Could Remastering Silicon Be the Answer to the Global Chip Shortage?
October 03, 2022
There hasn’t been another time in recent memory where semiconductors have become as critical to fueling the electronics industry’s economic framework. This has become abundantly clear with the global chip shortage, which continues to distress industry sectors from automotive to consumer electronics.
In addition to holding back global economic growth and making life difficult for consumers and businesses worldwide, the shortfall in manufacturing capacity is uneven, affecting legacy process nodes far more than mid-performance nodes.
While semiconductor experts have been hard at work on scoping solutions, the situation has looked insoluble — simply put, semiconductors are extremely hard to design and manufacture; supply chain effects are very difficult to absorb due to this lack of flexibility.
Enter silicon remastering, a new AI-driven design framework with the potential to transform the global chip supply chain. To understand how, we must acknowledge the root of the problem: an imbalance in manufacturing capacity. Process nodes built on legacy silicon technologies are in extremely short supply. With them running out, using past technologies to replenish them is no longer a viable option.
At the same time, the market is overflowing with mid-performance nodes that are between 12 and 16 nanometers (nm). This means that each year, up to 25 million wafers, or 10% of global capacity, sit unused. The chip supply crisis makes a strong economic case for repurposing and retargeting older chips toward this available capacity — or in other words, moving chip designs between nodes.
So why not move these older designs to different process nodes? Unlike other products, these semiconductor products were designed specifically for legacy silicon technologies. While redesigning them is a prohibitively complex, costly, and labor-intensive process for a human chip designer (talent shortage aside), artificial intelligence (AI) offers a way to absorb both this process and the market’s excess capacity.
Old Is New: The Concept of Silicon Remastering
As the chip shortage persists, we cannot get away from the fact that in the very near future, more industries are going to use silicon technology, and ones that already use it, such as automotive, are going to need more of it. This means that companies will have to start designing chips that factor in the need for future remapping — or remastering — to a newer technology.
To help the world absorb excess capacity, the overall semiconductor design process needs to become more flexible. At Synopsys, we use AI algorithms to help re-optimize existing chips for different nodes — in a matter of weeks as opposed to months.
It all started with curiosity. Can AI learn how to optimize a design for a certain silicon technology and then leverage this learning to re-optimize the design to another technology while maintaining the level of quality? A round of experimentation showed us that it can. We used AI to re-optimize a 40nm design to 10nm, with excellent results. We call this silicon remastering.
And it goes beyond re-optimizing an existing design to a new node: remastering can also work with derivatives to an existing design whether it is for the same or different technology according to customer needs, such as updating an IP version, adding or removing features, or trying to achieve a different performance vs. power trade-off.
Moving to a new node opens up capacity and optimization potential. It prolongs the life of an original design that is likely to have been the fruit of much hard work, with the added benefit of learning from it directly. Ultimately, this will inform the goal of designing new chips that can be readily remastered.
How AI Can Reshape Old Chips into New Products
Silicon remastering is a prime example of AI delivering a clear business solution. Today, AI-driven solutions for chip design have been proven to achieve power, performance, and area (PPA) targets faster and with significantly less engineering effort across the design team. AI can save companies many millions of dollars that would be otherwise spent on a less efficient process or lost due to the disruption of the chip shortage.
The impossibility horizon — sustained by the slowing of Moore’s law — will inevitably be influenced by rapid advancements in materials, devices, software, and architecture. This “SysMoore” era that we find ourselves in offers unprecedented opportunity for innovation. The emergence of autonomous design instruments — super-tools fusing together hundreds of algorithms precision-guided by AI —offer new opportunities for circuit designers and a new wave of architectural vitality.
The chip design process uses automation tools to simulate and verify a design before it heads to the fab, rendering the chip in digital format. Assuming the design becomes reality, AI can assimilate the original and re-engineer it for a new fabrication process. When chips are designed with remastering in mind, AI will also be able to deliver spin-off designs as well as complete remakes.
Why Silicon Remastering Is Critical for Building Tomorrow’s Capacity
Remastering represents a near- and long-term solution because it creates both physical capacity and a deeper understanding of the design. Moving a design from an old node to a new one frees up capacity on that old node. The process teaches us about the design, tools, and similarities between technologies.
As industry conversations note: why spend a fortune on building new fabs for old technology when you can spend a lot less by employing AI to do the heavy lifting of migrating old (but good) designs to a modern equivalent?
Given the growing appetite for chips across various sectors, including but by no means limited to automotive, the current shortage is not a temporary issue but an opportunity to rethink future semiconductor roadmaps with flexibility, optionality, and personalization.
Solving this will require a collaborative effort from the industry with big implications. In the meantime, remastering is already going a long way to rebalancing the world’s skewed manufacturing capacity and is expected to continue to do so for years to come.