AI Financing Risk – Are Banks Looking for the Escape Hatch?

By: Russ Kamp, CEO, Ryan ALM, Inc.

I’ve spent considerable time recently questioning whether the extraordinary investment in artificial intelligence (AI) will ultimately produce acceptable returns for investors. As I’ve stated before, I don’t question the technology’s potential to transform our economy, improve productivity, and create enormous opportunities. But there is a significant difference between an exciting technological advancement and a profitable investment. According to the Financial Times, several major banks are increasingly looking to purchase credit insurance to protect themselves against potential losses on loans financing AI infrastructure.

Please think about that for a moment. The banks providing billions of dollars to finance this incredible expansion are looking for ways to transfer some of that risk to someone else. Shouldn’t that get investors’ attention?

The Financial Times reports that Barclays has explored insurance protection involving more than $10 billion in AI infrastructure loans, while HSBC has also investigated similar arrangements. JPMorgan Chase supposedly considered the strategy but ultimately decided against it. These arrangements generally allow banks to transfer 20% to 50% of their exposure to insurance companies, freeing up capacity to make additional loans. However, that protection isn’t cheap! According to the report, insurance premiums can consume 50% to 70% of the lending income associated with the insured portion of these loans.

Furthermore, some insurers are reportedly approaching their own concentration limits for data-center exposure. Are we simply transferring risk from one part of the financial system to another? Credit insurance can be a perfectly legitimate risk-management tool, but it doesn’t eliminate the underlying risk. It merely changes who bears the potential losses.

Compounding the financing challenge is yesterday’s significant increase in U.S. Treasury yields. Following another disappointing inflation report, the 10-year Treasury yield climbed approximately 12 basis points to get back to nearly 5%, while the two-year yield rose roughly 18 basis points. It’s largest one day move in nearly 18-months. This development is important because the enormous AI infrastructure buildout depends upon access to capital at a cost that allows projects to generate acceptable returns. The higher Treasury yields increase the baseline cost of borrowing, while widening credit spreads can make financing even more expensive.

We’ve already witnessed substantial credit deterioration among some highly leveraged AI infrastructure providers. Now banks are seeking additional protection against their exposure. What happens when the cost of financing rises at the same time that lenders become increasingly selective about which projects they are willing to support? The answer could be delayed or eliminated construction, reduced investment, lower profitability, and ultimately disappointing returns for investors. More to come!

Bitcoin and the Nasdaq-100

By: Russ Kamp, CEO, Ryan ALM, Inc.

Nearly five years ago, I produced a post titled “Not a Correlation of 1, But Certainly Strongly Positive”, in which I suggested that a pension plan CIO’s observation that Bitcoin enjoyed “a low correlation to every other asset class” was perhaps misguided. I highlighted the correlation of Bitcoin to domestic equities and more specifically the NASDAQ-100. Well, that correlation has only grown during the last several years, and yesterday’s price action certainly highlighted this fact.

For those believing that Bitcoin was a diversifier that would act to support a pension plan’s asset base during periods of equity market stress, we have yet to see that materialize.

Importantly, the correlations highlighted above do not include yesterday’s price activity when both the NASDAQ-100 (-6.4%) and Bitcoin (-5.8%) received a gut punch. It certainly appears to me that Bitcoin may offer limited downside diversification during periods of significant technology-sector stress. Furthermore, Bitcoin has not proven to be an inflation hedge during this most recent period of higher inflation. In fact, Bitcoin has fallen nearly 40% in value since peaking in October 2025 (>$126k), while inflation continues to push higher.

If Bitcoin is not going to provide an inflation hedge, and it truly doesn’t act like a diversifying asset, what is it that you are getting? Once again, it appears to me that you are just getting more exposure to equities, and specifically technology, at a time when the technology sector is dealing with issues related to the massive investment in everything AI.

Ryan ALM believes, and has promoted the fact, that the only true pension inflation hedge is cash flow matching (CFM) because CFM matches and fully funds the pension plan’s unique actuarial liability projections, which include an inflation factor. No generic asset could ever replicate or correlate accurately with a pension’s unique actuarial projections.

Caveat emptor!

Pension Plan to be Stable in X Years – Really?

By; Russ Kamp, CEO, Ryan ALM, Inc.

