When we talk about the supply chain for electronics, we often look at it from a production, distribution, and manufacturing perspective. As a result, we focus on the manufacturing end of things, and the discussion often boils down to a binary decision between too much global production capacity, or too little capacity resulting in shortages. While it is true that certain chips have seen, and will continue to see, demand that outpaces current fabrication capacity, other advanced components face a different type of shortage.
The semiconductor industry in North America and Europe is reaching a crisis point not only due to lacking fabrication capacity, but also due to a designer talent shortage. And now the industry is pushing forward with a new design approach that Synopsys calls silicon remastering. The approach represents an important advancement on an expensive design tactic, and it could help companies stay competitive in the face of a long term challenge surrounding access to design talent.
The Chip Supply Chain is About Talent AND Production
Every time we have a chip or component supply shortage, the problem is typically short-lived. Going back to the 2000s, chip supply shortages related to excessive demand and the need for new production capacity can last anywhere from 1-3 years. Coincidentally, this is the amount of time required to bring a new manufacturing facility online, or re-tool an existing facility to accommodate demand for different products that might be at a more advanced node.
That’s the production side of the semiconductor supply chain. Because the time involved in bringing a facility to the point where it can address market demand is relatively long, designs are typically ready to place into production immediately. What if the tables were turned, and there was a shortage of ready-to-produce designs with waiting and available fabrication capacity? This is a longer term challenge the industry will grapple with over the coming years.
According to Deloitte, the US and European sectors of the semiconductor industry require a major infusion of talent in order to meet product release timelines and projected demand for new products. The Semiconductor Industry Association (SIA) has found that enrollment in computer science and electrical engineering at US institutions is down from its all-time high in 2016. The talent shortage is already bad enough, and it’s only expected to get worse. Other industry associations and media outlets worldwide (e.g., in Japan, Korea, and China) come to similar conclusions.
Current and projected semiconductor industry worker shortage by country
AI Enters the Design Side of the Supply Chain
The biggest driver of design decisions in the semiconductor space is design for manufacturing (DFM). PCB designers and component designers have to continue grappling with this, sometimes as a totally new concept. But for semiconductor engineers, DFM forces every chip to be designed specifically for compatibility with its fabrication process.
The result is that tooling cannot be used across different products, and that tooling is expensive and time-consuming to create. Similarly, optimizing a design for multiple fabrication nodes is equally difficult, but it is one way to overcome the supply chain challenge that continues to persist around products at lower technology nodes. Some of these products with long lead times as of Q3-Q4 2023 include:
- Specialty ASICs, typically with analog interfaces
- PMICs, including automotive PMICs
- Smaller microcontrollers
- Miscellaneous digital components (line drivers, interface converters, etc.)
Many of these products were originally built to be fabricated in older technology nodes. If they can be re-optimized for a newer technology node, they could be put into production in a newer process for which there is available capacity. However, the existing talent shortage creates a resource bottleneck: there simply aren’t enough people to do the work manually and reach PPA optimization within a reasonable time frame.
Given the huge amount of design data any semiconductor company has, the design factors that lead to the greatest PPA optimization results are difficult for humans to sort through. AI has become the design engineer’s data aggregation and pattern recognition tool, helping sort through the infinite possible design permutations and land on the best possible physical design. The newest AI-based tools are using reinforcement learning to provide continuous improvement in chip optimization based on real observed results from manufactured products.
This means designers can focus on other engineering tasks for which AI is not ideally suited, such as:
- Feature and interface selection
- Locating and debugging design errors
- Improving architecture to provide greater efficiency
- Differentiating products for target markets
When you’re choosing functionality for your new product and you need to optimize your design for maximum performance, use the complete set of system analysis tools from Cadence. Only Cadence offers a comprehensive set of circuit, IC, and PCB design tools for any application and any level of complexity.