Cisco Systems Inc. has provided more details about the timing of revenue from expected orders related to artificial intelligence, but the networking giant appears to be a late bloomer in the AI space.
Wednesday, Cisco
CSCO
We have repeatedly commented from the previous quarter that orders related to AI infrastructure construction products are expected to reach $1 billion in fiscal 2025. “The bulk of that $1 billion, let me be clear, orders will turn into revenue in fiscal 2025,” Cisco Chief Financial Officer Scott Herren told analysts on an earnings call. Cisco’s fiscal year 2025 begins in August of this year and ends in July 2025.
But executives also said that its pipeline has continued to grow and has actually tripled in the past 90 days. In an interview with MarketWatch, Herren said “pipeline” refers to opportunities that the company’s salespeople identify during ongoing communication with customers.
Comments on AI and its potential were a bright spot in a somewhat weaker quarter despite disappointing guidance and news of about 4,200 job cuts, about 5% of the workforce. Cisco also said there has been “an increased level of vigilance and scrutiny of transactions,” creating a great deal of uncertainty for many companies.
Compared to some of our chip manufacturing partners like Nvidia Corp.
NVDA
and Advanced Micro Devices, Inc.
AMD
,
Both companies already have and each project billions of dollars in revenue from sales of AI-related chips, but Cisco’s projections are on a much smaller scale. Cisco reported that its fiscal year 2023 revenue was $56.9 billion.
In building the infrastructure for today’s AI systems, the networking standard technology used in large, high-performance data centers is Infiniband, an Nvidia proprietary technology.
However, this situation could change in the next year or so as companies start building their own internal infrastructure to support AI in their own data centers, as opposed to companies like Metaplatform. .
meta
We’re spending billions of dollars on AI infrastructure to support our web services.
“Long term, the people who are building these large-scale training models don’t want individuals to lock themselves into a specific protocol, and that’s what Infiniband is. [Nvidia CEO] jensen [Huang] “He came to us and said, ‘We want to partner with Cisco,’ because he sees the same thing that we see: that ultimately, Ethernet is going to win in this space.” Herren said, referring to industry standard networking protocols. “Today we have Infiniband. Ultimately, Ethernet will win, not only on the front end, but also on the back end, where training takes place.”
Cisco’s current forecast for AI-related revenue includes little in the way of product sales generated by its recent partnership with Nvidia, and Herren said those eventual sales are a tailwind. “Enterprises will build their own AI infrastructure and use web scalers for some of their AI needs,” he said, adding that enterprises will build AI infrastructure for their own information and data that they want to keep secure. He added that it would be. He declined to give any further predictions about what future revenue Cisco could earn as a result of its partnership with Nvidia.
“We’re not going to do that yet, but there’s no question that it’s an upward force,” he said.
Cisco certainly has networking competitors eyeing the same pie. That is, companies looking to build their own AI-enabled data centers. However, it seems too early to write off the company as having missed the boat in the AI field.
