The biggest foundry in the world belongs to Taiwan’s TSMC. This is the company that turns Apple’s chip designs into actual chips such as the A15 Bionic (which runs the iPhone 13 series and features 15 billion transistors in each chip). TSMC is also responsible for the M1 chip line which includes the M1 Ultra powered by its 114 billion transistors. The M1 Ultra is made by combining two M1 Max chips.
The next major process node will be 3nm which Apple hopes to use on next year’s iPhone 15 series. TSMC expects to have five N3 nodes over the next three years. Instead of releasing a new node every two years which was the custom for the foundry and the industry, TSMC will now introduce a new node every two and a half years increasing to every three years with the N2 (2nm) process node. This data is known as the node introduction cadence.
TSMC will continue to use FinFET field-effect transistors for its 3nm process node while Samsung will debut its gate-all-around transistors with its 3nm chips). On the other hand, TSMC won’t begin using gate-all-around transistors until it starts to ship its 2nm chips. TSMC experts its more cutting-edge chip designing customers to quickly demand N2 when available while the foundry’s less tech-demanding customers will probably decide to stick with a 3nm process node for the next few years to come.
The N3E node reduces power consumption by 34%, or delivers an 18% performance bump. By 2024, the foundry expects to offer its N3P node focusing on performance improvements. And N3S is the density-focused version of the 3nm node. Density tells us how many million transistors can fit into a square mm of space. High density chips allow for more circuits to be placed on a chip while delivering higher operating speed.
The 3nm process nodes will be the last from TSMC to deliver the FinFET transistor-based process nodes. These transistors use a design shaped like a “fin” which is where they get their name. TSMC will introduce its 2nm process node technology in 2025.