EXPLORINEER
en

UK — The Register

Backbone of a beast – charting Nvidia's rise to networking titan

Chips1 min read

Nvidia, best known for graphics chips and AI hardware, has grown into a major networking company. Its networking sales now rival those of Cisco, a long-time leader in network gear.

What happened

Nvidia has become a big force in datacenter networking. It now sells more Ethernet gear for datacenters than any other vendor. This article traces how the company got there, starting with its 2019 purchase of Mellanox.

Key numbers

  • Nvidia's networking revenue topped $17.5 billion in Q2 (the second quarter of its financial year).
  • Cisco's total revenue in the same period, across all products, was $17.3 billion.
  • Nvidia bought Mellanox in 2019 for $6.9 billion.
  • Recursion's data sets total more than 50 petabytes, which is a huge amount of data.

How Nvidia got here

Nvidia bought Mellanox to gain networking skills. Mellanox was known for ConnectX network cards and InfiniBand switches, which are fast links that connect computers in a cluster. Nvidia realized that big data centers want whole clusters of machines that work together, not just single chips or servers. Jensen Huang, Nvidia's CEO, calls this approach co-design, which means building chips, networks, and software to work well together.

Why it matters

Customers like Recursion say that buying compute and networking from one vendor makes systems easier to set up, tune, and fix. This strength helps Nvidia win AI business. It also puts Nvidia in direct competition with older network companies such as Cisco.

What to watch

  • Whether Nvidia keeps growing its networking sales faster than rivals.
  • How Cisco and other network makers respond to Nvidia's push into Ethernet.
  • Whether more AI customers choose one vendor for both chips and networks.

Glossary

Mellanox
A networking company that Nvidia bought in 2019 for $6.9 billion. It made fast network cards and switches.
InfiniBand
A very fast network technology that links many computers together in a cluster.
Ethernet
The most common type of wired network technology used in homes and datacenters.
Cisco
A large company that makes networking gear such as switches and routers.
Recursion
An AI company that uses machine learning to help find new medicines.

Same story, different eyes

Ad

Add 150 words. Learn 10 a day.

Dicter word books send only the words you learn to your review notes.

More in this topic

ChipsThe Verge

Sam Altman says it’s better for RAM to be in a data center than on your desk

Asked about the ongoing RAM shortage, OpenAI CEO Sam Altman said in a Cheeky Pint podcast interview that it is "terrible" for the world's limited DRAM supply to sit in personal computers used only about five percent of the time. He argued that maximum utilization in the cloud is needed in a supply-constrained environment, and noted that memory makers Samsung and SK Hynix have long-term deals with cloud operators and projects like OpenAI's Stargate that have reserved much of the world's RAM output over the next few years.

ChipsTechNode

AI inference chip startup Nano-core Chip raises over RMB 1 billion in Series C2 funding

Hangzhou-based AI inference chip developer Nano-core Chip announced it raised more than RMB 1 billion in Series C2 funding, led by JianTou Investment with participation from Hangzhou Capital and Kaitai Capital. The company plans to use the proceeds for product development and preparation for volume production. It said its first three chips have completed tape-out, with customer sampling and initial volume shipments planned for early 2027.

ChipsThe Register

PC chip plunge wipes out a quarter of Intel’s value

Intel's share price fell almost a quarter on Thursday after the company reported a steep drop in PC processor sales, despite Q3 revenue rising 21 percent year on year to $16.5 billion on strong datacenter and AI growth. CEO Lip Bu Tan predicted the PC chip market will shrink 12 percent next year and forecast a major CPU boom for datacenters, but executives admitted Intel may struggle to meet demand due to supply constraints such as silicon wafer sourcing.

ChipsZDNet Korea

[Profile] Ahn Hyun, President of SK hynix Americas

SK hynix announced on the 30th that Ahn Hyun, its Chief Development Officer (CDO) president, will be appointed head of the newly created Americas division, a president-level organization based in the United States, effective November 1. The company also released Ahn's career summary, which includes a 1967 birth year, a nuclear engineering degree from Seoul National University, and roles including NAND development planning group head and head of the Global AI Research Center.