For the first time in two decades, NVIDIA will become the starting customer of the most advanced TSMC process-1.6-Nm A16.
Typically, the company relies on architectural improvements and switches to new buttons later than Apple or Qualcomm.
But the rapid growth of the market AI is forced to change the approach: NVIDIA itself will receive the first chips on the A16, which will bring gaafet semiconductor balls and the spring system with data supply from the back of the crystal.
These innovations will significantly increase the efficiency of energy and expansion of solutions for data centers.
© WCCFTECH
According to the TSMC plan, the mass production of A16 will start half in 2026 and the first NVIDIA products on this basis are expected in late 2027 or early 2028. It is likely that we are talking about the Feynman line, where the company plans to introduce the entire architectural innovation package.
For TSMC, this is an important contract: NVIDIA has become the largest player in the field of AI and the early approach to A16 will fix the leadership of Taiwan manufacturer in the advanced chips market.