Ullal expects ESUN-based scale-up trials to begin in 2027 with rackscale systems of 128-144 accelerators, scaling to about 1,000 across a row.
she expected for scale up trails based on the ESUN protocol to start in 2027, initially with rackscale systems with a radix of 128 or 144 accelerators within a rack and with as many as 1,000 accelerators across a row.
Ullal forecasts Arista will network 10 gigawatts of AI cluster capacity and link ten or more accelerator types by end of 2027, with over a hundred Etherlink customers already.
Here in 2026, Arista already has more than a hundred customers using its AI-centric Etherlink switches, and by the end of 2027, she is forecasting that the company will provide networking for 10 gigawatts of AI cluster capacity and will be linking ten or more different kinds of accelerators.
Jayshree Ullal said Arista's third-quarter results exceeded even her own high expectations, reflecting strong AI and cloud networking demand.
CEO Jayshree Ullal said the results surpassed even my high expectations.
Ullal says strong momentum across customer sectors shows Arista's growing relevance in the generative and agentic AI era.
This strong momentum across customer sectors highlights our increased relevance for the generative and agentic AI era,
Ullal says the industry is shifting from compute-centric stacks to network-centric architectures, and Arista aims to break down barriers between compute and networking.
The industry, we believe, is shifting from compute-centric vertical stack to a holistic network-centric approach. Arista is committed to unleashing and breaking down the barriers between compute and networking.
Ullal describes an AI super cycle, with over 100 Etherlink customers and a forecast of more than 10 GW deployed across 10+ accelerator types by end of 2027.
In 2026, we are in a super cycle of AI, having already surpassed 100 Etherlink customers. By the end of '27, we are forecasting greater than 10 gigawatts of deployment across 10 or more different types of compute accelerators. What a stark inflection in just 3 years.
Ullal argues the shift to distributed inference demands smarter networking, making the network a cornerstone of token efficiency.
We are witnessing the migration from highly centralized training compute to disaggregated and distributed inference, prefill and decode, demanding much more intelligent networking and colocation of compute, storage and KV cache resources rather than simply focusing on raw feeds and speeds and flops of compute. A well-designed network is becoming a foundational cornerstone for optimal token efficiency.
Ullal expects scale-up Ethernet trials in 2027 and production in 2028.
We expect scale-up trials to begin in 2027, driving Ethernet to larger radix of 126 to 144 within a rack and across 1,000 across racks, with production expected for scale-up in 2028.
Ullal guides 2027 revenue to about $18.5B, with AI fabrics doubling to ~$7B and campus reaching ~$2B.
Our first look at 2027 guidance range puts us in the 38% to 40% range for growth annually to approximately $18.5 billion. Within this envelope, we expect our AI fabrics target to be the fastest-growing segment and to double to approximately $7 billion, while our cognitive campus represents a healthy growth of 60%, much faster than the market to approximately $2 billion in 2027.
Ullal expects scale-across to grow beyond one third of the mix in 2027.
scale-across, we think, will be at least 1/3. It will probably go higher in '27. And so the split this year is roughly 70-30. The split next year may be -- I can't venture to guess, but I think the 70 may come down and the 30 would go up when it comes to scale-across because it's a very strong use case for us.
Ullal says pluggable optics will dominate, favoring open CPO only in the 2028-2030 timeframe.
We, first of all, strongly believe in copper when you can and optics if you must. So we think the world is going to remain in a very majority fashion on pluggable optics. We don't want to build proprietary co-packaged optics. We think they'll come on to the market in the '28 to '30 time frame. It's not imminent, but meanwhile, long live pluggable optics.
Ullal says scale-up is tough for lacking software advantages but Arista is well positioned to connect non-NVIDIA accelerators.
scale-up has been a tough market for us because it doesn't have all the software advantages that we bring, but it absolutely has the hardware skills and all the non-NVIDIA accelerators that Arista is in the best position to connect to.
Ullal says EOS's low vulnerability count positions Arista to monetize AVA in coming years.
Everybody else is dealing with a game of whack-a-mole where they've got thousands of vulnerabilities and no AVA can possibly keep track of it. So I think we feel in a very good position to monetize AVA in the next few years.
Ullal rejects the idea that the 2027 guide is conservative, calling it far above earlier projections.
when some things are growing 60% and 100%, and then you say the overall growth is 38% to 40%, then obviously, some of the things are not growing as fast, right? So I wouldn't read any conservatism into these numbers. These are wildly higher and optimistic than we had once projected.
Ullal says the industry is shifting from a compute-centric vertical stack to a holistic network-centric approach, driven by increasingly heterogeneous AI chip options.
"The industry, we believe, is shifting from compute-centric vertical stack to a holistic network-centric approach,"
Ullal argues that a well-designed network is becoming a foundational cornerstone for optimal token efficiency.
"A well designed network is becoming a foundational cornerstone for optimal token efficiency,"
Ullal says Arista's record growth and profitability across customer sectors show its growing relevance in the generative and agentic AI era.
“Arista registered another momentous quarter with the highest year-over-year revenue growth since our IPO and record profitability, surpassing even my high expectations,” stated Jayshree Ullal, Chairperson and CEO of Arista Networks. “This strong momentum across customer sectors highlights our increased relevance for the generative and agentic AI era.“
Ullal said customers view networking as the center of how their infrastructure operates.
Chief executive Jayshree Ullal said customers see networking as the center of how their infrastructure runs.