Although the new product has not yet been released, as the market is excited for the AI event, the stock price of NVIDIA, the "universe leader", has risen to nearly US$900, with a market value of more than 2.2 trillion. It seems to be only a matter of time before it replaces Apple and becomes the second largest company in the world by market value.

The new NVIDIA GTC conference will be held from March 18th to 21st. AI expert Karl Freund predicts that investors in this GTC are expected to see the new generation of flagship GPU B100, a new inference platform, details of automotive chips in cooperation with MediaTek, and progress in the software business.

New hardware products are expected to become the protagonist

NVIDIA's new generation flagship AI processor Blackwell GB100 is expected to be officially unveiled at this GTC conference. As the successor of the Hopper series, the B100 is regarded as a powerful tool for large-scale language model (LLM) training and inference.

It is reported that the B100 will double the performance of the previous generation product, provide 25% greater high-bandwidth storage capacity, and be equipped with a specialized accelerator module. It is expected to become the main revenue driver of Nvidia in 2025. In addition, there are rumors that more flagship products will be launched in the H100 and H200 series.

In the field of AI inference, the Lovelace L40S chip may not be able to meet the ultra-large memory required for large-scale LLM inference. Analysts expect Nvidia to launch a more affordable new inference platform based on GDDR7 memory to make up for the shortcomings of its existing product portfolio.

Automobiles and edge AI may become new growth points

In addition to the data center market, Nvidia will further develop its efforts in emerging areas such as automobiles and edge AI. In 2022, Nvidia raised its automotive AI order guidance to US$11 billion for six years. However, given the sluggish performance of the electric vehicle market last year, especially the poor progress in electrification and intelligence of large car companies including Mercedes-Benz, whether the guidance will be adjusted is one of the focus of the industry.

In addition, Nvidia may also announce details of cooperation with mobile chip giant MediaTek in the automotive field. The two parties plan to jointly develop chips, integrate Nvidia GPUs and MediaTek SoCs, and open up a new track for automotive artificial intelligence.

Software ecosystem is indispensable

In addition to hardware, the software ecosystem is an important link for NVIDIA to further explore business opportunities in the AI industry. Last quarter, Nvidia disclosed for the first time that its software business scale had reached US$1 billion, covering enterprise AI tools, Omniverse digital twin platform, etc.

In the future, NVIDIA will work to further monetize its software products. The current perpetual license price of NVIDIA Omniverse Enterprise is US$3,595 per CPU socket, while Enterprise Business Standard support is priced at US$899 per license per year, with a single user paying up to US$5,000.