Jensen Huang stepped away from his position as CEO of Nvidia, according to a November 14, 2026 Tom's Hardware Premium roundup. The departure marks a significant shift for the company, which transformed from a gaming GPU manufacturer to an AI-focused business over the past five years. The roundup also covered AMD's MEXT predictive memory engine technology, which the company stated runs on 145 servers and aims to retain 99% of server performance while using half the DRAM, with plans to expand to 2,000 servers in 2027.
AMD MEXT Technology Runs on 145 Servers, Targets 2,000 in 2027
AMD shared details about MEXT, a predictive memory engine designed to address industry-wide DRAM capacity constraints. The company stated the system offers 99% of a server's performance while running on half the DRAM. AMD reported that 145 servers currently run with the system in place, with plans to expand that figure to 2,000 next year. The technology is powered by a transformer model and uses flash storage.
Google Announces TPU 8i Max and TPU 8tx Chips for AI Workloads
Google expanded its 8th-generation TPU lineup with two new chips designed for long-context agentic AI and security workloads. TPU 8i Max doubles the specifications of the original TPU 8i by placing two chips on one package. TPU 8tx incorporates AMD EPYC processors for specialized security workloads, replacing the Arm-based hosts used in other TPU variants.
Cognichip Revealed AI Chip Design Tools in September
Cognichip's Chief Product Officer discussed the company's AI-powered chip design tools in an interview covered by the roundup. In September, the company revealed a suite of proprietary AI models named Artificial Chip Intelligence (ACI), developed specifically for semiconductor engineering. The tools automate labor-intensive tasks, though they have not yet been used in a shipping product.
Nvidia Export-Compliant RTX 5500 Pro Sees Over 6 Million Units Ordered
The roundup reported that over 6 million units of the RTX 5500 Pro, an export-compliant GB202-based AI accelerator, have been ordered. Chinese AI labs face challenges obtaining Nvidia-based compute due to export controls. Some companies have moved to renting compute in Southeast Asian countries like Malaysia and Singapore. Black market B300 rack prices reached $2.8 million, and Nvidia partners face a 51% non-Chinese quota requirement.
FAQ
What did Jensen Huang do at Nvidia?
Jensen Huang stepped away from his position as CEO of Nvidia, as reported in a November 14, 2026 Tom's Hardware Premium roundup. The company transformed from a gaming GPU manufacturer to an AI-focused business over the past five years under his leadership.
How does AMD's MEXT technology reduce DRAM requirements?
AMD's MEXT predictive memory engine aims to offer 99% of a server's performance while running on half the DRAM. The system is powered by a transformer model and uses flash storage. AMD stated that 145 servers currently run with MEXT, with plans to expand to 2,000 servers in 2027.
What are Google's new TPU chips designed for?
Google announced TPU 8i Max and TPU 8tx as additions to its 8th-generation TPU lineup. TPU 8i Max is designed for long-context agentic AI workloads and places two chips on one package. TPU 8tx is designed for security workloads and incorporates AMD EPYC processors instead of Arm-based hosts.