I first posted about this phenomenon 5-years ago. It is still so relevant today. I can’t tell you how many times I’ve read articles about underfunded pension systems that are “on track” to be fully funded in x # of years. The most recent one claims that the fund will be stable by 2048 (should I hold my breath?). However, what these articles fail to highlight are the actions required to achieve a fully funded status. First, the plan sponsor must make all annual actuarially determined contributions, which usually increase every year. Second, the assets must earn the actuarially determined ROA consistently for the next 22 years. But those are HUGE ifs, especially when one looks at the tremendous uncertainty that investors must wade through today – oil, inflation, interest rates, geopolitical crisis, the AI trade, etc.

What is the probability that the ROA will be achieved on a consistent basis? Furthermore, the sequencing of returns can play a meaningful role in whether the return objective is met, especially since the DB plan in question is a closed plan that will have greater outflows tomorrow than 20 years from now. What this plan needs to do is to secure the promised benefits for the next 10 years or so through a cash flow matching (CFM) mandate. Doing so will BUY TIME that will insulate the plan from potentially adverse market conditions that would harm the plan’s alpha assets. Buying time (extending the investing horizon) for these alpha assets permits them to grow unencumbered, as they are no longer a source of liquidity.

Furthermore, buying time enhances the probability that these growth assets will achieve their forecasted ROA since they are not diluted to make benefit payments. A Guinness Global research study on history of the S&P 500 tells us that dividends and dividends reinvested account for 47% of the total return on a rolling 10-year horizon since 1940 and the contribution to the total return is even greater for 20-year periods.

Actuaries do a terrific job despite the uncertainty in forecasting contributions, benefit payments, and expenses far into the future. But they aren’t magicians. There should be an expectation built into their forecasts that those inputs will not be achieved as predicted, especially market-based forecasts. We’ve seen this scenario play out time and time again. Given the volatility associated with predictions of this kind, perhaps a greater emphasis should be placed on the contributions, which reduces the reliance on markets, thus improving the likelihood of success. Just a thought!

What Would You do…

By: Russ Kamp, CEO, Ryan ALM, Inc.

Election day is finally upon us. I really couldn’t take one more political ad, especially given how nasty most of them have become. I’m not sure that I watched many in which the candidate talked about what they’d previously accomplished and what they hoped to do once elected. Most of the commercials were just about bashing their opponent. Something has got to change.

If I were running for office to secure Pension America – I actually ran in six prior elections (won 5) for two different boards of education and town council – this would be my platform. I would propose that every defined benefit pension plan have as its core holding a cash flow matching (CFM) portfolio in lieu of active core fixed income. The balance of the plan’s assets could then be managed as they are today, or perhaps more aggressively since CFM will expand the investing horizon increasing the probability of those assets achieving their desired outcome. My promise to you (no fingers crossed behind my back): Ryan ALM, Inc. would,

  • Secure the promised benefits at low cost and with prudent risk,
  • Dramatically improve the monthly liquidity to meet those promised benefits (and expenses) chronologically,
  • Extend the investing horizon for the alpha assets,
  • Eliminate interest rate risk for the CFM assets, as they are defeasing benefits that are future values,
  • Keep the plan from forcing liquidity when selling assets to meet benefits is not advantageous,
  • Significantly reduce the cost of those future promises effective on the day the portfolio is constructed, and
  • Provide all of this at modest cost

Like many politicians promising the moon, these “benefits” may sound too good to be true, but this is what CFM does all of the time! To show you that we mean what we say, we are always willing to provide plan sponsors with a free analysis to highlight what CFM could do for your plan.

What do you say? Do we have your vote? I believe that your plan participants would say, “YES!”

AI Financing Watch: It’s All About Free Cash Flow!

By: Russ Kamp, CEO, Ryan ALM, Inc.

I’ve begun producing a roughly weekly update on AI financing given the incredible fundraising that is occurring through the debt markets. Here is some perspective on the latest comings and goings.

The important AI funding development isn’t simply that AI spending remains enormous, as it clearly does. However, investors are now rewarding companies that can fund that spending from cash generation and those that don’t get severely penalized.

Microsoft provided a positive story, reporting that Azure growth was 48%, quarterly revenue was $92.3 billion and, most importantly, $18.6 billion of free cash flow or roughly $10 billion above expectations. However, Microsoft still spent $45.2 billion on capex during the quarter and expects Azure growth to continue to accelerate. As a result, Microsoft’s shares rose more than 2% after the announcement.

