If you are a consistent reader of this weekly missive, you know that I occasionally open with an analogy to set the course of my article. Well, you will have to bear with me on this one…
In the mid-1800s in the newly forming America, word started to spread that gold has been found out west. In response, many (from all over the world) saw this news as the chance to change their family’s fortunes forever. So, many of those who headed to the Western U.S. (and Canada), sold their farm or sold their business – a then steady and proven source of income – to buy a claim at a price that only makes sense if it’s sitting on a rich vein of gold. The logic for the action was not based on valuation analytics. Rather, selling the farm/family business and buying a claim was about “gold is real, fortunes are being made, and if I don’t get in now, I’ll miss the greatest opportunity of my lifetime.” The gold rush itself was real, and a few people did strike it rich. But most claims bought at the peak produced little. The people who reliably made money were the ones selling shovels, supplies, and land to the hopeful, often at inflated prices. My point is simply: being right that a transformative opportunity exists is not the same as being right that your stake, at your price, will pay off.
When an opportunity feels historic (and may even BE historic), investors stop anchoring to valuation and start anchoring to fear of missing out. The upside may be real, but the price can still be too high to earn a good return, even if the big story turns out to be true.
Which brings me to artificial intelligence.
Two Questions That Are Not the Same Question
Let me say this clearly up front, because it is the part people tend to make immediate assumptions about: in my view, artificial intelligence may well be a genuinely transformative technology, and I am certainly not writing to argue otherwise. The productivity possibilities are real, the capital being deployed is real, and the changes already showing up in how businesses operate are real.
But in the world of investing, we have other factors to consider like, is this technology truly “transformative?” and is this possibly transformative technology a sound investment at today’s price? These are two entirely different questions even if they feel similar. And conflating them is one of the more expensive mistakes an investor can make. Warren Buffett may have summed this up most eloquently when he said “price is what you pay, value is what you get”. Price matters when you invest. So, even if artificial intelligence and the related platforms (Claude, ChatGPT, Gemini, etc.) are truly transformative, that does not mean the benefits of the transformation should increase the value of all companies in the S&P 500, nor does it mean there won’t be companies (even industries) and, therefore, investors, that will be negatively impacted.
History is fairly generous with examples here. Railroads reshaped commerce in the 9th century and yet, still bankrupt a remarkable number of railroad investors along the way. Commercial aviation transformed how the world moves and has been, over the decades, a famously difficult place to earn a durable return (1) The internet did approximately everything its most breathless advocates promised — just not on the timeline, and not for the shareholders, who paid 1999 prices for it. In each case, the technology was not the problem. The price was. The transformation arrived but it simply arrived after a lot of investors had already paid for it in advance.
What the Price Is Already Telling Us
So how much of the AI future has already been pulled into today’s prices? Nobody knows precisely – I know I do not… but I would submit anyone who tells they do is not being honest with you (at best) or is entirely delusional (at worst). However, we can look at a few metrics and concepts that are measurable rather than speculative.
Concentration is one. At the end of 2025, a universe of 28 direct AI stocks now represents roughly 50% of S&P 500 market capitalization, and a broader group of 42 AI-related companies has generated something in the range of 65% to 75% of S&P 500 earnings, profits, and capital spending since late 2022. (2) Whatever else that is, it is a market that has made a very large, very specific bet. No judgement, just information.
Spending is another. Consensus expectations have the largest cloud and infrastructure companies spending roughly $770 billion on capital expenditures in 2026 — an amount approximately equal to 100% of their cash flow from operations — with net debt across that group up around $170 billion since the start of 2025. (3) I want to be careful here, because that second figure is not automatically alarming. Companies borrow to build all the time, and building is generally how value gets created. But it does change the character of the investment. When a company funds an investment out of spare cash, the future has room to disappoint. When it funds that investment out of every dollar it earns and then some, the future has considerably less room to disappoint.
The Scoreboard Replaces the Story
What I find genuinely interesting about this particular moment is that for roughly two years mentioning AI on an earnings call was, by itself, treated as good news. Fortunately, that has changed. Investors have moved from rewarding the story to asking for the scoreboard — where is the revenue, what is the margin, when does the spending produce a return? That is a healthy development, in my view. Skepticism is not pessimism. Asking a company to show its work is not the same as betting against it.
It does, however, mean something practical for portfolios: the range of outcomes within the AI theme may be widening rather than narrowing. Being “right about AI” as a technology may not be sufficient. The harder work is figuring out which businesses actually capture the economics of it — and what you are paying for that possibility. After all, great technology may improve all sorts of experiences (and, even productivity) but that does not mean it is a rising tide that will raise all ships.
As an investment advisor, I have experienced clients, friends (okay, as you might expect “friends of Matt” is a pretty short list… but nevertheless) and acquaintances ask: what are we supposed to do about AI? Well, not panic, for starters. Fear of missing out and fear of participating in a future correction are just two sides of the same coin. Maybe “buy anything AI-related” or “sell everything and hide in the garage with the second freezer” are both too extreme. Like many/most big ideas there is far more nuance than the headlines suggest and far more variety of outcomes than offered by the talking heads.
With that in mind, a few things I think are worth considering, and I offer these as general principles rather than advice for any specific situation:
Know what you already own. Many investors who have never intentionally bought an “AI investment” have substantial exposure to this theme through ordinary broad-market index funds. That is not a flaw (or strength) in the funds, rather, it is simply how they are built and worth understanding, because it may mean your portfolio is less diversified than the number of holdings suggests.
Distinguish conviction from certainty. You can believe a technology will reshape the economy and still acknowledge you do not know what it is worth today. Those two viewpoints can certainly be compatible. In fact, I would argue holding both at once is the mark of a serious investor.
Let rebalancing or systematic allocation adjustment do the arguing. A disciplined rebalancing or allocation adjustment process makes decisions about position sizing without requiring you to make a prediction about the future. It is unglamorous… but that is largely the point.
Match your time horizon to your expectations. If your reason for owning something is a benefit that arrives in 2033, your patience needs to extend to 2033. If your money is needed in 2027, that is a different conversation entirely.
The Honest Ending
I do not know how much of the AI future is in today’s prices. Neither does anyone else, though you would not always know that from reading the financial press. What I do know is that the willingness to pay now for benefits that arrive later is not a defect in human nature so much as a feature of it. It is the same instinct behind planting trees, funding a 529, and putting money in a retirement account at 30 for a self you cannot quite picture. That instinct builds things. It also, on occasion, overpays for them.
The work of investing — the actual work, the part that is not fun to talk about at parties — is holding that instinct and a calculator at the same time. Believing in the future while staying honest about its price. At East Franklin Capital, that is a good part of what we do: helping clients participate thoughtfully in the economy’s long arc without confusing enthusiasm for analysis. If you would like to talk through how this theme shows up in your own portfolio — and for many people, it shows up more than they expect — let’s have that conversation.
