Multi-die systems define the future of semiconductors
However, some business leaders are keeping pace with the latest developments in the sector. Multi-die technology is still a puzzle to many executives. A recent survey by MIT Technology Review Insights asked business leaders about their awareness of this design strategy—and found that 62% of respondents were either indifferent, unaware, or only somewhat aware of this technology’s capabilities. know.
Some chip-dependent industries clearly need to keep a close eye on advances in semiconductor technology: automotive companies, artificial intelligence firms, hyperscale data processing organizations and smart device makers, to name a few. But because advanced semiconductors are fundamental to modern-day business operations, even executives whose jobs don’t directly touch the technology should be mindful of chip design trends—including those that Will define the next chapter of
why semiconductors matter
While the global semiconductor shortage that began in 2020 had proximate causes in natural disasters and geopolitics, its effects drew widespread attention to the fact that nearly every industry depends on chips. And pandemic-related ripple effects aside, the silicon status quo has been in flux for some time. New technologies such as artificial intelligence and machine learning (AI/ML), which require greater computing efficiency and performance, have put traditional systems under strain in recent years.
With the rise of the Internet of Things (IoT), consumers have also come to expect intelligence in everything from refrigerators to lightbulbs. Innovators are responding accordingly. Our survey found that nearly one-third (31%) of business executives plan to improve their companies’ existing smart products, and nearly one-third (29%) will soon add AI/ML capabilities to their products. intend to. Only 9% of respondents said they are not producing IoT or connected devices.
However, this type of technology requires strong edge computing and on-device processing, which in turn requires more and more efficient hardware performance. To complicate matters, the cloud data centers powering this compute shift are also energy consumers. This is another area where traditional silicon lags: stability. The cost of producing more silicon than is needed is not only bad for business—it also has an environmental impact. And while there is an ongoing push toward net-zero carbon emissions within the semiconductor supply chain, the industry is not yet on track to meet the emissions standards set out in the UN’s 2016 Paris Agreement.
An industry shift towards multi-die design may be part of the solution to these challenges. Instead of a single monolithic chip (“system on a chip”), a multi-die design consists of a collection of chips (chiplets or dies) connected in a sophisticated package (“system of chips”), which may include stacking blocks in 3D. . Configuration for higher density. Multi-die system designs are capable of supporting the rollout of AI/ML on a large scale, and they can improve silicon yields while reducing waste during chip manufacturing.
When it comes to the commercial use cases of multi-die systems, Patrick Moorhead, founder, CEO and principal analyst at the global technology consulting firm Moor Insights & Strategy, notes that these custom designs may soon be the first for companies looking to stand out. There can be a significant difference. out among competitors. “As more people are looking at more custom silicone as a way to differentiate what they bring to the table, that is what businesses should be looking at,” he says. “Chiplets enable smaller companies with smaller pocketbooks to use semiconductors for unique competitive advantage.”
Gerry Talbot, Corporate Fellow at semiconductor company AMD, breaks down the business value of chips for a wide range of use cases for the technology. “I do not think so [business leaders] would be so excited about the technology,” he says, “as much as the application and the enablement of a unique user experience that could help sell their product.”
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by the editorial staff of MIT Technology Review.