U.S. demand empties Mac inventory

Mac desktops are suddenly a supply story — not because Apple launched something new this week, but because a weirdly specific kind of buyer showed up all at once. Developers and small teams have been buying Mac mini and Mac Studio machines as local AI boxes, and Apple now says supply for those models could stay tight for several months. The interesting part is not just the shortage. It’s what kind of shortage this is. ### Why are Macs selling out? The short version is memory. Running AI models locally gets expensive fast, and Apple’s unified memory setup makes higher-end Mac mini and Mac Studio configurations unusually attractive for people who want to run models on their own desks instead of renting cloud GPUs. Apple said on its April 30 earnings call that Mac mini and Mac Studio may take “several months” to get back into supply-demand balance. ### Why these Macs in particular? Because they hit a sweet spot. A Mac mini is cheap by workstation standards, quiet, power-efficient, and easy to stack or deploy in batches. A Mac Studio pushes further for buyers who need more memory headroom. That does not make them better than a rack of Nvidia servers — obviously not — but it does make them very good for local inference, agent workflows, testing, and small-team development where cloud bills pile up fast. Apple’s own quarterly results also showed Mac revenue rising to $8.4 billion in fiscal Q2 2026, up 6% year over year, which lines up with a real demand bump rather than a rumor cycle. ### What actually disappeared? The high-memory SKUs are the ones getting squeezed first. MacRumors reported on May 5 that Apple removed Mac mini configurations with 32GB and 64GB of RAM from its online store, and also cut the Mac Studio’s higher-memory options, including the M3 Ultra model with 256GB RAM. Earlier reports in April showed some Mac mini and Mac Studio configurations marked “currently unavailable,” which usually means Apple has stopped taking orders for that exact build. ### Is this just a website glitch? Probably not. The pattern has been consistent for weeks — first long shipping estimates, then “currently unavailable” labels, then full removal of some configurations from Apple’s configurator. Cult of Mac and AppleInsider both saw lead times stretching from weeks to months depending on memory choices, with base models often easier to get than upgraded ones. That is exactly what you would expect if the bottleneck is memory-rich configurations rather than the whole product line. ### So is the real bottleneck DRAM? That looks like a big part of it. Multiple reports tie the crunch to tighter global memory supply and higher DRAM pricing as AI server demand soaks up components. That matters because the Mac mini and Mac Studio story is not “Apple forgot to build enough boxes” in the simple sense. It’s “the exact parts needed for AI-friendly Mac configs are also in demand elsewhere,” which is a nastier problem. ### Does Apple itself frame it as AI demand? Yes. Apple explicitly tied the imbalance to stronger-than-expected demand, and third-party coverage of the earnings call consistently points to local AI workloads as the driver. Apple did not announce some grand “AI Mac” pivot here. Turns out the market did that part for them by deciding these machines are practical local model hosts. ### Could this also be about an M5 refresh? Maybe a little, but that does not explain the whole pattern. Refresh timing can make Apple less eager to restock every possible variant, and some observers have floated that idea. But when Apple is also warning about months-long imbalance and the worst shortages cluster around high-memory builds, demand plus component pressure is the cleaner explanation. ### What’s the bottom line? This is a small but telling AI market shift. The shortage says more people now want personal or team-owned AI compute than Apple seems to have planned for. If that keeps up, Macs stop being just developer laptops and desktops — they become part of the local AI infrastructure stack.