Extreme Networks bets on agentic AI networking as supply chain uncertainty weighs on margins – News

Despite the seven straight quarters of revenue growth and sales and earnings results that topped analysts’ estimates in the most recent quarter, Extreme Networks Inc.’s stock is down by a third since its Sept. 11 high.

Chief Executive Ed Meyercord (pictured) said supply chain uncertainty tied to surging investment in artificial intelligence is creating near-term margin pressure across the networking industry, but his company is managing the cycle through pricing flexibility, more targeted purchasing and alternative sourcing strategies.

In an interview with News, Meyercord said investors have been concerned that large AI infrastructure buildouts are consuming key components such as memory, driving up costs and potentially restricting supply for networking vendors. He described the issue as “an industry phenomenon. We’ve seen this before.”

The company is preparing for the current supply-demand imbalance to last another 18 to 24 months as semiconductor prices rise and manufacturers race to expand capacity. However, he said Extreme has tools to offset the impact.

‘Critical infrastructure’

One is the ability to raise prices, which it did, by around 7% last year. Meyercord also noted that networking is “critical infrastructure” that is less vulnerable to economic cycles. Extreme’s small size relative to giant rival Cisco Systems Inc. can also be an advantage because the company is “chasing a container, not a container ship,” allowing it to secure smaller volumes of components more easily than larger rivals.

In the short term, Meyercord said, Extreme Networks is benefiting from some market turmoil triggered by Hewlett Packard Enterprise Co.’s planned acquisition of Juniper Networks Inc. Uncertainty about HPE’s plans for its new subsidiary has unsettled channel partners and customers, he said.

“It’s helpful to Extreme,” he said. Juniper’s style had “more personality” than HPE, and he expects cultural clashes to stretch the full integration process out to his long as two years.

Consolidation has already opened up some recruiting opportunities. Meyercord cited two recent hires from Juniper, including Gos Hein van de Wouw as the head of Europe/Middle East/Africa and Joe Spencer as global channel chief.

Meyercord also pointed to recent changes in Cisco’s channel program as another competitive opportunity. The overhaul has made Cisco a less profitable option for many partners outside the top tier. “It’s a little harder to make money,” he said. “If you’re not kind of in the top 50, then you might consider alternative vendors.”

AI-first platform

Extreme’s technology strategy is centered on an “AI-first” networking platform, which Meyercord said is shifting from traditional machine learning toward agentic AI systems designed to automate network management tasks.

“The first generation was machine learning,” he said, describing predictive maintenance capabilities. But he said Extreme is now building a unified platform with “an agentic AI core,” which he said can automate tasks such as troubleshooting, license management and network planning.

Meyercord said Extreme’s “service agent” can reduce troubleshooting time from days to minutes by pulling device logs, analyzing network conditions and generating recommended fixes while also showing the steps it took to reach its conclusions.

“The thing that’s great about the service agent is it shows you everything it has done,” he said.

Meyercord said the company plans to make its AI agents available to ecosystem partners, enabling integration into broader enterprise workflows through what he described as an “agent exchange.”

He pointed to potential integrations with ServiceNow Inc. for trouble-ticket workflows and location-based applications in retail, such as identifying a shopper’s position in a store to deliver targeted promotions. “This is where Extreme is going to have an advantage,” he said.

Meyercord said Extreme is closely aligned with Microsoft Corp. and Amazon Web Services Inc., noting that the company has expanded its AI team in Seattle to about 50 people working with the hyperscalers’ engineering groups. He said he views cloud giants more as ecosystem enablers than competitors, arguing that marketplaces will create new distribution channels for Extreme’s agents.

Shift to edge

In customer deployments, Meyercord said, most AI-driven networking demand today is concentrated in the data center, as enterprises rely on hyperscalers for model training and experimentation. Over time, he expects more AI workloads to shift toward the edge, creating demand for dedicated networking capacity and fabric-based architectures.

Wi-Fi 7 is quickly gaining momentum, with customers responding positively to the standard’s performance improvements and energy efficiency. Meyercord said Extreme was “first to market” with Wi-Fi 7 products in partnership with Broadcom Inc. and is seeing adoption ramp. “We’ve had a lot of early success, and we’re seeing that technology ramp significantly,” he said.

Looking ahead, Meyercord said the company expects AI-driven automation to help relieve overextended information technology teams, not by eliminating staff but by enabling network professionals to focus on higher-value projects and permitting more work to be virtual.

“AI is going to provide a welcome addition to a bunch of team members that are virtual,” he said.

Photo: Extreme Networks