Intel has already committed to its tick-tock cadence where a tick would represent a reduction in the fabrication process, tock would represent enhancement while continuing using the previous manufacturing process. A leaked roadmap by WikiChip now confirms that Intel will indeed stick to its two years cadence and will reach 1.4nm by 2029.
As seen in the roadmap, it is also seen that Intel will invest in a yearly cadence of + / ++ nodes resulting in 10nm+++ in 2021. It is from this slide that Intel has for the first time unveiled that it is working on 3nm and 2nm in pathfinding and 1.4nm in research. While intel will have 10nm+++, they will simultaneously have 7nm process node.
What these slides don’t show is how Intel aims to achieve this shrink in the fabrication node. When they will forfeit their trigate 3D transistor (Intel’s version of finfet) to gate-all-around. Insertion of EUV is also on the cards for 7nm process nodes.
Intel has also shown backporting opportunities at every current node to the previous node’s ++ version. This could mean that Intel either intent to backport features from new process node to the previous process node or complete new chip could be backported to a previous mature process node.
Intel through this roadmap has committed for the next 10 years to stick to existing technologies and make improvements to adhere to Moore’s law. However, historically speaking, they have delayed their process node by very much as Intel has yet to touch 7nm. Any delay in a future node would mean delay in 1.4nm which Intel hopes to ship in 2029. The pressure is really on Intel as they are now lagging compared to other foundries like TSMC and Samsung.
What we still dont know what lies beyond 1.4nm as the quantum disturbance is now causing each shrink in nodes more difficult then before. Maybe Quantum computers will make the next breakthrough by then.