Yesterday, the Intel Q2 2020 earnings report brought more grim news for the company’s advanced manufacturing processes. Its next-generation 7nm manufacturing process now lags a full year, with those parts slated to see the light no later than the end of 2022.
Intel’s 14nm barrier
Intel’s struggles with 7nm development and manufacturing follow what can be generously described as an unsuccessful transition to 10nm. In March this year, Intel Chief Financial Officer George Davis described the company’s 10nm process (used in its current line of CPUs for Ice Lake laptops) as “[not] the best node Intel has ever had, “he goes on to say that 10nm Intel would be” less productive than 14nm, less productive than 22nm … it will not be as strong a node as people would expect at 14nm or what they will see in 7 nm “.
These struggles for higher clock speeds and better 10nm performance rates have forced Intel to continue to rely on its 14nm aging process, now so old and frequently revised that it is often referred to as “14nm ++ ++ “. Ice Lake 10nm laptop CPUs are far from useless; Due to their higher integrated GPU performance and power efficiency, they are a premium choice for devices with limited battery. But even on laptops, Intel Ice Lake 10nm competes directly against Intel Comet Lake 14nm, with the highest performing Intel parts coming from the above process.
When Davis admitted to these setbacks with 10nm in March, he resisted 7nm as Intel’s opportunity to regain parity with other manufacturers, making this week’s news that 7nm has fallen six months earlier is particularly grim. Assuming there are no more mishaps, this places Intel’s eventual 7nm debut at roughly the same time as the projected 3nm parts from rival foundry TSMC. (TSMC is one of the world’s largest foundries; Intel AMD’s rival is one of TSMC’s clients.)
Intel’s nomenclature for process size is slightly different from TSMC: a 7nm Intel process is approximately the same actual transistor density as a 5nm TSMC process. But that distinction, while important, does not nullify projections of Intel being a complete process behind its competition for another complete architectural cycle.
Fighting for alternatives
With its internal manufacturing so far behind, Intel is beginning to look for “more aggressive” outsourcing strategies. As one of the world’s historically largest and most successful chip foundries, Intel has used third-party fabs in the past for very little of its production. It is usually only based on third-party fabs for cheap non-CPU products based on older processes.
Given his continued struggles with 10nm and now 7nm, CEO Bob Swain says the company is looking for a more “pragmatic” approach to using third-party foundries. This could mean more critical components, such as GPUs or even CPUs that come from outside Intel, with the company’s advanced multi-chip packaging technologies used to add different dies in a single package.
Relying on third-party sources for mission-critical parts exposes Intel to a host of pressures it has previously avoided by keeping things at home: outsourcing components can mean tight margins and even supply guarantee struggles, with Intel bound to compete with its competitors like AMD and Nvidia for their production capacity in the same third-party manufacturers.
In March, Intel planned to shorten the 10nm production cycle, relying on a fast 7nm rollout to regain parity with its competition. But with this week’s 7nm rollback announcement, it’s getting 10nm running again: The company now says it will boost its 10nm CPU send by 20 percent and get another “full performance node” from the 10nm process. That means we will search more than 10nm before seeing a 7nm debut.
Intel’s first 10nm desktop CPU architecture, Alder Lake, is expected to hit the market sometime in the second half of 2021.