Then there was Alphabet, which produced similar results, but the market’s reaction was quite different. Alphabet generated spectacular Google Cloud growth with revenue up more than double to $30.7 billion and backlog order revenue at $561 billion. Yet quarterly free cash flow was negative $9.1 billion, as AI and cloud investment continued to surge. Alphabet also expects capex to finish 2026 at between $195–$205 billion. They are also projecting that spending will increase substantially again in 2027. Unlike Microsoft, the market’s reaction was far less supportive, as Alphabet’s shares declined nearly -13% yesterday. The loss removed nearly $500 billion of market cap.

Regarding the debt markets, Microsoft raised roughly $25 billion in its first U.S. bond offering since 2017. Demand exceeded $100 billion, and the offering spans maturities from two to 40 years. Given that Microsoft remains free-cash-flow positive, it shouldn’t be surprising that Microsoft’s 40-year tranche was reportedly priced at only about 105 basis points over Treasuries, versus roughly 130–140 bps for comparable long bonds from Alphabet, Meta and Oracle. Clearly the bond market is still extremely willing to finance AAA-quality AI investment at relatively modest spreads.

Beyond Microsoft, the scale of AI-related borrowing remains incredible. FTSE Russell reports that Amazon, Alphabet, Meta, Microsoft and Oracle had already issued $218.1 billion IG debt by early October, which is more than double the $107.3 billion issued during all of 2025.

Speaking of negative free cash-flow, Amazon reported that trailing-12-month free cash flow had swung to -$42.4 billion, following Q3 capex of $77.3 billion, which exceeded expectations by roughly $20 billion. Amazon still expects that spending will be meaningfully more in 2027. Amazon shares are currently down about 3.3%.

Lastly, Intel, a second-order AI play, reported yesterday’s Q3 results in which they announced that Intel’s Data Center and AI revenue jumped 78% year-over-year to $7.3 billion. However, it remains a secondary source of company revenue. The balance of the business missed expectations and the stock is currently down -28.7% today!

39-Years Ago!

By: Russ Kamp, CEO, Ryan ALM, Inc.

Black Monday’s 39th anniversary came and went with little fanfare. I can recall quite vividly the events surrounding the U.S. equity market’s largest single day decline and wondering at the time if I was going to have a job by the end of that week. For those of you who were working on Wall Street (or even born) at that time, you’ll remember that the Dow declined 108.35 points (-4.6%) on Friday, October 16, 1987, making for a very long weekend, as we wondered what might happen next. I certainly didn’t expect a 508-point decline (-22.6%), and I suspect that most didn’t either. Importantly, the crash didn’t emerge from a weak market. The Dow had risen an extraordinary 44% in the first seven months of 1987, before peaking in August.

As I reflect on that time, I wonder if there are similarities to today’s investing environment, including valuations, interest rates, market concentration, and a powerful theme driving market enthusiasm that should give us some pause.

1. Valuations: Current valuations are stretched, and perhaps more so than in 1987. The Shiller CAPE is currently about 40.7, versus a historical mean around 18. The only comparable recent example was the 1999–2000 technology bubble (44). Another period of note was in September 1929, when the CAPE was about 32.6. If you aren’t a fan of the Shiller CAPE, another current valuation measure puts the S&P 500 at roughly 21× forward earnings. Like any valuation measure, they aren’t timing tools, but they do tell us investors are paying a very high price for future earnings that might limit future gains.

2. Interest Rates: U.S Treasury yields have risen significantly since bottoming in 2020, achieving a level that makes bonds serious competition for stocks. In 1987, the 10-year Treasury note began the year with a yield of 7.2%. By September 30, 1987, that yield hit 9.63%, providing an extremely attractive alternative to equities. Today the absolute yield is much lower at 5.11% (9:13 am) but that’s up considerably from the near-zero-rate environment in which many of today’s equity valuations were established.

3. Market Leadership: Today’s S&P 500 has become increasingly dependent upon a handful of mega-cap technology/AI companies. S&P Global specifically warns that the Magnificent Seven increasingly behave like a single AI-related risk factor, rather than seven independent sources of diversification. In a note from Glen Eagle, they point our that the five largest companies in the S&P 500 are currently so massive that they are worth the same as the bottom 434 stocks in the index combined.

4. Powerful Narrative: Today’s investing narrative surrounding AI may be more obvious than that which was driving markets in the mid-’80s, but not necessarily more powerful. In 1987, markets had experienced years of deregulation, financial innovation, globalization, and economic expansion following a prolonged period of rising rates, that began to reverse in 1981. As I’ve written in previous blog posts, I don’t question AI’s transformative potential. I question whether today’s prices and enormous capital expenditures can generate sufficient economic returns to justify those investments.

