Chip manufacturers have been racing for years to shrink their process node technology in an effort to improve performance, efficiency, and cost of processors made at their foundries.
Earlier this year Samsung announced it had already begun manufacturing chips using a 3nm process node. Now the company is laying out its plans for the next five years: Samsung expects to begin mass production of 2nm chips by 2025 and 1.4nm chips by 2027.
Samsung says it also plans to triple production capacity for “advanced nodes” by 2027, which means the company won’t just be cranking out more efficient processors, but it could have the capability to produce more of them than ever before.
Of course, a die shrink is only one of the things chip makers can do to improve performance of next-gen processors. And so Intel (which spent years trying to move beyond the 10nm node) has been making the case that what one company calls a 5nm or 7nm chip might not have all the features you’d find in a competing chip on a higher process node. So the company decided last year to stop using nodes to describe its manufacturing process (an “Intel 4” chips, for example, will be manufactured on a 7nm node).
That said, AMD’s latest mobile and desktop chips have been manufactured using a smaller node than Intel’s latest chips… and while Intel’s chips often continue to take the lead in single-core performance, they also tend to be far more power hungry than AMD’s chips while offering more of a mixed bag in multi-core tests.
So maybe there’s something to this whole die shrink thing after all.