The data center of the future will be used to crunch the skyrocketing demands of artificial intelligence (AI) and machine learning as well as scientific computing and data analytics.
Because of this, the data center is becoming a fast-growing and promising market but one that will require high-performance computing (HPC) with tens of thousands of compute nodes needed to meet the demands of the market.
Worldwide spending on AI systems is forecast to reach $35.8 billion in 2019, an increase of 44% over the amount spent in 2018, according to new research from market research firm IDC. Spending on AI systems will more than double to $79.2 billion by 2022 with a compound annual growth rate (CAGR) of 38% during the forecast period, IDC said.
As more demand is needed, companies are circling their wagons to be in the best position and, as a result, Nvidia Corp. announced on March 11 that it signed an agreement to acquire intelligent interconnect device and service maker Mellanox for $6.9 billion.
According to Luca De Ambroggi, senior research and analyst director for artificial intelligence at market research firm IHS Markit, other key players targeting the data center and HPC domains such as Xilinx and Intel also participated in the auction for Mellanox, which confirms the importance of HPC in the market.
Data centers will likely become much more structured and efficient than they are today to address a high cost pressure and to match aggressive performance requirements, resulting in sharp growth of processing requirements and data analytics in future equipment. The Nvidia acquisition of Mellanox reflects this, showing that the computing market is no longer just about the central processing unit (CPU), general processing unit (GPU) or field programmable gate arrays (FPGA), De Ambroggi said.
While Nvidia has already showcased its computer hardware acceleration through its GPU technology, tackling the overall end-to-end data center architecture will require the right software as well.
“Mellanox will contribute with their hardware and software solutions such as InfiniBand and Ethernet product not by just addressing fast operations and networking within servers, but also address another key aspect that is fast storage I/O ops,” De Ambroggi said.
However, challenges between Nvidia and Mellanox and their disparate data center business models may arise as their teams converge.