China’s Chitu AI challenges Nvidia, boosting domestic chip independence

China is making strides in AI chip self-sufficiency with the open-sourcing of Chitu, a large-scale model inference engine developed by Tsinghua University-backed AI startup Qingcheng.AI. The company has also launched a full-scale deployment solution, positioning Chitu as an alternative to Nvidia’s Hopper series and a crucial driver in China’s evolving AI ecosystem.

Continue reading

GTC 2025: Nvidia unveils AI GPU roadmap for 2026-2028 with Rubin and Feynman architectures

During the keynote address at Nvidia’s GTC, CEO Jensen Huang revealed the latest AI products and technologies, and shared insights into the product roadmap for 2026 to 2028. Huang announced that the Blackwell Ultra NVL72 will be launched in the second half of 2025, followed by the unveiling of the Vera CPU and Rubin GPU in the second half of 2026. The company plans to introduce Rubin Ultra and a new architecture codenamed after American physicist Richard Feynman, showcasing strong confidence in the future of AI.

Continue reading