TSMC has shared new plans for its next-gen chip packages, where the company plans over 1 trillion transistors on its next-gen A10 process node by the year 2030... identical to what Intel has planned, and announced last year.
Taiwan Semiconductor Manufacturing Company (TSMC) shared the news during the recent IEDM conference, where it lists its newer N5 and N4 process nodes, as well as N3 and N3E process nodes. N5 and N4 began their lives back in 2020, while N3 and N3E process nodes started out this year in 2023, while N2 and N2P process nodes will be unleashed in 2025.
NVIDIA's upcoming H200 AI GPU is a monolithic chip with 80 billion transistors, while AMD's upcoming Instinct MI300X AI GPU uses a chiplet design with a whopping 153 billion transistors using a blend of 5nm and 6nm IPs on the MI300X. However, TSMC teases it will have 200 billion transistors on a single monolithic chip with its N3 and N3E process nodes very soon.
- Read more: TSMC is so ahead of the game, Samsung might not catch up until 2030
- Read more: Intel targets 1 trillion transistors on a single package by 2030
- Read more: Intel CEO: we want 1 trillion transistors in a single package by 2030
From there, TSMC plans its new N2 and N2P process nodes for some time between 2025 and 2027, while its new A14 (1.4nm) and A10 (1nm) chips will have over 1 trillion transistors by the year 2030. TSMC will have a bunch of new semiconductor technologies it'll use to cook up its leadership in the semiconductor industry.
- Read more: NVIDIA H200 AI GPU: up to 141GB of HBM3e memory with 4.8TB/sec bandwidth
- Read more: AMD Instinct MI300X: new AI GPU with 192GB of HBM3 at 5.3TB/sec bandwidth
- Read more: AMD Instinct MI300X AI GPU teased: 153 billion transistors, 192GB of HBM3 memory
TSMC teases that we'll see a next-gen monolithic GPU with over 200 billion transistors next year, which I'm sure the world can't wait to see. NVIDIA's next-gen "Blackwell" B100 AI GPU is going to be unleashed next year, so the AI GPU fight will only intensify in 2024 (and beyond).