Apple has discontinued its 128GB Mac Studio configuration, just two months after killing the 512GB option. The move points to larger supply constraints as the entire tech industry scrambles for memory chips to power on-device AI features. What looks like a product lineup adjustment is actually a symptom of the AI hardware crunch.The Mac Studio was Apple's workstation for professionals who needed serious computational horsepower. The high-memory configurations were particularly popular with AI researchers, video editors, and developers working with large datasets. Now the maximum configuration tops out at 96GB, a significant step down for users who actually needed those resources.The official explanation is "supply constraints," which is technically accurate but incomplete. The real story is that every tech company simultaneously decided they need to run large AI models locally on consumer devices. Apple is betting big on on-device intelligence with Apple Intelligence, Google is pushing local AI features, and Microsoft is building AI into Windows. They're all competing for the same memory chips.This is what happens when an entire industry pivots at once. Memory manufacturers can't instantly scale production to meet a sudden spike in demand, especially for the high-bandwidth memory needed for AI workloads. The chips that would have gone into high-end Mac Studios are probably being allocated to products with higher volume or strategic priority.For professional users, this is frustrating but predictable. Apple has a history of discontinuing configurations that serve niche markets when components become constrained. The company would rather ensure availability of mainstream configurations than maintain options that only a small percentage of users need. That's a rational business decision, but it leaves professionals scrambling for alternatives.The broader question is whether this supply crunch is temporary or structural. If on-device AI becomes standard across consumer electronics, demand for high-bandwidth memory will remain elevated. We might be looking at a multi-year period where high-memory configurations are limited or expensive across the industry.The local AI frenzy is creating real hardware bottlenecks. The technology is impressive, but the physics of semiconductor manufacturing doesn't care about product launch timelines.