I’m certainly not suggesting that equity markets may see another -22% drop in a single day, as I witnessed 39-years ago. Importantly, today’s markets have circuit breakers. A 7% S&P 500 decline triggers a Level 1 halt, while a -13% triggers Level 2, and -20% triggers Level 3, ending trading for that day. These circuit breakers don’t prevent a major decline, but it makes a repeat of October 19th’s –22.6% single day decline much harder to replicate.

Final thought: In 1987, the market crash itself created the crisis. Today, I wonder whether the capital cycle could create the equity correction, as everything AI is consuming enormous amounts of capital and driving interest rates up in the process. Hyperscalers are spending hundreds of billions of dollars, with potentially trillions of $s yet to be invested. Data-center developers are borrowing heavily well before they are likely to see a return on that investment. Power infrastructure requires still more capital, as capacity seems inadequate to meet future demand. As I wrote about last week, CoreWeave debt has repriced dramatically, where their bonds issued at a yield of 9+% during the summer are now trading at roughly 16%. If AI infrastructure returns ultimately fail to exceed these increasingly expensive financing costs, investors may begin questioning the earnings assumptions embedded in AI-related equity valuations. If that happens, all bets are off!

Are Cracks Forming?

By: Russ Kamp, CEO, Ryan ALM, Inc.

On September 30, 2026, I wrote my first post regarding my concerns about capacity within the AI trade, as hyperscalers and other businesses associated with AI and data center buildout were rapidly seeking financing through the credit markets. Since that post, I’ve produced one additional update as well as track daily developments in the credit markets that might impact not only Ryan ALM’s business, but bonds/spreads in general.

I’m taking the time to write today because there has been a meaningful development, and this may be the clearest credit-market warning I’ve seen since I began tracking the AI buildout.

The critical development is CoreWeave’s bond selloff. CoreWeave is a publicly traded cloud-computing company built specifically around GPU-intensive AI workloads rather than the general-purpose cloud model of AWS, Azure or Google Cloud. It provides GPU computing, storage, networking and software used to train and run large AI models.

Here’s the story. CoreWeave issued bonds in June at a yield of roughly 9.7%, which are now yielding around 15%! Reuters describes the move as a test of the market’s willingness to continue financing highly leveraged AI infrastructure and says investors are increasingly questioning whether future cash flows justify the debt required to build it.

And CoreWeave isn’t an isolated indicator. Data-center bonds associated with projects backed by major AI participants, including Meta Platforms, Oracle and Microsoft have fallen below par as investors reassess leverage, project the potential economics of these buildouts, and the enormous amount of financing still required.

Demand for the physical infrastructure remains extremely strong. Schneider Electric reported triple-digit data-center demand growth today and raised its 2026 forecasts again; Q3 organic revenue increased 16.9%. So this is not evidence that demand for AI capacity has suddenly disappeared. It is, however, evidence that investors are becoming less willing to finance that capacity at almost any price, and rightfully so.

For AI equities, I would consider this a genuine warning signal. Companies dependent upon repeated external financing should now be valued very differently from cash-generative hyperscalers. CoreWeave’s shares are down about 15% this week even as the underlying demand story remains strong.

For bonds, we’re seeing the beginnings of exactly the differentiation that we anticipated. A 15% yield represents substantial credit, refinancing, and execution risk. If this repricing spreads from “neoclouds” into SPVs and data-center project debt, the marginal cost of AI capital could rise substantially.

For data-center real estate, this adds a second scarcity test to the one we’ve been discussing. A project increasingly needs both secured power and secured economical financing. Having electricity but needing 12%–15% debt doesn’t necessarily produce an economically viable data center.

Ryan ALM, Inc.’s Concern: Stagflation

By: Russ Kamp, CEO, Ryan ALM, Inc.

The U.S. capital markets are sending some interesting, and perhaps conflicting, signals. Treasury yields have risen dramatically, energy prices remain elevated, diesel supplies continue to be constrained, and corporate credit spreads are beginning to widen. Yet inflation expectations remain relatively well anchored and economic activity has held up surprisingly well. What are we to make of all this? I believe we may be witnessing the early stages of an environment tilting toward stagflation.

