Nvidia will acquire fellow chipmaker Mellanox Technologies for $6.9 billion in a deal that will boost Nvidia’s data center business and better position it to compete against Intel in that space.
Mellanox has headquarters in Sunnyvale, California, and Yokneam, Israel, and is perhaps best known for its high-speed networking interconnects. The two companies have partnered over the years on data center servers that combine Nvidia graphics processing units (GPUs) and Mellanox interconnects. They claim that between them they count every major cloud service provider and computer maker as customers.
The companies expect the transaction to close by the end of this year.
Nvidia reportedly outbid other companies including Intel, programmable silicon company Xilinx, and Microsoft, all of which also had their eye on Mellanox for its strength in artificial intelligence (AI), high-performance computing (HPC), and other big data and analytics workloads.
In an earlier interview, Charles King, president and principal analyst at Pund-IT, noted that “Mellanox has a rich portfolio of high performance Ethernet and Infiniband interconnect technologies and products, and a healthy list of data center vendors, including Dell Technologies and HPE that leverage those offerings in their servers, storage arrays, and other gear.” Any vendor that purchased Mellanox “will add significant muscle to its data center portfolio and deepen its relationships and sales opportunities among numerous data center customers,” he added.
Mellanox pioneered InfiniBand interconnect technology, which along with its high-speed Ethernet products is now used in more than half of the world’s fastest supercomputers and in many hyperscale data centers, the chipmaker claims.
In a statement about the acquisition, Nvidia CEO Jensen Huang said his company will use Mellanox’s technology to optimize data center-scale workloads to achieve better performance and lower operating costs for customers.
“The emergence of AI and data science, as well as billions of simultaneous computer users, is fueling skyrocketing demand on the world’s data centers,” Huang said. “Addressing this demand will require holistic architectures that connect vast numbers of fast computing nodes over intelligent networking fabrics to form a giant data center-scale compute engine.”
Nvidia-Mellanox Data Center Biz
The two chipmakers also worked together to build the world’s two fastest supercomputers, Sierra and Summit, operated by the U.S. Department of Energy. Both use Nvidia GPUs and Mellanox interconnects.
The acquisition follows a new program Nvidia launched last month with North American data center operators to gives companies access to its deep-learning DGX systems through a network of colocation partners.
It also comes on the heels of a particularly disappointing quarter that CEO Jensen Huang described as “unusually turbulent and disappointing” in a shareholder letter. Nvidia’s fourth quarter fiscal 2019 earnings were $2.2 billion, well below its earlier guidance of $2.7 billion.
Data center revenue came up short, and a number of deals did not close, Huang said in the letter. But, he added that Nvidia expects new data center GPU applications and partnerships will drive growth this year.