A large number of GPUs rushed in!TSMC 5nm capacity utilization rate is close to full load

Just the latest news, according to the industry sources, a large number of Nvidia orders of the most urgent process (SHR) have poured in, and the production capacity utilization rate of TSMC 5nm process platform is pushed to close to full load.Nvidia emergency orders include H100, A100, H800, A800.Sources added that TSMC's SHR for Nvidia AI GPU will last a year.Previously Nvidia's order has boosted TSMC 7/6nm process utilization rate.TSMC had previously only expected to receive a large number of Apple orders, and did not expect that Nvidia's emergency orders would pour in large quantities.

It is reported that Nvidia A100 uses TSMC 7nm process node, and H100, which started shipping last May last year, is based on TSMC 4N process nodes (optimized version of the 5nm process node for Nvidia).Since its launch, Nvidia has seen huge demand for its H100 and DGX H100 GPU servers.Although Nvidia has been working hard to meet the needs, the recent artificial intelligence boom has a huge impact on the company's efforts to meet the demand.

After all, the latest news showed that some agents revealed that the price of Nvidia A100 has risen since December last year. As of the first half of this year, the cumulative price of 5 months has reached 37.5%; the cumulative increase in the A800 price increased by 20.0%during the same period.At the same time, Nvidia's GPU delivery cycle has also been stretched. The previous delivery cycle was about one month, and now it is basically three months or longer.Even some new orders "may not be delivered until December."

Nvidia announced the company's latest financial report yesterday.The report shows that the net profit of Nvidia's first fiscal quarterly revenue was 2.043 billion US dollars, an increase of 26%compared with 1.618 billion US dollars in the same period last year, and an increase of 44%compared with US $ 1.414 billion in the previous fiscal quarter.

According to the business department, Nvidia's first -quarter data center business revenue was US $ 4.28 billion, setting a new record in the history of the company, an increase of 14%compared with the same period last year, and an increase of 18%compared with the previous fiscal quarter.

*Disclaimer: The news of this article originated from the Electronics Times Finance News Agency. It was organized by Huaqiang Microelectronics. The news was for exchanges and learning without official confirmation.