NVIDIA's AI "new ace" is here! Vera Rubin has been delivered to giants like OpenAI, and the next round of the computing power competition has begun

According to the Zhitong Finance APP, NVIDIA (NVDA.US) has announced that computing systems based on the latest Vera Rubin architecture have been delivered to several large artificial intelligence (AI) companies and are about to be put into actual operation. The company emphasized that the new generation of products is progressing as planned and is expected to further consolidate its leading position in the AI chip market.

Ian Buck, NVIDIA's Vice President and General Manager, stated at a media briefing held at the company's headquarters: "We have fully entered the mass production phase, and all major customers are deploying the relevant systems."

The launch progress of the Vera Rubin products has been closely monitored by investors and analysts, as there were previous market concerns about potential delays or production bottlenecks. Although NVIDIA CEO Jensen Huang had previously dismissed these concerns, the market remains vigilant about whether NVIDIA can maintain its technological advantage as competitors accelerate their efforts.

Data shows that this year, the benchmark index for the chip industry has risen by about 66%, but NVIDIA's stock price has only increased by about 9%. In contrast, the stock prices of Intel (INTC.US), Arm Holdings (ARM.US), and AMD (AMD.US) have all more than doubled.

Currently, NVIDIA remains the dominant supplier in the AI accelerator market. AI accelerators are primarily used to run AI software in data centers. However, AMD and Broadcom (AVGO.US) are actively competing for more market share, and large data center customers like Amazon (AMZN.US) are also developing their own chips to reduce reliance on NVIDIA products.

Against this backdrop, NVIDIA faces pressure to prove that its new products can be delivered on time and outperform competitors. The company has held several briefings at its headquarters in Santa Clara, California, emphasizing that the new generation of devices not only offers higher performance but also significantly reduces deployment difficulty.

NVIDIA stated that ChatGPT developer OpenAI plans to "massively" adopt the Vera Rubin system in the third quarter. Companies such as CoreWeave (CRWV.US), Google Cloud under Alphabet (GOOGL.US, GOOG.US), Azure under Microsoft (MSFT.US), Meta Platforms (META.US), and Dell Technologies (DELL.US) have also begun using the relevant systems.

Despite investors being relatively cautious about NVIDIA's stock performance this year, the market expects its revenue growth rate to accelerate further. Analysts predict that NVIDIA's revenue for the current fiscal year will grow by 82% year-on-year, reaching $393.4 billion, higher than the previous estimate of about 65%; total revenue for the next fiscal year is expected to exceed $500 billion.

NVIDIA is expected to contribute about one-third of global chip industry sales while maintaining a high level of profitability. The market expects the company's gross margin for the current fiscal year to reach about 75%.

NVIDIA executives stated that the company is launching a series of new technologies to improve product performance and deployment efficiency by enhancing hardware design, manufacturing processes, and cooling systems.

In previous generations of products, computing trays accommodating AI accelerators, central processing units, and network components typically required manual assembly, including connecting numerous cables. This process is time-consuming, especially when factory workers are just starting to familiarize themselves with the new design, leading to a relatively high failure rate The new generation of systems has basically eliminated internal cabling, primarily achieving component interconnection through circuit boards or specially designed connectors. NVIDIA stated that tasks that previously took hours for humans to complete can now be done by robots in minutes, significantly reducing the failure rate.

This improvement is partly due to the comprehensive adoption of liquid cooling technology for major components. The liquid cooling system reduces the need for fans and air circulation space, thereby increasing the component density within the cabinet.

NVIDIA announced that the token processing capability of the new generation AI accelerator cabinet NVL72 will reach ten times that of the previous generation products. Tokens are the basic units for measuring AI computing tasks, and the relevant performance data was measured by the data center operator CoreWeave.

NVIDIA also unusually compared the performance of the new product directly with competitors. The company claims that when running Python programming language code, the new Vera processor can achieve speeds up to 1.8 times that of the AMD Turin processor. Both OpenAI and Anthropic are early customers of this product.

In addition, NVIDIA has established several testing sites near its headquarters in Silicon Valley for the rapid development and validation of the hardware, software, and supporting technologies required for AI data centers. Customers can test software and services on new hardware in these facilities before the equipment is officially delivered.

At a small testing data center located in Sunnyvale, California, NVIDIA engineers showcased a set of Vera Rubin servers being tested by OpenAI. NVIDIA is sending a clear signal that the new generation of AI computing devices is ready to handle the world's most critical artificial intelligence workloads