
New money, same AI arms race
AMD has officially wandered deeper into the AI infrastructure rabbit hole. The chipmaker joined Israeli networking software company DriveNets in a $410 million funding round, giving it a seat at the table for the part of AI that doesn’t get the headlines but absolutely keeps the lights on: networking.
Why this matters more than it sounds
DriveNets builds software that helps telecom operators and data centers run high-speed AI networks using off-the-shelf hardware instead of pricey proprietary gear. Translation: if GPUs are the celebrity chefs, networking is the kitchen staff making sure dinner doesn’t get served cold.
That matters because modern AI deployments are basically massive team sports. Thousands of GPUs need to talk to each other fast, and the more models grow, the more the networking layer becomes a bottleneck. So when AMD puts capital into a company like DriveNets, it’s not just a side quest — it’s a bet that the next wave of AI spending will spread beyond chips and into the plumbing connecting them.
Follow the capex trail
The timing lines up neatly with the broader AI spending splurge. Microsoft, Amazon, Alphabet and Meta are all expected to pour huge sums into data centers, networking gear and compute hardware, which means the winners might not just be the chip names everyone already knows.
DriveNets said the round should help it expand inventory and meet demand for Ethernet-based networking systems, and it claims more than $1 billion in customer business. If that holds up, the AI infrastructure pile may be getting taller — and a lot less chip-only than the market first assumed.
Big picture: AMD’s move says the AI trade is evolving. First came the chips. Now comes the stuff that lets the chips actually chat with each other.
