Kim Nok-won, CEO of Deep

DeepX 1st generation AI semiconductor ‘DX-M1’

‘Butter Benchmark’ experiment comparing DeepX AI Semiconductor DX-M1 (left) with overseas competitors’ products. Heat control performance is evaluated by placing butter on top of each product and observing how quickly the butter melts over time. This is an experiment designed by Kim Nok-won, CEO of DeepX, to easily explain the superiority of the product to customers. [영상출처=딥엑스]

“At the end of December this year, we will release the first mass production wafer of Deep We are developing a mass-produced product and are in discussions with over 140 global companies, including HP.”

Kim Nok-won, CEO of DeepX, a domestic fabless company specializing in semiconductor design, recently met with reporters at the ‘Semiconductor Exhibition (SEDEX 2024)’ held at COEX in Seoul and introduced the AI ​​semiconductor ‘DX-M1’ that the company is developing.

The reason CEO Kim emphasized wafers is because they are an important indicator for checking the stability and efficiency of products prior to mass production. He emphasized, “We received a prototype from Samsung Foundry in June of this year, and it sufficiently demonstrated mass production in terms of performance and power consumption,” and added, “We secured both yield and quality in the 5nm (nanometer, one billionth of a meter) process.” .

On the 23rd, at the DeepX booth at the ‘Semiconductor Expo (SEDEX 2024)’ in COEX, Gangnam-gu, Seoul, DeepX CEO Kim Nok-won is explaining the ‘Butter Benchmark’, a heat control and low-power performance test of the DX-M1. This is an experiment designed by CEO Kim himself to effectively convey low-power performance to customer executives. Photo = Reporter Choi Seo-yoon

DX-M1 is a semiconductor for edge computing that can process AI tasks quickly and efficiently. It can be used in home appliances such as electric rice cookers, all systems equipped with cameras, and even data center servers. “The number of robots will increase explosively in the next 10 years,” he said. “The possible applications are endless.”

For example, in an unmanned payment system, consumers must directly select agricultural products such as apples or green peppers and enter the price. By applying DeepX’s AI technology, the system can automatically recognize products and assign accurate price tags, making 100% unmanned operation possible. CEO Kim said, “If we introduce this system, we can build a settlement system with only AI semiconductors costing about $100, which can reduce labor costs and increase efficiency at the same time.”

At this event, DeepX conducted a melting experiment by placing butter on a chip (semiconductor). Since low power and heat suppression are essential for AI semiconductors applied to home appliances, this is a comparison with competitors’ products. In the ‘Butter Benchmark’ experiment, the temperature of a competitor’s product rose to 60 degrees in 10 minutes and the butter melted, but the DX-M1 maintained its shape for more than an hour. CEO Kim explained the low-power performance of the DX-M1 and said, “Butter melts at 30 to 36 degrees, but the DX-M1 suppresses heat generation.” He added, “In electronic devices or robots, heat generation can shorten battery life and reduce device performance,” and “DX-M1 is a product that solves these problems.”

Kim Nok-won, CEO of DeepX. Photo = Reporter Choi Seo-yoon

CEO Kim said that DeepX’s low-power design differentiates it from NVIDIA, a powerhouse in AI semiconductors. He said, “In the AI ​​semiconductor market, there is no absolute leader in the field that Deep “We are filling the gap,” he said.

CEO Kim was confident that DeepX is maintaining the technological gap with global competitors through more than 300 patents. He emphasized, “In the NPU field, we have more patents than NVIDIA, Qualcomm, and Intel,” and “Within 20 to 30 years, we will grow DeepX into NVIDIA in the on-device AI semiconductor field.”

Reporter Choi Seo-yoon sychoi@asiae.co.kr