that's returning to share that's being returned to shareholders is much lower. It's about 63%. That reflects the fact that that capex and AI is draining cash flows might otherwise have gone to shareholders. So at this point we have a sense at least of the big picture which is AI capex is creating a negative cash flow. It's increasing book capital. In the aggregate it might actually be pushing up earnings at least in the near term.

24:45 And in the aggregate it's pushing down cash flows that companies have available to return to shareholders. Yeah. Which sets us up for a broader question which is all of this is being done in the hope that AI will pay off. Let's play it out. Let's say the AI boom is real, that there is a payoff coming, that there is a big market out there. Here's what would you expect to see. You should see the earnings at the AI builders start to rise over time. Right now, they're not making much in in revenues and earnings from their AI capex, but that should change if AI is a real is the real deal.

25:20 Your earnings should go up. Does that mean that these companies market cap should go up? Not necessarily. And here's why. When you invest two 2 and a half three five trillion in AI capex making 50 billion in income is not enough right because that capital require has to earn at least its cost of capital let's give it an 8% cost of capital not an unreasonable number 8% of 5 billion is 400 billion if you don't generate $400 billion in earnings and cash flows you're not making enough of a return to justify the cost of capital so AI might succeed in terms of being real and delivering during a boom. But if companies collectively have trouble earning their cost of capital, the market value will have to get written down. The accountants will not write it down because earnings are good and therefore you're making money. There's no write-offs. The cash flows should start improving in these companies because as the earnings start to come in and capex starts to level off, the net capex will start to get lower. The cash flows should come back. But the builders even if they emerge as successes from this will be a will be very different as businesses. They'll be capital intensive businesses. A meta Microsoft and Alphabet 10 years ago were capital light high growth businesses. If they succeed they'll become capital very different game for investors to play and for the companies to manage. That's if the boom is real. What if the boom fizzles? What if there's no revenues and earnings at the end? Then we have a problem. The first is the earnings that the AI builders expected in the long term will never materialize. The accountants will wake up and say this was a mistake and start writing down the trillions of dollars of capex, big heads. They you can call them extraordinary in one time, but they're still heads. And if the AI capex was funded disproportionately with debt, we have a truncation risk problem.

27:16 What am I talking about? that the companies might be unable to make their debt payments and the pain will be felt both by the companies which will go into distress as well as the lenders to these companies. The cash flows may eventually recover at the eye builders, but it'll take a while to get there. And the market will exact its own pain for the for from these companies. Partly because you've had all of these hundreds of billions of dollars of capex with nothing to show for it. And partly because you lose trust in management that did that and told you to believe but never delivered. So I think that both scenarios, both pathways remain open and somewhere in the middle might be the endgame but it's worth thinking about the consequences.

27:59 So you think what do I do as an investor as I see it you have four possible games you can play. One is you can go all in on the AI story winning but AI company AI is going to you know drive disruption. Not only is it going to disrupt a lot of business, it's going to happen in the near term and it's going to be total. Hey, that's the Leo Ashen Brener thesis, right? He of course created a hedge fund. He bought the companies who would benefit from the AI story winning. Sold short on the companies would be disrupted. The mistake of course that he made and I talked about this before was that he had too much conviction. So he levered up to the hilt and the leverage brought him down. So maybe you can do what Leo did without the leverage. So that's one one possible pathway. The second is you can say well crowd wisdom is still wisdom.

28:51 The market is building in a consensus in AI where it thinks AI will win and be big but it'll take longer than than the AI optimists think it will and be perhaps smaller than they think it's going to be. That's being built into the market. you can go along with the market consensus and implicitly that's what you're doing when you buy an index fund of all US equities or S&P even the S&P 500 fund is you're going with the market consensus. The third is you can trust the AI story in its broad in its broad contours and say you know I think AI is going to win but it's you know there going to be lots of barriers blocks to get over it's not going to be as profitable as people think it is which will mean you're more cautious about jumping in today because you think the market's being too upbeat about the future not assessing the risk. What will this mean? you will leave the bulk of your money still where it is, which might be in the market, but new cash coming in. You might hold as cash saying, "I'm going to wait to see how this plays out." And there's a fourth possibility. You might think that this is all entirely built on smoke and mirrors that the AI stories is going to crash and burn. And there are two things you can do. One is you can sell all your stocks and put your money in cash or if you don't trust cash because inflation is out you can look at other collectibles, cryptos, whatever strikes your fancy. I let's call that the paranoid pathway. The other is you can decide to stay in the market and try to take advantage of investors that you think are being too gullible by doing what? By selling short by by reversing the LEO story. So, you're going to sell short the companies that benefit from AI and you're going to buy into the companies that are supposed to be disrupted and hope to make money from the crashing out story. I mean, I am not going to prejudge this, but I I mean, it behooves me behooves me to tell you where I am in this process.

30:55 With the bulk of my portfolio, I'm going to go with the market consensus. I do hold five of the mag seven. Four of those companies are big AI capex companies. So whether I like it or not, implicitly I'm assuming that at least in the aggregate, we're not going to crash out. At the same time, the cash that's coming into my portfolio, I'm leaving as cash. My cash balance in my brokerage account is higher than it's been historically because I'm not sure I buy into the thesis as completely as I would need to for that money to go into the market as well. Now, in doing this, I know I've left money on the table. In fact, my Nvidia investment, I've sold over four blocks. It's probably going to keep going up, at least in the near term, and I'm okay with that because I need to sleep well at night, and I'm okay with leaving money on the table.

31:44 So, your assessment then has to depend on what not only what you think about AI, but what helps you sleep at night. One more thing, if you're, you know, I'm I'm I'm too old to think about this as a key component, but if you're younger, a key component of your portfolio is your human capital, the present value of the wages you will make, and you got to factor in what the AI threat will do to that. If you're in a business or an industry where AI is coming for you, you have a human capital impact, you've got to factor in. You got to invest in things that are not affected by AI as much. So you got to find a way of hedging at least partially the possibility that you could lose your job. We live in interesting times and you can see why the Chinese use that as a curse because interesting times can make us stressed out. So I wish you the best in thinking through this and I hope that this session has helped frame your thinking on how AI is playing out in financial statements in earnings and ultimately in how we price equity.

32:48 Thank you and take care.