Google: AI Server Memory Costs Exceed 75%, Dual Hardware/Software Strategy

The fact that memory now dominates AI server costs signals a structural shift in where value accrues in the AI hardware stack. Google's move to pair custom silicon with software-level memory management suggests that optimization, not just raw compute, is becoming the competitive battleground. This affects cloud providers and chip designers alike, as they face pressure to innovate around memory efficiency. The open question is whether this dual approach can meaningfully bend the cost curve, or if it merely redistributes spending within the system.