$AAPL getting back into servers would be a pretty big shift. What makes this more interesting is that Apple wouldn’t be going back to the old server business exactly as it was. The idea being explored is an AI system built around Apple’s own M-series chips, while using $NVDA NVLink Fusion to connect into NVIDIA’s ecosystem. And the target market is pretty specific: private AI inference for developers, enterprises and governments. That part matters. A lot of companies want AI capabilities without sending sensitive workloads and data entirely into public clouds. If Apple can combine its silicon, software ecosystem and privacy positioning with NVIDIA’s high-speed interconnect technology, there’s a potentially interesting enterprise product here. The proposed timeline is 2029, so this is still very early. But Apple building its own AI infrastructure while NVIDIA provides part of the connectivity is definitely not something I expected to see.