As technology improves, so too does the manufacturing process behind said technology. Graphics cards get faster, TVs get sharper, fridges get smarter etc. One of the reasons for this is that the actual chips behind some of these technologies are able to pack more power in them thanks to smaller process nodes.
For instance, if you can pack the same amount of transistors into a smaller space, then you can effectively increase the amount of transistors in the same space as well. Nvidia’s latest Ampere graphics cards, the RTX 30 series, use a 7nm process node. AMD’s Ryzen 5000 series also uses a 7nm process node, and Intel’s upcoming Rocket Lake processors use the 14nm node size.
As those new products improve, generally the node size decreases. That’s why the 3nm process node is a big deal in the chip manufacturing industry, and TSMC is leading the charge first by starting production on the 3nm node size later this year - almost 1 year ahead of schedule. But there’s only one problem: Apple has bought all of them.
That’s not necessarily a bad thing though, since TSMC is still producing 7nm and 5nm node sizes. In fact, Apple is kind of a Beta tester for the new 3nm node, as there is a chance the new wafers won’t even work at all, or at least not up to expectations.
TSMC will be producing a total of 140,000 wafers per month for the 7nm node size this year, and 105,000 for the 5nm node. For 3nm? They’re just manufacturing a batch of 30,000 instead.
On top of that, TSMC is upping production of all 3 wafer sizes in the years 2023-2024, at around 160,000 wafers per month for both the 7nm and 5nm node sizes each, and 105,000 for the 3nm node.
Node size 2021 production (wafers per month) 2023-2024 production (wafers per month) 3nm 30,000 105,000 5nm 105,000 160,000 7nm 140,000
So whilst Apple will potentially get a much faster and more efficient chip than anyone else months before anyone outside of Apple can get their hands on it (3nm node size will offer roughly a 25%-30% reduction in power consumption compared to the 5nm node size, or roughly a 10%-15% performance improvement with the same power consumption), it’s not exactly like they will be dominating the market.
It just means PC users won’t be able to take advantage of the new 3nm node size until a bit later than expected.
Plus, Samsung is projected to start production on their own 3nm node size in 2022, and Nvidia has already made a deal with them for their RTX 30 series graphics cards, and AMD might be outsourcing their GPU and APU production to Samsung in order to help with stock issues. So it’s not the end of the world for PC users.
What do you think? Is it good that Apple will Beta test the 3nm node size? Will waiting a little bit longer for the 3nm node be that bad? Let us know your thoughts!