The second quarter reflected a continuation of what we saw in the first quarter in most respects. The one notable exception is oil. As the U.S. neared a compromise with Iran, oil prices have returned to within a few dollars per barrel of where they were before the war broke out. Meanwhile, the “AI loser” stocks (think software, IT services, data providers) kept losing and the “AI winners” (semiconductors, semi cap equipment and other electronic equipment) kept winning. That relatively small group of sub industries in the MSCI World ex USA Index contributed roughly 35% of the index return despite an average weight of only 7%. Talk about punching above your weight!
Our holdings in ASML (ASML) and Samsung (SSNLF) were significant contributors in the quarter. You may have noticed we continue to hold Samsung and ASML in similar weights to last quarter despite the significant run-up in their shares in a relatively short period of time and wondered why. We are value investors. Shouldn’t we be reducing our holdings substantially given the price moves? While that is typically the case, it’s not here. We (and consensus) underestimated how massive the positive impact from AI would be on their fundamentals. In short, we were wrong, but in the right direction. This occurs when we are surprised positively by the fundamentals. Clearly, this is more fun than mistakes where fundamentals are not tracking your thesis (wrong in the wrong direction), but they can be just as painful and act as future errors of omission if you are too eager to declare victory. In this case, we underestimated what this upcycle would look like for the semiconductor ecosystem. This is where our mistake management system really paid off. It worked to slow us down as evidence emerged that our forecasts were wrong. In fact, Samsung has already been reviewed twice in front of the group this year, each time triggered by a more than 10% increase in our estimate of value.
At the risk of stating the obvious, let me take a moment to provide context for what’s happening. The demand for AI has been staggering with the leading AI labs, excluding Google (GOOG), generating a revenue run rate over $80B – four times the revenues a year ago. This led to enormous demand for compute and, in the case of Samsung and ASML, leading edge memory known as “HBM” or high bandwidth memory. For now, at least, AI compute requires a significant amount of HBM which consumes three times the capacity of lagging edge, more commoditized memory. This has a downstream effect of leading to shortages of supply there as well. The result has been industry revenue and profits now running at multiples of prior peaks already and yet, continuing to grow at impressive rates—for now.
It is worth being specific about why we own Samsung and ASML within a much broader universe of 2026 AI winners. Many of the sub-industries that rallied hardest this year were commodity memory, chip foundries and equipment suppliers that have historically been textbook boom-bust businesses: price takers with thin margins and little control over their own economics between cycles. We did not buy Samsung or ASML because we foresaw an AI supercycle; we plainly did not. We own Samsung for its scale and process leadership in memory, built over decades and now most visible in HBM. We own ASML because it is, in practice, the only supplier of the extreme ultraviolet lithography tools the entire industry needs to make leading-edge chips, a position closer to a monopoly than a commodity. Those competitive advantages were the reason we underwrote both businesses long before HBM entered the vocabulary of every equity analyst.
So how are we handling this in our values? The most significant change is giving full credit for the near/medium-term cash flows—which have more visibility—from what has emerged as a super-cycle for the ages. This matters because the earnings levels are large as a percentage of the current market capitalization. In our view, the low current trading multiples signal the market’s skepticism about the sustainability of these cash flows. While we believe a cash windfall of this magnitude is likely temporary, that’s a separate question from whether the competitive landscape has structurally changed. Memory producers have historically been price takers in the spot market and today they can essentially name their price (within a range) in take-or-pay contracts locked-in for several years in advance. It remains to be seen whether this new competitive paradigm will last, but one cannot ignore that the dynamic today has changed. This adds a positive skew to the range of outcomes, so we are giving partial credit for somewhat higher returns going forward but the cash flow windfall remains the primary driver of the cumulative change in our values year-to-date. We are going to continue monitoring the situation closely to make sure our values continue to be our best guess given how quickly things are evolving with AI.
We never seek to be wrong in the right direction, but when evidence emerges that we are, we have a process to deal with it, and it was delightful to witness it work this quarter. We cannot promise we will continue to own Samsung and/or ASML next quarter or next year, but we can commit to sticking to our process, whatever the future brings.
A Farewell and Thank You to Intertek.
I just wanted to take a minute to acknowledge the terrific job done by the board of Intertek as an offer emerged from private equity during the quarter. Management and the board had all the typical reasons to say no, such as self-help margin opportunities and portfolio pruning. Importantly, when the price reflected the opportunity and risks of those efforts, they accepted the offer. Too often boards view any self-help or portfolio pruning as reasons not to sell the company because they ought to be positive and therefore it seems no price can be substantial enough until they complete this work. What they often fail to recognize is that these opportunities existed under their watch and do come with execution risks, particularly if that management team has been in place for some time. The Intertek board did not fall into that trap, and they ought to be commended and widely copied!
Thank you for your continued partnership in the Harris | Oakmark International strategies.
Tony Coniaris, CFA Portfolio Manager
Original Post
Editor’s Note: The summary bullets for this article were chosen by Seeking Alpha editors.