The problem, as I’ve written before, is that much of today’s inflation isn’t being created by excessive demand. It is being driven by supply issues, particularly within energy. The Fed can raise the Fed Funds rate by 25 basis points, 50 basis points, or perhaps much more, but none of those actions will produce another barrel of oil, gallon of diesel, or unit of refining capacity. Clearly, we aren’t going to plow fewer acres, ship less food, or stop moving goods around the country simply because interest rates have increased. Energy sources are embedded in virtually everything we produce and consume. Higher interest rates don’t fix that problem.

What higher rates can do is reduce demand elsewhere. Unfortunately, that means households get hit from both directions. They pay more for gasoline, food, utilities, and transportation because of the current energy supply issues, while simultaneously paying more for mortgages, auto loans, credit cards, and other financing because of the Fed. Businesses face the same squeeze through higher operating costs combined with a higher cost of capital. Is that really the prescription we want when the original inflationary driver is primarily a supply problem?

The markets are beginning to reflect this tension. The 10-year Treasury yield has moved well above 5% (currently 5.27% at 12:07 pm), producing an extraordinarily high real yield given that longer-term inflation expectations remain near the mid-2% range. At the same time, credit spreads have begun widening, while oil remains expensive and diesel markets remain stressed. Credit spreads aren’t signaling a crisis today, but the combination of these many factors bears careful watching. If credit conditions continue deteriorating while energy costs remain elevated, we’ll have an increasingly powerful stagflation cocktail based on stubborn inflation accompanied by weakening economic activity.

The Fed should obviously be concerned about inflation. However, I still question whether monetary policy can effectively address the inflationary environment that we’re experiencing without creating other problems in the process. The Fed can diminish demand. It cannot create supply. If energy prices remain elevated while real interest rates stay near more restrictive levels and credit conditions continue to deteriorate, we shouldn’t be surprised to see economic growth suffer. We’ll continue watching Treasury yields, inflation expectations, oil and diesel prices, and credit spreads closely as we monitor conditions for stagflation. Current economic conditions are beginning to get very interesting.

Kudos to Florida and Its Liquidity Statute

By: Russ Kamp, CEO, Ryan ALM, Inc.

You are not likely to know about Florida Statute §112.661(6) unless you reside in Florida and sit on a pension board, and even then, you still might not have heard or read about this statute. It became effective October 1, 2000, as part of Chapter 2000-264, Laws of Florida, enacted through CS/SB 372, “Investment of Public Funds”, which Governor Jeb Bush approved on June 14, 2000,

This provision is the only one of its type among the 50 U.S. states. Florida should be commended for its inclusion in the management of public pension plans. The subsection for this statute is titled “Maturity and Liquidity Requirements.” 

It provides: “The investment policy shall require that the investment portfolio be structured in such manner as to provide sufficient liquidity to pay obligations as they come due.”

It then goes a meaningful step further: “To that end, the investment policy should direct that, to the extent possible, an attempt will be made to MATCH (the statute’s word, not mine) investment maturities (that says bonds to me) with known cash needs and anticipated cash-flow requirements.” As Phil Rizzuto would say, “Holy Cow”! If given the opportunity, I couldn’t have written this requirement any more clearly.

Again, the Investment Policy Statement (IPS) should direct that an attempt will be made to match investment maturities through bonds with known cash needs and anticipated cash-flow requirements. That sounds a lot like Cash Flow Matching (CFM) to me.

So, I ask, within your (IPS) do you have an explicit Liquidity Policy? If so, does it consider Florida Statute 112.661? At the most recent FPPTA conference in Orlando, I was asked to speak to liquidity management practices currently used in Florida. What I discovered is the there are two fundamentally different philosophies:

Asset-driven liquidity

“When we need cash, where can we get it?”

Liability-driven liquidity

“We know when benefits must be paid, so why don’t we arrange the assets today to produce the cash when those payments occur?”

Here are the seven strategies that I was able to identify that are currently being utilized. I’d be interested to hear from those of you that might have adopted some other means to secure liquidity.

Only #7 falls under the category of liability-driven liquidity, which comes closest to adhering to statute 112:661. Given the rapid rise in U.S. interest rates, creating a cash flow matching portfolio for some period of time (10-years) not only SECURES the necessary liquidity, it extends the investing horizon for the plan’s residual assets, while dramatically reducing the cost to fund those future benefits. We are always happy to provide a free analysis on what CFM could accomplish for you and your fund. Don’t let this wonderful rate environment pass you by.

Is AI Investment Reaching A Natural Limit?

By: Russ Kamp, CEO, Ryan ALM, Inc.

