Senior executives at leading electronic design automation (EDA) manufacturer Cadence said the United States should invest more in cutting-edge semiconductor packaging technology to ensure the United States remains at the forefront of artificial intelligence (AI).Aye) and other emerging technologies may maintain their leading position.
Bloomberg News reported that Anirudh Devgan, CEO of Yihua Computer, said in an interview in Taipei: “Manufacturing is extremely important, (whether it’s 3 nanometers or 2 nanometers), but we also need to invest in the future. That’s the trend of the future.” ) 3D.Wafer, “Yihua is one of the three leading companies in semiconductor design software. Its customers includeHuida(Nvidia) and other AI hardware companies.
Advanced chip packaging technology helped Huida develop an industry-leading AI accelerator, but limited suppliers hindered Huida from production. As computing demands continue to grow and transistors shrink, the manufacturing costs of advanced chips are becoming higher and higher. Assembling chips with 3D architecture and solving physical limitations has become a major priority.
As transistor shrinkage approaches physical limits, one way semiconductor designers can improve chip performance is to stack so-called “chiplets” to create multifunctional 3D chips. This approach helps create powerful chips that can handle generative AI workloads at a reasonable cost. Dagen said it should be a priority for the United States to lead in this regard.
He added: “You don’t even need your own fab to do the packaging, because sometimes you can collect chips from different fabs. For the US government, it’s great to have this kind of technology.”
Huida is an excellent example. The company’s technology is clearly ahead of the industry, but TSMC’s packaging capacity is limited, which limits the supply of chips. TSMC Chairman Liu Deyin reiterated last week that the shortage of AI chips will last for 18 months.