Letter to Investors | PDF

“The more things change, the more they stay the same.” That old maxim is proving true again as the financial markets are experiencing the curious phenomenon of dual gold rushes. The first is a rush to own the yellow metal itself. The second is a race to make money in artificial intelligence (AI). These two movements dominating the news are not related, in my view, but each reflects important trends that investors need to be aware of.

As to the barbarous relic[1] (gold), deglobalization and US government policies emanating from both parties over the last 10 years are pressuring the dollar down relative to other currencies. In this regard, gold is a currency, one that cannot be manufactured at will by any central bank. And speaking of governments, foreign reserve managers worldwide are diversifying their foreign reserves by rebuilding their gold reserves after spending most of the last 55 years reducing their holdings. They’ve been busy buying gold for at least the last four years, and we expect this to continue. We’ve been arguing for over a decade that deglobalization is dollar-bearish and will worsen inflation. These realizations are finally coming to pass, influencing the gold market. For several years we have allocated to gold in Confluence strategies that involve commodity exposure (Asset Allocation and Global Hard Assets), and we expect to continue holding these positions.

The other gold rush is even more complicated. In modern parlance, the phrase gold rush denotes a race by investors to get rich by investing in new technologies that seem to have “can’t miss” potential. This, of course, is an analogy to the great gold rushes of the 19^th century. The two most famous were the 1849 California gold rush and the 1896 Yukon gold rush, but there have been many others in US history, starting with the North Carolina gold rush of 1799. In fact, a new discovery somewhere precipitated another gold rush about every 20 to 30 years. It’s a historical oddity that stock market “gold rushes” occur at about the same frequency. Starting with the railroad boom of the post-Civil War era, we have had similar “gold rushes” for automobiles, radio, aeronautics, semiconductors, personal computers, the internet, and cell phones. Now, it’s AI.

Each one of these technological advancements has eventually proven to be every bit as economically and socially important as early investors thought they’d be. Unfortunately for those gold rush speculators, such booms (bubbles?) suffered from two common faults. First, in their haste to participate in the new technology, investors threw money at the stocks of many companies that eventually failed. It’s a fact that winners in new technological races are greatly outnumbered by the losers. Second, investors have tended to badly overvalue the future worth of these businesses. For example, we saw this on display in the internet bubble of 1999-2000. Not only did many of those dot-coms fail, but the stock prices of the surviving internet and telecom stocks set highs that were not equaled for many years. The NASDAQ Composite Index set a high in 2000 that was not equaled until 2015.

We’re seeing the same thing in the AI race today. Our own Thomas Wash reported in a recent Asset Allocation Bi-Weekly[2] that a study by MIT showed 95% of the companies investing in AI were earning a zero return. OpenAI, the private company that produces ChatGPT, was recently valued by investors at $500 billion, which is over 38 times the revenue they expect this year. Of course, the company is unprofitable in the extreme and does not even expect to be cash flow positive for another four years.

Is anyone making money on AI? Yes, the companies selling the products needed to build out the AI server farms: companies that make semiconductors, servers, cooling systems, data centers, electric generating equipment, etc. In our gold rush analogy, these are the makers of the picks and shovels needed to prospect for gold. Those businesses made a lot of money, at least until the rush ended, even though most of the miners didn’t. The same is true today, and the “picks and shovels” stock prices are also extremely high. Nvidia, the maker of the chips everyone wants for these servers, has a market value of $4.4 trillion (greater than that of the entire German stock market), which is merely 22 times its expected 2025 sales.

We expect that AI will prove to be every bit as economically important as its fans expect, but we caution investors not to over-invest in the sector. Such overly enthusiastic early excursions into economy-changing technologies of the past often ended badly. While we own a few AI-related stocks in our equity portfolios (and within the ETFs in our Asset Allocation portfolios), great businesses that we’ve owned for many years, we have stayed well diversified among many industries. We’re not speculators, but long-term investors. In a time of technological excitement, that’s a distinction that investors need to remember.

These Technology and Communications sectors are now worth 45% to 50% of the major US stock indexes, an amazing figure that, in my opinion, overvalues their economic value over the next years. This rapid run-up in prices has pressured downward the valuations of virtually every other sector of the stock market, leaving the stocks of many excellent businesses anywhere from reasonably valued to downright cheap. These include many outstanding dividend-payers. We cannot predict how long the current AI boom will last (nor how long the more modest valuations of other stocks will remain), but we encourage you to keep focused on your long-term objectives and concentrate, as we do, on quality businesses that will survive the boom-bust cycle.

We appreciate your confidence in us.

Gratefully,

Mark A. Keller, CFA
CEO and Chief Investment Officer

[1] A little over 100 years ago, John Maynard Keynes, the well-known British economist, famously dubbed gold “the barbarous relic.” The moniker stuck.

[2] “The AI Arms Race: Navigating the Divide Between Promise and Profit,” October 6, 2025.

View PDF