I’ve been spending a lot of time recently thinking about the incredible amount of capital being thrown at Artificial Intelligence (AI). There is no question that AI has the potential to dramatically change how we live and work. But does that mean that every dollar being invested in AI infrastructure is going to generate an acceptable return? I’m not so sure. As interest rates continue to rise, I think investors need to start asking a different question. Instead of asking, “How big can AI become?”, perhaps we should be asking, “How much AI capacity can actually be built economically?”

AI investment is staggering. Microsoft, Amazon, Alphabet/Google, Meta, and Oracle (the Hyperscalers) are at the center of this massive infrastructure buildout. Until recently, these companies generated so much cash that they could fund most of their capital spending internally. That situation is changing rapidly. AI-related capital expenditures are consuming an enormous percentage of their operating cash flow, and the hyperscalers are increasingly turning to the bond market and other forms of financing to keep the spending machine going. Four of the five major hyperscalers have issued significant amounts of bonds in 2026, with Microsoft being the notable exception. Collectively, hyperscaler borrowing has already reached roughly $200–$220 billion this year. Wow!

Why should we care? Because the cost of money matters and this massive investment could profoundly impact equities, bonds, real estate, private equity, private credit, etc. DB Pension plans need to take notice! When Treasury yields were 1%-2%, financing massive data centers and other AI infrastructure was relatively inexpensive. Today, Treasury yields have climbed above 5%, and the 30-year Treasury bond is closing in on 5.7%, while some long-term hyperscaler debt is being issued at 6% or more. Oracle has recently  issued long-dated debt carrying coupons approaching 8%. This reality changes the economics dramatically. It isn’t enough for a $10 billion or $20 billion AI project to generate revenue. It needs to generate a return sufficient to compensate investors for the cost of the capital, operating expenses, electricity, depreciation, technological obsolescence, and the risk associated with the project. What is that return likely to be and where is that return going to come from?

Capital isn’t the only potential constraint. AI requires enormous amounts of electricity, generation capacity, transmission, transformers, land, cooling, water, semiconductors, construction, and, of course, data centers. A large AI campus can require hundreds of megawatts of electricity. High-density data centers can cost roughly $14-$16 million per megawatt, meaning that a 500 MW facility could cost approximately $7.5 billion to construct before considering the broader power infrastructure necessary to support it. We keep hearing about seemingly unlimited demand for AI. Fine! But there certainly isn’t unlimited electricity, grid capacity, construction capability, or CAPITAL. Why does the investment community seem to assume otherwise?

Furthermore, the AI trade may be creating its own headwind. Think about this for a minute. Massive AI capital expenditures consumed free cash flow from most of the hyperscalers. Declining or exhausted free cash flow creates a need for external financing. Greater borrowing produces more corporate bond supply. More bond supply can contribute to higher yields and wider credit spreads. Higher financing costs increase the hurdle rate on the next AI project. Eventually, some projects simply won’t make economic sense. That seems like the beginning of a vicious cycle to me.

As mentioned previously, the implications extend well beyond technology stocks. Equity investors need to determine whether these enormous capital expenditures are actually producing an acceptable return on invested capital. Bond investors are being asked to absorb hundreds of billions of dollars of new AI-related debt and need to be compensated appropriately. Real estate investors financing data centers must compete against a >5% risk-free Treasury yield while dealing with higher construction and financing costs. 

I’m certainly not suggesting that the AI boom is about to end. But I do believe that the AI investment thesis may be entering a very different phase. The first phase was about AI models, semiconductors, hyperscalers, and data centers. The next phase may increasingly be about the scarce resources necessary to support all of that growth, beginning with capital and including electricity, generation, transmission, transformers, cooling, powered land and water. The winners may ultimately be those controlling the scarce resources rather than simply those spending the most money.

Markets have an interesting habit of believing that trends can continue indefinitely. From my 45-years in the investment industry, I’ve come to appreciate that they don’t. There is always a natural capacity to every investment. AI will prove to be no different. At today’s cost of capital, the important question isn’t how much AI infrastructure companies want to build. It is how much they can afford to build while still generating an acceptable return. Are today’s investors and pension plans adequately considering that distinction? I’m not convinced that they are.

I have an idea. While you wait for the AI thesis to play out, buy time (extend the investing horizon) by creating a cash flow matching (CFM) portfolio that will secure the monthly benefits and expenses for some time – say 10-years. This will enable that AI thesis to perhaps generate the desired return while it grows unencumbered.