inference-cost
5 articles tagged with this topic
Nvidia's Moat Is Spilling from GPUs to the Networking Layer
August 2026: Nvidia's AI advantage spills from GPU compute into data-center traffic scheduling. The networking layer is becoming the new moat source.
Nvidia's 70% Guidance: AI Capex Is Far From Peaking
Bloomberg Aug 27: Nvidia guided ~70% next-fiscal-year revenue growth, rebutting 'AI capex peak' narrative. Forward guidance—Huang's industry signal.
DeepSeek Standardizes Weekend Off-Peak Pricing: Monetizing Idle GPUs
From Aug. 23, DeepSeek will charge weekend API calls at off-peak rates, via demand shaping to monetize idle GPUs—an AWS Spot model for LLM inference.
Behind Nvidia's 15% Price Hike: The Inference Cost Curve Begins to Reverse
Nvidia raised prices 15%+ on HBM costs. AI inference cost-decline curve hits the memory wall—HBM supply sets token economics for 12-18 months.
Anthropic Recruits Google TPU Veteran, Custom Silicon Gambit Begins
Anthropic recruits Google TPU's Amir Salek—first concrete anchor for custom silicon. Covers token economics and vertical integration implications.