Slicast
Tuesday, September 15, 2026
DarkSubscribe
Slicast
AI Infrastructure · News & Analysis
Home › Chips & Hardware › Report
Chips & Hardware · Report

Vera Rubin NVL72 architecture achieves 67x better performance per dollar in agentic inference workloads versus competing accelerators.

Superior agentic inference cost-efficiency could drive adoption of this architecture for long-context, reasoning-heavy AI agent deployments.
Trade pressSlicast · September 14, 2026 at 22:08 UTC · US · Source: SemiAnalysis
importance 65
Read the original
Companies in this story
NvidiaCapex history →
In-depth analysis
CommentaryNvidia's $10 Billion Anthropic IPO Stake, September 2026CommentaryBroadcom vs Nvidia: Custom XPU Versus Merchant GPU in the Inference Build-Out
Related reports
Chips & HardwareSamsung invests in Euclyd's $231 million Series A funding round; Euclyd develops AI inference chips positioned as an Nvidia alternative.2026-09-15Chips & HardwareSamsung invests in and adopts Euclyd AI accelerator chip, broadening inference AI chip optionality beyond Nvidia.2026-09-15Chips & HardwarePerplexity launches Portable Computer, a local AI agent for Windows PCs with NVIDIA RTX GPUs that handles multistep task2026-09-15Chips & HardwareSamsung leads $230 million funding round in unnamed Dutch AI chip startup seeking to build Nvidia alternatives for inference and training.2026-09-15Chips & HardwareAI inference chip startup d-Matrix, backed early by Samsung and SK, formally joins Nvidia's ecosystem.2026-09-15
Vera Rubin NVL72 architecture achieves 67x… · Slicast