Gordon Moore’s law cadence means that the number of transistors in a dense integrated circuit doubles every approximately two years. However, Intel has now confirmed that the next transistor density manufacturing process of 10nm has been delayed until 2017 and the company will stretch the lifespan of its 14nm process with its “Kaby Lake” CPU lineup.
Apparently, Intel, which was co-founded by Gordon Moore himself, admits that higher density wafers will now take two and a half years to make instead of the clean two it used to take in the past. As a result, Intel now has to take into account developing not two but three CPU families using the same process technology.
The company believes that a third family of CPUs will be a smooth transition to the next process manufacturing technology. However, the truth remains that moving to 10nm isn’t an easy feat, and Intel confirms that technological issues have made the new process technology unattainable until the near future.
This way, the “Cannonlake” processors made using the 10nm manufacturing process will clearly be delayed until the second half of 2017, while the next year will help Intel improve and polish off its 10nm process and ensure a steady high-volume of new chips at launch.
Intel considers that the more you scale down the lithography, the more difficult it is to make it functional. The company also confirms that the new process technology uses FinFET transistors with enhancements and immersion lithography with a lot of multi-pattern steps.
It’s very likely that Intel won’t take any shortcuts just to get to the 10nm in time. The new FinFETs will feature smaller transistor fin pitch, transistor gate pitch, as well as interconnection pitch compared to 14nm technology in a bid to maximize transistor density.
Taking this new Tick-Tock cadence into consideration, Intel admits that after the next three product families based on the 14nm and 10nm processes, the real change of pace will come when developing the 7nm process with EUV (Extreme Ultraviolet Lithography) in 2020.
However, it is worth noting that IBM managed to spectacularly leapfrog the 14nm and 10nm process technologies by announcing the world’s first 7nm FinFETs this summer. Although IBM also considers that mass EUV printing is still a thing of the future, its process technology breakthrough means that patents can be sold, and through a trans-corporate effort, a solution to 7nm mass production will arrive at the turn of the decade.
KitGuru believes that the rumored 10nm “Ice Lake” CPU will probably arrive in 2018 and have its lifespan stretched to three years. Unless Intel finds a way to reliably use EUV printing by 2019.