Could AI Help Solve The Global Chip Shortage?

In half, sure. We may faucet into unused manufacturing capability utilizing AI, and Synopsys believes it has the AI that might do it.
Today’s chip scarcity is holding again world financial development, fueling inflation, and making life tough for shoppers and companies alike. Everything from ovens to Porsches are affected. The causes are advanced, however one of many points is that the shortfall in manufacturing capability is uneven, affecting legacy course of nodes way over mid-performance nodes reminiscent of 12 and 16nm, the place there’s a surplus. Yes, regardless that Apple, Intel, Qualcomm, and NVIDIA push the envelope to provide chips in essentially the most superior course of nodes, the US Department of Commerce says 8-10% (20-25 million wafers) of complete world fab capability goes unused yearly. Until now, investing in retargeting older chips to obtainable capability simply hasn’t made financial sense. Things could also be altering…

Re-balancing Capacity and Demand
As Intel CEO Pat Gelsinger lately famous, “The lack of ability to transition nodes effectively and shortly signifies that capability stays unused. Since constructing new fabs on previous nodes doesn’t make techonomic sense, to mitigate the provision concern, the trail ahead is transferring designs between nodes”.

As Intel seeks new prospects for its foundry companies, Synopsys AI instruments might help remaster designs … [+] to absorb extra capability.


As Intel seeks new prospects for its foundry companies, Synopsys AI instruments might help remaster designs to absorb extra capability. 
However, transferring designs from node to node, or “retargeting,” is impractical. Scarce engineering sources are busy designing new silicon, which drives future income and earnings, so retargeting tasks are a decrease precedence. To make issues worse, retargeting, usually, is nearly as advanced as designing a chip, to start with. But now, AI could provide a path ahead, enabling corporations to refresh silicon shortly and affordably, producing new, sooner chips, decrease energy, and cheaper than what’s within the area in the present day. Synopsys has years of expertise with utilizing AI in chip design, and is now taking the following step.

Silicon Remastering
Synopsys refers to this as “Silicon Remastering”: using AI algorithms to routinely redesign a chip for a unique node, reaching months or years of labor in simply weeks whereas doubtlessly saving prospects a whole lot of thousands and thousands of {dollars}. Company co-founder and co-CEO Aart de Geuss mentioned this functionality in his keynote handle on the annual ISSCC convention, explaining that this method may mitigate provide points in comparatively quick order by bringing 25 million new wafers on-line or greater than a yr’s value of recent capability. “Within 5 years, we are going to see a change of the worldwide chip provide chain that may higher facilitate using capability, and we imagine silicon remastering will probably be crucial expertise,” Dr. de Geus mentioned.

Remastering silicon echoes the method and choices obtainable to, say, remastering previous audio. Engineers can take the older recording and create newer variations benefiting from new audio applied sciences to enhance sound high quality. Or they’ll add tracks, reminiscent of background strings, to make a brand new product at a fraction of the price of re-recording the piece. The Cirque de Solei “Love” soundtrack is a good instance. Well, one can do the identical with silicon; simply transferring to a brand new course of node can provide entry to new capability and provide additional optimizations.

From real-world outcomes, within the picture beneath, remastering produced a greater resolution with extra efficiency, smaller die space, and decrease energy. These outcomes have been achieved in only a few weeks, not months, and solely required one engineer as an alternative of a group of designers. This light-touch method ought to entice corporations like Intel to shift manufacturing of older chips to mid-range applied sciences like Intel16 capability.

Synopsys was capable of remaster an older system to a more recent course of node, and improve frequency at … [+] decrease energy utilizing AI.


Silicon Remastering may open up demand for Intel and different foundries with newly reborn chips. The corporations who remaster their legacy elements may save a whole lot of thousands and thousands of {dollars}. And the customers of these elements will get higher efficiency and energy consumption. I believe Pat Gelsinger mentioned all of it when he commented that “I’ll make them as many Intel 16 [nanometer] chips as they need”.
Disclosures: This article expresses the opinions of the writer, and isn’t to be taken as recommendation to buy from nor put money into the businesses talked about. My agency, Cambrian-AI Research, is lucky to have many, if not most, semiconductor companies as our shoppers, together with NVIDIA, Intel, IBM, Qualcomm, Esperanto, Graphcore, Synopsys, Cerebras Systems and Tenstorrent . We haven’t any funding positions in any of the businesses talked about on this article. For extra data, please go to our web site at

Recommended For You