Back to home

inference

7 articles tagged with this topic

NVIDIADSX MaxLPS

NVIDIA's DSX MaxLPS: AI Data Center Competition Shifts From GPU Count to Watts

NVIDIA's DSX MaxLPS shifts AI data center KPIs from GPU count to tokens per watt. Power, not silicon, is now the bottleneck.

Aug 212 min read
googlegemini

Google Changed Not the Quota, but the Pricing Logic of AI

Google’s Gemini quota tallying change looks administrative on the surface, but signals a deeper shift from per-use AI pricing to resource-intensity pr

Jul 182 min read
chinaenergy

When the Power Grid Starts Pricing Tokens for AI Inference

At WAIC in Shanghai on July 17, China’s energy regulator shifted the AI debate from training capacity to inference load as a real power constraint.

Jul 182 min read
byt edancechina

Byt eDance Doubles Down on Infrastructure , Not Models

ByteDance raised its 2026 AI infrastructure budget 25 % to R MB 200 billion, sign aling that China 's top consumer platforms are in ternal izing AI su

May 92 min read
cereb rasipo

Cerebras Price H ike: More Than Just IPO Momentum

Cerebras plans to raise its IPO pricing range— superf ic ially strong primary market demand, but fundamentally capital markets repr icing 'non-Nvidia

May 82 min read
coreweavegpu

CoreWeave Is No Longer Just a GPU Landlord

Core Weave's Q1 was labeled 'transformational' by its CEO— not just for revenue and margin gains , but for demand expanding from AI-native to trading,

May 82 min read
hon-hainvidia

Hon Hai's 30% Growth Is Not Just a Contract Manufacturing Story

Hon Hai's April 2026 revenue rose 29.7% Y oY. Surface narrative : AI server demand. Real signal : Nvidia's supply chain hasn 't entered dest ocking. W

May 52 min read