Financially conflicted. How Nvidia nudged Micron onto the AI train. Kültürkampf & Chinmaxxing Russian elites. - Adam Tooze | Substack

This article explores how NVIDIA's dominant position in the AI sector prompted Micron Technology to shift from a traditional memory supplier to a key player in the AI chip market. As AI data centers e... Read original →

NVIDIA’s architectural dominance in AI is compelling Micron to evolve from a cyclical DRAM vendor into a strategic node within the AI hardware stack. Technically, the tight integration of HBM3E with CoWoS packaging has transformed memory from a passive component into a critical bottleneck for compute density—forcing Samsung and SK Hynix to accelerate 3D TSV and hybrid bonding R&D. On compliance, Micron’s five-year supply deal with NVIDIA locks revenue but heightens exposure to U.S. secondary sanctions; if HBM falls under BIS export controls, production lines in Taiwan, China and Korea face disruption. Strategically, Samsung may counter with 3nm GAA logic-HBM vertical integration, while AMD partners with SK Hynix on MI300X-custom stacks to break the NVIDIA-Micron axis. Over the next 18 months, HBM capacity—not AI chips—will be the true delivery bottleneck, signaling a valuation paradigm shift: memory firms are now priced on architectural indispensability, not commodity cycles.