Report: Intel Will Build DG2 at TSMC on ‘Enhanced’ 7nm Node


This website may receive affiliate commissions from the back links on this page. Conditions of use.


One of the most significant questions struggling with Intel as we transfer into 2021 is the degree to which the firm will count on 3rd-celebration foundries for its primary-edge merchandise. Above the previous calendar year or so, Intel has announced that it would faucet foundry companions for a wider range of products and solutions, but still left open up which businesses experienced gained which company.

In accordance to Reuters, Intel intends to manufacture DG2 — its future purchaser graphics card — at TSMC, on an increased 7nm node that hasn’t even been named nonetheless. TSMC has offered 7nm in a few flavors — N7, N7P, and N7 . N7P was the unique N7 node with further performance enhancements, whilst N7 released EUV lithography. The introduction of EUV was a main phase all on its possess. TSMC may perhaps have created a new 7nm node for its personal purposes the firm at times introduces new variants of experienced nodes. 7nm is nonetheless near more than enough to the main edge to plausibly be refined in some fashion.

Alternately, Intel could have paid TSMC to implement a distinct version of the node that fits its personal objectives. Reuters claims the node will be a lot more superior than Samsung’s 8N, which Nvidia tapped for Ampere.

Intel CEO Bob Swan gave interviews ahead of CES 2021, in which he appears to be charting a middle ground amongst Intel yanking all of its current 3rd-get together semiconductor production back into its very own fabs and heading purely fabless. In accordance to Swan, Intel’s aim is to preserve utmost adaptability in its methods:

[W]e may well outsource a lot more it signifies we may perhaps use more obtainable 3rd-bash IP, it indicates we may perhaps make stuff for many others, not just, i.e., be a foundry ourselves. And is there a circumstance where by we could be making use of somebody else’s procedure technological know-how in our fabs? That is attainable. The vital is, as the field evolves, how do we leverage the innovation? Not just within just our 4 partitions, but the innovation going on in the business as a complete, and be quite flexible and adaptable to consider benefit of all those [innovations] together the way.

Intel’s DG1 GPU, presented in the Xe Max products line.

These responses are not also distinctive from what Swan has reported in the previous, and they indicate that Intel will announce a blended system in which it retains its possess foundries, but clarifies which merchandise will be created at other firms on January 21. It does not really issue if Intel builds DG2 at TSMC, just like it does not actually issue that Mobileye carries on to use TSMC for its personal merchandise. When people today feel about Intel, they really do not feel about GPUs or automotive computing. What is genuinely heading to drive headlines and opinions about no matter what method Intel announces on January 21 is not the place the enterprise builds its GPUs, IoT, automotive, or storage solutions. The perceived influence will flip on the place future CPUs will be developed and what foundry prospects (if any) win which products.

In accordance to 2020 facts from IC Insights, TSMC was the second-biggest foundry, with the equivalent of 2.five million 200mm-equivalent wafer begins per month. Intel was calculated at 817K wafer starts off for every thirty day period. Even though this leaves TSMC of course dwarfing Intel, not all of TSMC’s foundries are even able of creating the sort of components Intel requirements. Any attempt by Intel to change its organization to TSMC wholesale would also require a spectacular ability growth on TSMC’s part.

Glance for much more specifics on Intel’s upcoming chip strategy to arrive on January 21.

Now Study:

This website may generate affiliate commissions from the one-way links on this web page. Conditions of use.

Leave a comment

Your email address will not be published.