Two of the most important names in the artificial intelligence supply chain are deepening their alliance. NVIDIA, the dominant maker of AI chips, and SK hynix, one of the world's leading memory manufacturers, have announced a multiyear partnership to co-develop the next generation of memory tailored to NVIDIA's AI ambitions. The collaboration is designed to align SK hynix's memory development with NVIDIA's hardware roadmap, which spans a wide range of products. These include the Vera Rubin AI supercomputers, Vera central processing units, RTX class machines and Jetson Thor robotics platforms. In simple terms, the more powerful AI systems become, the more they depend on fast and abundant memory to feed their processors, and the two companies want that memory ready in step with each new chip. There is a second strand to the deal. SK hynix plans to use NVIDIA's tools to speed up semiconductor simulation and to build so called digital twins, virtual replicas of its factories that allow operations to be tested and optimised in software before changes hit the real production line. That promises faster development and more efficient manufacturing. The partnership reflects a broader reality of the AI era. Building the data centres and devices that run advanced models is a marathon, with long development cycles and enormous demand for specialised components. Locking in a close working relationship between a chip designer and a memory supplier is a way to manage that complexity and to secure supply. For the wider industry, the tie up is another sign that the AI buildout is reshaping how the biggest technology firms cooperate. Rather than working at arm's length, designers and suppliers are increasingly planning their roadmaps together, betting that the appetite for AI computing power is only going to grow. Source: [Crescendo AI](https://www.crescendo.ai/news/latest-ai-news-and-updates)