Innovation has become the focal point for the data center chip market with chipmakers launching new chips. Moreover, there is fierce competition with major tech giants planning for their chips. Advanced Micro Devices (AMD) is the chipmaker that launched its new series of processors that offer improved performance along with consuming less power. Qualcomm also announced its data center AI chip. The competition gets intense as major tech giants aim to develop their chips. Google and Huawei are among giants, looking forward to developing their chips. These activities are taking the data center chip industry to higher levels. According to the research firm Allied Market Research, the global data center chip market is expected to reach $15.64 billion by 2025. Following are the trends of the industry that would shape the future:
Innovative Chips by Major Players
Launching new chips for data centers is one of the prime strategies adopted by market players to stay ahead of the curve and gain a competitive edge. The strategy has been adopted recently by the chipmaker, Advanced Micro Devices (AMD). It has launched the AMD EPYC 7002 series that provides better performance, at the same time, consuming less time. Also, it is equipped with advanced security features, improved bandwidth for memory, and offers 23% more instructions per clock. Dr. Lisa Su, the President, and CEO at AMD, outlined that the firm sets a new standard for modern data centers with the release of 2nd Gen AMD EPYC processors that provide the high-quality performance along with less total cost of ownership for various set of workloads. The adoption of its new processors is increasing as customers prefer EPYC to meet their server computing needs. Google and Twitter are among its customers. Google has already implemented 2nd Gen AMD EPYC processors in data centers. Bart Sano, Google Vice President of Engineering, highlighted that these processors help the company in innovating. The scalability in computing, memory, and input/output performance would enhance the ability of Google to keep innovating for its infrastructure and offer customers the flexibility to opt for the best VM.
On the other hand, Twitter is expected to deploy the processors by the end of this year. In addition, HPE and Lenovo stated that they would also be utilizing the newly-launched chips. Securing the giant tech companies is a big win for AMD to gain a competitive advantage.
Another tech giant Qualcomm Inc. announced its chip for data centers to accelerate artificial intelligence (AI) processing. The firm plans to utilize the mobile expertise and competence of designing chips with its latest manufacturing technology. Keith Kressin, a Qualcomm executive, said that the critical attribute of chips would be power efficiency. With a rise in competition in the smartphone market and legal battles threatening the licensing business, the tech giant strives to enter the new markets. The semiconductor companies have been endeavoring to either optimize the current designs of chips or provide innovative approaches for tech giants such as Facebook, Google, and others. Joe Spisak, the product manager at Facebook, highlighted that the firm’s computers make nearly 200 trillion predictions each day. This rise in workloads makes it difficult for companies to keep up with surging requirements of power. He added that the requirements two-folds every year. This outlines the urgency of innovative solutions. Many companies such as Intel Corp. have acquired small firms for producing various types of chips that would help AI accelerators. Moreover, it has been striving to add more capabilities for its processors and improve its efficiency to manage the generation of floods of data. Kressin outlined that the firm would offer more details on the product by the end of this year. The chip is designed for inferencing work, making decisions by analyzing data streams. He mentioned that the chip is 50 times more powerful in processing the AI workloads than the flagship mobile chip.
Many companies have turned toward developing their own chips for data centers. The chip making giants such as Nvidia should have a reason to worry about its business as its big customers have opted for the development of their chips. However, this is not the case. Jensen Huang, the CEO of Nvidia, is not concerned about it. Amazon, Tesla, and Facebook are among their big customers that buy graphics cards and began their chip-development projects. Huang stated that the conversation with huge customers is intensifying, and more conversations will help with more significant customers. As there is a wave of interest in AI chips rose across the globe, Nvidia was one of the chipmakers that benefited the most. As huge companies progressed toward deployment of AI chips in their data centers, Nvidia’s stocks multiplied by ten times in less than three years. Google began the development of its own data center chips powered by AI in 2016. However, the company has not started selling its chips. Moreover, Amazon utilizes its data chips in its centers for offering services to the third-party developers. It has been observed that the traditional chipmakers pose a bigger threat than its customers.
On the other hand, there is a trade war between the U.S. and China, intensifying the need to develop its own chips for Huawei. It has announced its plans to invest in developing data center chips. Huawei recently organized the ‘Kunpeng Computing Industry Summit’ event in Beijing and highlighted the plan to invest $435 million (514.7 billion KRW) for developing data center chips and other necessary infrastructures over the next five years. Eric Xu, the Rotating CEO of Huawei, highlighted that the firm would focus on innovation of its data center chip ‘Kunpeng’ along with AI chipset ‘Ascend’ processor. Also, the advanced IT infrastructure will be established as it develops a computing system for Kunpeng. The firm aims to enter the data center CPU market through the expansion of the lineup of the Kunpeng processor. It will possibly collaborate with a British semiconductor design firm known as ARM. This firm offers its IPs (Intellectual Property) on semiconductor design. Huawei’s Kunpeng 920 processor is designed based on ARM. As the firm focuses on the development of new chips, it is evident that it would utilize IPs developed by ARM.
server racks -DepositPhotos