Samsung’s use of AI for chip design signals industry shift away from dominant suppliers

The news: Samsung is using artificial intelligence (AI) to automate the highly complex process of designing cutting edge computer chips, per Ars Technica. The Korean tech company, which is the world’s top chipmaker by revenues, hopes to accelerate R&D of future processors, and competitors like Google, IBM, and NVIDIA are following suit.

  • Using an AI approach in new software from Synopsys, a leading chip software design company, Samsung is creating complex chips, like the Exynos processor, which will run its smartphones, tablets, and PCs.
  • The protracted global chip shortage has disrupted various industries by causing delays to product launches, limiting supply, and raising prices for manufacturers, which has in turn been passed on to consumers.
  • Furthermore, the shortage has exposed the dangers of relying on a few dominant chip suppliers, and pushed manufacturers to consider making their own chips.

Here’s how it works: Samsung’s use of Synopsys’ AI employs a machine-learning technique called ‘reinforcement learning’ to work out chip designs. Reinforcement learning involves training an algorithm to perform a task through reward or punishment; it has yielded apparently good results as an effective way of capturing subtle and hard-to-codify human judgment.

While this methodology has been pioneered with Samsung, it has the potential to become an industry standard for manufacturers like NVIDIA and IBM, which are looking to ramp up bespoke chip design.

  • AI has the potential to change how chips are made. Google released a paper outlining how it used AI to develop its Tensor chip, which is coming to their Pixel line of smartphones, and has replaced traditional Qualcomm processors.
  • Synopsys has its own advantage too: years of cutting-edge semiconductor designs that can be used as the foundational data set to train an AI algorithm.

The opportunity: If companies like Samsung, Google, NVIDIA, and IBM can successfully harness AI to develop future chips, they could inspire other companies to do the same.

  • Developing chips in-house could lead to the creation of novel chips for tailored applications, as well as less reliance on a handful of chip suppliers.

"Behind the Numbers" Podcast