Intel vs. TSMC Process Nodes Battle: Speed or Density – Which Wins?

TechInsights and SemiWiki recently unveiled significant insights into Intel and TSMC’s forthcoming 18A (1.8nm-class) and N2 (2nm-class) process technologies at the International Electronic Devices Meeting (IEDM). Reports from TechInsights suggest that Intel’s 18A technology could deliver superior performance, while TSMC’s N2 might boast a higher transistor density.

According to analysis by TechInsights, TSMC’s N2 technology achieves a high-density (HD) standard-cell transistor density of 313 MTr/mm^2. This density significantly surpasses that of Intel’s 18A at 238 MTr/mm^2 and Samsung’s SF2/SF3P at 231 MTr/mm^2. Although these figures align well with the SRAM cell sizes for the 18A, N2, and N3, as well as TSMC’s projections for N2 and N3, there are additional factors to consider.

Firstly, the focus here is solely on HD standard cells. Almost all contemporary high-performance processors operating on cutting-edge nodes utilize a combination of high-density (HD), high-performance (HP), and low-power (LP) standard cells, along with features like TSMC’s FinFlex and NanoFlex.

Secondly, the comparative data on Intel’s and TSMC’s HP and LP standard cells is not clear. While it’s reasonable to assume that N2 leads in transistor density, this lead may not extend as significantly to HP and LP cells. Thirdly, both companies showcased the performance, power, and transistor density benefits of their 18A and N2 processes over their predecessors at the IEDM event, yet direct comparisons between these technologies remain elusive.

In terms of performance, TechInsights posits that Intel’s 18A could potentially outperform both TSMC’s N2 and Samsung’s SF2 (previously known as SF3P). However, the methodology used by TechInsights for this comparison, which involves using TSMC’s N16FF and Samsung’s 14nm processes as baselines and incorporating announced improvements, might not provide completely accurate predictions.

Moreover, Intel’s 18A is designed with a focus on high-performance processors, emphasizing performance and power efficiency over HD transistor density. Notably, 18A incorporates PowerVia, a backside power delivery network, potentially giving it a performance and density edge over TSMC’s N2, which lacks this feature. However, not all 18A chips will necessarily utilize PowerVia.

Regarding power consumption, analysts from TechInsights suggest that chips built on TSMC’s N2 process will likely use less power than similar chips on Samsung’s SF2, given TSMC’s historical advantage in power efficiency. The power efficiency of Intel’s 18A remains to be seen, although it is expected to offer benefits in this area.

Lastly, Intel plans to start mass production of its 18A technology in mid-2025, coinciding with the launch of its Core Ultra 3-series ‘Panther Lake’ processors later that year. TSMC, on the other hand, aims to begin volume production of N2 by late 2025, with the first N2 products likely hitting the market by mid-2026 and mass-market items anticipated in Fall 2026. Samsung, meanwhile, has only indicated that its SF2 will enter high-volume manufacturing sometime in 2025, without specifying further.