⚡SemiAnalysis: Custom ASICs Will Outgrow Nvidia GPUs in 2026
TL;DR
TrendForce data cited by SemiAnalysis shows custom AI ASIC shipments growing 44.6% in 2026 vs 16.1% for merchant GPUs — the first inversion of the AI era. ASICs hold a 40–65% TCO edge in multi-year inference deployments per SemiAnalysis and Bernstein.
TrendForce data cited by SemiAnalysis shows custom AI ASIC shipments growing 44.6% in 2026 vs 16.1% for merchant GPUs — the first inversion of the AI era. ASICs hold a 40–65% TCO edge in multi-year inference deployments per SemiAnalysis and Bernstein.
Key Points
Custom AI ASIC shipment growth: 44.6% in 2026 vs merchant GPU at 16.1%
Estimated 40–65% TCO advantage for ASICs in long-term inference workloads
Buying criteria has shifted from peak FLOPS to cost-per-token, power, utilization
Google TPU, AWS Trainium, Microsoft Maia and Meta MTIA all hitting full production scale
Nvidia still holds ~81% of data-center AI chip revenue per IDC
Why It Matters
Inference is the new TAM, and inference workloads are where ASICs win on TCO. Nvidia's pricing power is intact for now, but the gross-margin profile of the next decade is being decided this year.
Quick Facts
Frequently Asked Questions
Why does this matter?
Inference is the new TAM, and inference workloads are where ASICs win on TCO. Nvidia's pricing power is intact for now, but the gross-margin profile of the next decade is being decided this year.
What happened?
TrendForce data cited by SemiAnalysis shows custom AI ASIC shipments growing 44.6% in 2026 vs 16.1% for merchant GPUs — the first inversion of the AI era. ASICs hold a 40–65% TCO edge in multi-year inference deployments per SemiAnalysis and Bernstein.
Comments
Be the first to comment
Enjoyed this article?
Get it daily. 7am. Free. Reads in 5 minutes.