Samsung revealed its first-ever AI-designed semiconductor on August 15, 2021. Samsung is automating the incredibly complicated and delicate method of developing cutting-edge microchips with AI technology.
The South Korean behemoth was one of the first device manufacturers to employ artificial intelligence in the development of its products. In the latest software from Synopsys, a prominent semiconductor autodesk provider used by many industries, Samsung is including AI elements.
Synopsys president and co-CEO Aart de Geus adds, “What you’re witnessing here is the first of a true commercial processor architecture incorporating AI.” Companies, such as Google, have discussed using AI to build circuits. Since Synopsys collaborates with numerous firms, its tool, DSO.ai, may emerge to be the most far-reaching.
As per industry observers, the tool has the potential to greatly improve chip manufacturing and uncover innovative chip designs. Synopsys also has years of cutting-edge chip ideas that can be used to develop an AI algorithm, which makes it a significant asset for creating AI-designed devices.
Samsung admits that Synopsys AI software is used to build its Exynos processors, which are utilized in smartphones, including Samsung’s own labeled devices, as well as other gadgets. Samsung introduced its latest smartphone, the Galaxy Z Fold3, a foldable gadget, earlier this week.
The business hasn’t said whether or not the AI-designed chips are in manufacturing, or what devices they may be used in. AI looks to be altering the way chips are produced across the industry.
In a study released in June, Google explained how it used AI to organize the features on the Tensor chips it employs in its data centers to train and operate AI systems. The Pixel 6, Google’s next smartphone, will include a bespoke processor made by Samsung.
A Google spokesman declined to comment on whether AI was used in the development of the mobile processor. Nvidia and IBM are among the chipmakers experimenting with AI-driven chip design.
Other semiconductor-design software companies, such as Cadence, a competitor of Synopsys, are working on AI capabilities to help map out the blueprints for a new chip.
Artificial intelligence is ideally adapted to organizing billions of transistors across a chip, according to Mike Demler, a senior analyst at the Linley Group who studies semiconductor-design software. “It lends itself to these tremendously complicated problems,” he explains. “It will just become part of the computing toolkit.”
According to Demler, using AI is costly since it takes a lot of cloud computing resources to train a sophisticated algorithm. However, as computer costs fall and models grow more efficient, he expects it to become more affordable. Many jobs in chip design, he says, cannot be automated, therefore experienced designers are still required.
Semiconductors, as well as the equipment required to manufacture them, have become more valuable assets. The US government has attempted to limit the flow of chipmaking technology to China, a crucial competitor, and some lawmakers have proposed adding software to the export restrictions list.
The possibility of utilizing AI to tailor the software to operate more effectively on a semiconductor also exists in the approaching age of AI-designed processors. This might incorporate neural network methods, which are frequently employed in current AI and operate on specialist AI processors.