“While the industry has been talking a lot about co-packaged optics, these are still science experiments, and they’re very proprietary with individual vendors doing their own thing. We’ll embrace open CPO a few years from now, but we think XPO has a 10-year run, especially at 1.6T and 3.2T where you need liquid cooling and you need that kind of capacity. So, all the scale-up racks we’re talking about wouldn’t be possible without XPO or CPC or any one of those technologies,” Ullal said.
“Just as the last decade was greatly influenced by OSFP, the next decade will be greatly influenced by XPO,” Ullal said. “And remember, 99% of the optical market today that we connect to is all pluggable optics. So this is a very crucial invention and innovation, not just for Arista, but the industry at large.”
Enterprise AI: Calm before the storm?
When it comes to AI for enterprise network customers, Arista’s Ullal says it’s just getting started.
The company is seeing a shift from AI training toward more AI inference, “which means you don’t always need the GPU,” Ullal said. “You’re going to have high-end CPUs, and you’re going to have a smaller set of parameters and tokens to manage, and you’re going to have specific agentic AI use cases and applications. We’re seeing very, very early trials and stages. Nothing super big yet.”
Some customers are deploying clusters that are “more inference-based, more agentic AI, edge, inference-based as well,” Ullas said. “I think we’ll see more of that. This is the calm before the storm, if you will. As AI gets more distributed, I think it doesn’t need GPUs alone. It’s going to need more high-performance compute… I think it’s gonna take a couple of years to fully happen.”
