hyperscaler-capex - Next Waves Insight

Custom Silicon Is Winning Specific Workloads. What That Means for the AI Supply Chain.

April 3, 2026April 2, 2026 by Arjun Mehta

3 min readEvery major hyperscaler is running a significant fraction of AI inference on custom silicon — Google TPU v5, Meta MTIA, Microsoft Maia, Amazon Trainium. NVIDIA retains dominance in training, but the inference market is structurally bifurcating.

$300 Billion in AI Infrastructure Spending: Where the Hyperscaler Capex Is Actually Going

April 3, 2026March 26, 2026 by Sarah Chen

3 min readMicrosoft, Google, Amazon, and Meta will collectively spend over $320 billion on AI infrastructure in 2026 — but more than 60% is going into power infrastructure and data centre construction, not compute hardware. The binding constraint has shifted from chips to electricity.