Why Google’s New TPU Strategy Carries More Strategic Weight Than a Routine Silicon Update This note is not intended to be a chip-versus-chip architectural comparison, nor is it an attempt to rank Google’s latest TPU designs against competing silicon. That is a different exercise, and in many ways a less useful one at this stage.…

A fresh perspective is needed on foundry economics at the leading edge. For years, the industry lacked a clear cost-per-transistor framework, even as the economic foundation of Moore’s Law was becoming less durable. That framework held through N3E. It no longer holds at N2 and A16, and the evidence is now strong enough to reframe…

Neoclouds sit at the center of the AI infrastructure durability debate because they are increasingly the external channel through which hyperscalers secure additional power and compute to convert demand into revenue. CoreWeave ended 2025 with $66.8 billion in contracted backlog. Nebius signed a deal with Meta worth up to $27 billion. IREN landed a $9.7…

Prerequisite: 800 VDC: The Inflection Point Reshaping Datacenter Power and AI Infrastructure (March 2026). This report builds directly on that analysis and assumes familiarity with the power architecture transition it describes. 800 VDC: The Inflection Point Reshaping Datacenter Power and AI Infrastructure BEN BAJARIN · MAR 31 Read full story The Thesis   Our view…

This report is a companion to our memory report found here: Memory’s $200B Inflection BEN BAJARIN · FEB 19 Read full story The market still tends to frame AI infrastructure demand through compute. In that view, storage is a downstream beneficiary. More accelerators ship, more storage gets attached. We think that framing is becoming less…

We believe the debate in memory has evolved from whether AI has created a real super cycle (the market has largely accepted that point) to what kind of market memory is becoming, how long the current earnings window can last, and what the first real signals of normalization will look like. Our view is that…