AMD launches new AI chips, software for Data Center

At its recent Data Center & AI Technology Premiere event in San Francisco, Advanced Micro Devices Inc. (AMD) revealed a range of products aimed at competing with Nvidia Corp. and Intel Corp. However, despite the announcements, AMD’s stock faced some downward pressure.

During her keynote address, Lisa Su, the Chair and Chief Executive of AMD, introduced the Instinct MI300X accelerator. This accelerator is specifically designed to handle generative-AI workloads. Currently, customers are sampling one version of the accelerator, known as the MI300A. The MI300 was initially introduced by Su during her CES 2023 keynote as the world’s first data-center integrated CPU + GPU.

Similar to Nvidia, AMD has also been caught up in the excitement surrounding the mainstream adoption of artificial intelligence technology. This is largely due to the growing popularity of OpenAI’s generative-AI ChatGPT, supported by Microsoft Corp. The increasing use of large-language models and other AI technologies demands significant computing power, which is typically provided by data centers.

MI300Xs for Generative AI

During the event, Lisa Su announced that the MI300X accelerator enables customers to run generative-AI models with an impressive capacity of up to 80 billion parameters. She mentioned that the sampling of the MI300X will begin in the third quarter, with production scaling up in the fourth quarter.

Another notable introduction was the AMD Infinity Architecture Platform, which utilizes eight MI300X accelerators for generative-AI inference and training tasks.

In previous earnings reports, Lisa Su emphasized that AMD’s focus on AI extends beyond the cloud, specifically highlighting the MI300’s capability to significantly reduce the time required for the inference-modeling process from months to weeks.

AMD’s data-center and embedded offerings received a significant boost from their acquisitions of Xilinx for $49 billion (completed in February 2022) and Pensando, a data-center software company, for $1.9 billion (completed in May 2022).

During the keynote, AMD President Victor Peng, formerly the CEO of Xilinx, introduced the ROCm software ecosystem for data-center accelerators. This ecosystem aims to support the Instinct MI300 and foster an open AI software environment.

This move by AMD puts them in direct competition with Nvidia’s AI strategy, which relies not only on its hardware but also on its proprietary software ecosystem. This has led many analysts to view Nvidia as a software company with hardware as its primary business.

Lisa Su also introduced the fourth-generation Epyc data-center CPU, code named Bergamo, intended as a competitor to Nvidia’s Grace data-center CPU. AMD claims that the 128-core-per-socket CPUs excel in energy efficiency and require fewer servers. Su stated that the chip delivers 1.8 times the performance per watt compared to Intel’s fourth-generation Xeon platinum CPU.

Following the keynote, AMD shares experienced a decline of approximately 3%, ending the regular trading session down 3.6% at $124.53. In contrast, Nvidia shares closed the day up 3.9% at $410.22, making Nvidia the first dedicated chip maker to reach a market capitalization of $1 trillion.

Why AMD shares might have fallen, after these announcements!

The unveiling of AMD’s latest products, including the Instinct MI300X accelerator and the AMD Infinity Architecture Platform, has sparked anticipation in the tech industry. With the ability to handle generative-AI workloads and support up to 80 billion parameters, these offerings position AMD to compete head-on with industry giants Nvidia and Intel.

Than, Why AMD Shares fell following the announcements? If you check previous trends, market mostly welcome these kind of announcements so what have happened in this case!

The market always eagerly awaits the availability or launches of these cutting-edge products, but AMD’s shares experienced a decline following the announcements. It might be that despite the impressive technological advancements, investors may have been concerned about the intense competition in the AI market and the potential challenges in capturing market share from established players like Nvidia. Additionally, market dynamics and short-term fluctuations can influence stock prices, leading to temporary declines.

However, it is important to not forget that after these announcements and already existing line up will bring the long-term potential for AMD. The company’s past strategic acquisitions, such as Xilinx and Pensando, have empowered its data-center and embedded offerings, which enhances its competitive position. Furthermore, the introduction of the fourth-generation Epyc data-center CPU, known as Bergamo, demonstrates AMD’s commitment to delivering high-performance and energy-efficient solutions. Which eventually helps in the long run to the company’s success.

As the demand for AI technologies continues to grow and AMD solidifies its position as a key player in the market, the company has the potential to benefit from its innovative product lineup. Investors and industry watchers will closely monitor AMD’s progress as it navigates the competitive landscape and capitalizes on the opportunities presented by the expanding AI market.