Nvidia’s main goal is to design AI chips

Nvidia Jensen Huang’s main goal is to have AI design the chips that run AI.

AI-assisted chip design of the AI ​​H100 and H200 Hopper chips. Jensen wants to use AI to combinatorially explore the entire chip design. He wants AI-assisted cross-modular code design.

They would optimize the thousands of chip cores that make up GPUs.

AI is also used to optimize AI data centers. The new computing unit is the data center.

Nvidia creates massive digital twins of all of its data centers before they are built and implemented.

They pre-organized everything related to the supply chain and created prototypes of sections of the data center.

Nvidia had as its product a multi-layered project called data center.

Taalas and Etched are two AI chip startups aiming to challenge Nvidia’s dominance in the AI ​​hardware market. Etched uses ASIC chips. Taalas uses eASIC chips. They both put logic in hardware instead of using software on top of GPUs. It’s possible to achieve 100 to 1,000 times more energy efficiency than more generalized hardware where software drives AI.

It is likely that Nvidia will also use a section of chips with ASIC or FPGA capabilities to directly access the hardware’s computational efficiency.

Fully AI-designed and optimized chips and data centers will likely deliver 10,000 times better performance.