According to media reports, the new AI chip of DeepX, a South Korean artificial intelligence chip startup, has now changed its production cooperation from Samsung Electronics to TSMC. Previously, the startup had been working with Samsung.
As a start-up company focusing on the research and development of AI chips, DeepX has always taken innovation and technological breakthroughs as the core driving force for the company's development. Previously, DeepX has won wide recognition in the market for its unique NPU (Neural Processing Unit) technology and high-performance AI chips. DeepX's abandonment of Samsung and the use of TSMC's 3nm process for OEM production is a high recognition of TSMC's 3nm process technology.
As the world's leading semiconductor manufacturing company, TSMC's 3nm wafer process technology represents the highest level in the current semiconductor industry. This technology can provide smaller transistor size, higher integration, and lower power consumption, which provides the possibility of improving the performance and optimizing the energy efficiency of AI chips. DeepX chose to cooperate with TSMC because of its leading edge in process technology and stable manufacturing capabilities.
However, DeepX plans to entrust TSMC with the production of chips while maintaining its relationship with Samsung Electronics. DeepX's current portfolio also includes the M1 chip for automotive, the H1 chip for the AI accelerator developed for data centers, and the V-series NPU SoC for connected cameras and drones. The M1 and H1 chips are currently produced on Samsung's foundry's 5nm process, while the V-series is manufactured on a 28nm process, and it is expected that both will be produced in small quantities before mass production.
Pictured: DeepX in partnership with TSMC
DeepX's latest AI chip will integrate DeepX's latest NPU technology, with higher computing power and lower power consumption, which can meet the needs of various complex AI application scenarios. This NPU SoC is an edge device semiconductor that enables AI processing workloads to be completed at the endpoint, reducing reliance on cloud services and saving costs. By working closely with TSMC, DeepX will be able to ensure high quality and efficiency in the design and manufacturing process of the chips.
For DeepX, this cooperation with TSMC not only represents a technological breakthrough, but also represents an important shift in market strategy. By partnering with the world's leading semiconductor manufacturing companies, DeepX will be able to better grasp market needs and technology trends, accelerating product development and time to market. At the same time, this will also win more advantages and opportunities for DeepX in the future market competition.
It is reported that DeepX plans to produce engineering samples at the TSMC factory next year, and maintain the relationship with the Samsung foundry to continue to develop its product line. DeepX's portfolio includes the M1 chip for vision, the H1 AI accelerator for data centers, and the V-series NPU SoC for IP cameras and drones.
