A GPU, or graphics processing unit, is a processor built to perform many calculations in parallel. Although originally designed for rendering graphics, this structure is well suited to the matrix and vector operations at the center of deep learning. GPUs accelerate both model training and inference, and multiple units can be connected to handle larger models or datasets. Performance depends not only on raw arithmetic speed but also on memory capacity, memory bandwidth, interconnects, numerical formats, cooling, and software libraries. Consumer, data-center, and integrated GPUs serve different workloads, so the most expensive chip is not automatically the most efficient choice for every AI application.
NVIDIA is reportedly in advanced talks to provide a roughly $250 billion financing guarantee for a vast OpenAI data center in Ohio, potentially helping the ChatGPT maker secure better terms for one of the largest AI infrastructure projects ever proposed.
Google is designing Frozen v2, a chip that hardwires parts of its Gemini model into silicon and could serve six to ten times more tokens per watt, targeting deployment by 2028.
OpenAI launched Codex Micro, a $230 mechanical keypad built with Work Louder that gives developers physical controls for its Codex coding agent, its first hardware to reach market.
SK Hynix raised $26.5 billion in its Nasdaq debut, the largest US listing ever by a foreign company, as investors rushed to buy into the AI memory chip boom.
Meta plans to begin manufacturing its in-house Iris AI chip in September as it aims to double computing capacity to 14 gigawatts next year and cut its reliance on Nvidia.
Apple expanded its partnership with Broadcom in a multi-year deal worth more than $30 billion, its largest US manufacturing commitment, to produce over 15 billion American-made chips.
Nvidia’s Kyber rack for its 2027 Rubin Ultra chips has slipped more than a year to 2028 over a hard-to-manufacture circuit board, SemiAnalysis says, with no proven fallback.
OpenAI teased its first branded hardware, Codex Micro, a programmable macro pad built with Work Louder to give developers physical shortcuts for its Codex coding agent, launching July 15.
A record European heatwave is highlighting a growing threat to the AI boom: extreme weather that stresses data centers and the power grids they depend on at once.
SK Hynix filed to raise about $29 billion through a Nasdaq ADR listing set for July 10, roughly double earlier estimates, to fund its AI memory expansion.