Macro Pressure Sitting on Top of Strong Micro Earnings
The recent market action comes down to a lot of macro pressure overwhelming genuinely strong fundamental earnings on the micro side. Companies are delivering, yet the price action doesn't reflect it, because forces outside any single company's control keep pushing valuations around day to day. The result is heavy short-term volatility with several drivers stacked on top of each other.
Oil is one of them. The situation in Iran has moved oil prices, and that carries a specific consequence for the jewels of the AI supply chain, South Korea and Taiwan. Both are largely energy dependent, so higher energy costs feed straight into their currencies and their inflation. That adds another layer of volatility to the names most central to building out AI.
Rather than treat this turbulence as a signal to flee, it functions as a reason for investors to pause, step back, and look at the full AI picture. The useful question in that pause is where to position within AI, and specifically which companies can convert their AI exposure into durable earnings power rather than a one-time boost.
Memory as the Proof Point
High bandwidth memory, and the DRAM segment in particular, is where durable value is solidifying most clearly. Micron's numbers last week made the case: a strong quarter, strong results, strong guidance. Pull back to understand why memory accrues so much value and the answer sits in the enormous CapEx flowing into US data centers. There is roughly a trillion dollars of that spending, and almost 40% of it goes into memory. That single figure explains why memory is the proof point to look for when hunting for the pockets of the AI trade that are both exposed to the theme and actually capturing lasting value.
Memory is the obvious candidate, the one everyone already watches closely. A less crowded angle is the movement of data itself. With data proliferating, it matters both to store it and to move it, which draws attention toward networking and interconnect plays. Two names stand out here. Marvell sits on the networking fabric side. Semtech holds a strong position in signal integrity, and moving data well means preserving the highest possible integrity of the signal while it travels, keeping latency low, keeping power draw low, and holding the fastest speeds. Storing data and shipping it around are both necessary; signal integrity is what makes the second one work.
Check Writers and Check Takers
The migration of capital away from the hyperscalers and toward the bottlenecks has been described as check writers versus check takers. The hyperscalers write the checks; memory, AI infrastructure, and networking suppliers take them. Something worth noting happened in South Korea, where SK Hynix and Samsung, long the check takers, started acting as check writers themselves. Capital rotated out of those names, which had performed very well, and into industrials and renewable energy, since the fabs planned in Korea are meant to run on renewable power.
A similar pattern may play out in the US. The check beneficiaries taking advantage of all that CapEx are themselves spending heavily, and industrials have been performing well month to date, so some of that rotation into industrials looks likely to come through here too. Part of what's visible in the US is rotation out of AI entirely, with investors taking profits into the more defensive positions available in the current market. The more common move, though, has been rotation within AI. Interconnect, optics, and cooling companies performed well through the first two quarters of the year, then began lagging the memory names. That shift is driven by investors stepping back to ask who actually owns the long-term durable value, and steering more capital toward the memory companies and select semis.
How to Invest Through the Seesaw
The rotation under the surface has been fast and disorienting. Some mornings the memory names are down while everything else is up; some days hardware leads, other days software; sometimes the Mag 7 rises while the broader market falls; other days everything rallies in unison. Traders have said openly that the constant flip-flopping is doing their heads in. So how do you invest around the AI trade when the attention span under the surface is this short, and would it be better to capture the whole thing through a single index rather than pick a category that could be up or down the next day?
The answer is to focus on who actually captures the value. Over the past year a rising tide lifted all boats, and almost anything with AI exposure had a significant run. That indiscriminate lift is changing. The move now is toward more exposure to names that can deliver durable franchise value based on where they sit in the infrastructure stack, and less to names that simply rode the theme. The short-term volatility itself reads as opportunity, because the macro noise is not breaking the long-term AI build-out; it is only making the near-term trading setup harder. For names with high long-term conviction, these dips can be a genuinely good buying opportunity.
Where Macro Actually Bites the Micro
Everything in the year-end targets points to earnings strength, and this market has seen a positive earnings shock led largely by tech. There is a running debate about the narrowness of the earnings profile, with the counterargument, associated with Mike Wilson at Morgan Stanley, that it is broadening out, which is encouraging. This is also a market that rallied straight through an entire war and still hit record highs. So where does the macro actually weigh on the micro, and is the AI-driven inflation flagged by Apple about to become a real problem?
The pressure is a combination of items on an ever-growing list. First, investors are digesting the Fed decision and growing more comfortable with having less visibility into whether policy turns more hawkish. Second, the Iran conflict and its effect on energy prices and inflation matters precisely because it hits the critical supply-chain geographies that feed the US AI build-out. Third, memory inflation is starting to trickle through, and that in turn feeds back into the Fed decision and any policy adjustments that follow. All of it converged at once, and that convergence was quick to overwhelm the strong Micron print from the week before.
The Next 30 Days
Looking at the coming month toward the end of July, a large earnings season arrives. ASML and TSMC report, and those critical players will offer far more visibility into capacity lead times and whether the industry is leaning into that capacity. The hyperscalers report toward the end of the month as well. If the other value-capturing players in the space echo what Micron showed, this stretch is a real chance for the micro to push back against the macro that has been sitting on top of it.
