Search results for AI inference
AI & Enterprise
Google Cloud CEO says AI market to be reshaped in 1 to 2 years, economics will determine survival
Google Cloud CEO Thomas Kurian said a full-stack AI strategy, in which Google builds chips, data centres, foundation models and products in-house, can help narrow the gap with Amazon Web Services. He said owning IP, models and chips allows greater investment. Google Cloud’s market share rose to 14 percent from 7 percent in eight years, and the company expects revenue to exceed $70 billion this year.
Mobility
Tesla HW4 may soon be outdated as HW4 Plus with double memory unveiled
Tesla is preparing an upgraded successor to its HW4 self-driving computer, dubbed HW4 Plus. Elon Musk (일론 머스크) said on a quarterly earnings call that a new AI4.1 or AI4 Plus chip will double memory, raising RAM per chip to 32 GB and total system memory to 64 GB. He also acknowledged HW3 struggles with unsupervised FSD, citing memory bandwidth as a key bottleneck.
AI & Enterprise
SKT to work with Arm, Rebellion on AI inference server solution
SK Telecom is developing an AI inference server solution that combines a graphics processing unit and a neural processing unit. The company signed an MOU with chip designer Arm and AI chip startup Rebellion to jointly develop a solution that uses Arm’s Arm AGI CPU and Rebellion’s Revel Card, due in the third quarter. SKT plans to validate the solution at its AI data center and review running its A.X K1 model on it.
-
Industry
SK Hynix earnings surprise likely as NAND exports surge
-
Industry
AD Technology targets North American AI-RAN and DRAN markets
-
Industry
Mobilint to work with POSCO DX on NPU-based industrial AI development
-
Industry
Fadu unveils Gen6 controller, FlexSSD at China flash market summit
-
AI & Enterprise
KAIST professor involved in TurboQuant development says it will be core basis for running large-scale AI models
-
Industry
Google TurboQuant sparks memory chip selloff, shocking Hynix and Samsung investors
-
AI & Enterprise
AI agents reshuffle data centre CPU market as Nvidia and Arm join in
-
AI & Enterprise
Snowflake steps up strategy to run AI work inside its data platform
-
AI & Enterprise
Gimlet raises $80 million as it targets AI inference bottlenecks with multi-silicon inference cloud
-
AI & Enterprise
Samsung SDS launches South Korea\'s first B300 GPU service to boost enterprise AI inference
-
AI & Enterprise
Nvidia shares muted despite grand vision, Wall Street-Silicon Valley divide
-
AI & Enterprise
Tech Insight: Nvidia is now an AI infrastructure platform company
-
AI & Enterprise
Science ministry, Financial Services Commission hold joint meeting on \'K-Nvidia project\'
-
AI & Enterprise
Nvidia unveils Groq 3 LPU inference chip for multi-agent workloads
-
Industry
Jensen Huang visits Samsung booth, picks HBM4 and Groq LPU supply chain