Like many other suppliers of hardware and systems software, Cisco Systems Inc. is looking for ways to cash in on the AI revolution. So far, Cisco has had some success building out its AI pipeline, along with its OEM and ODM peers, but it's been measured in the billions of dollars, far short of the $90 billion-plus that Nvidia expects to generate from its data-center business this year.
This dichotomy shows that the hyperscalers and cloud builders, along with a few major players like Tesla, still account for the majority of AI investments two years after GenAI's explosive growth. And it remains to be seen whether corporations, governments, and research institutions, especially the top 50,000 companies that The Next Platform has consistently followed, will start taking out their checks and investing in AI initiatives that will significantly increase revenue and profits for companies like Cisco.
But Cisco executives say this is starting to happen.
“We continue to capitalize on the multi-billion dollar AI infrastructure opportunity,” Cisco Chief Executive Chuck Robbins said during a conference call with Wall Street analysts to discuss fourth-quarter fiscal 2024 numbers, which ended in July. “We have surpassed $1 billion in AI bookings to date from web-scale customers, and three of the top four hyperscalers have deployed our Ethernet AI Fabric, leveraging Cisco's validated designs for AI infrastructure. We expect to reach another $1 billion in AI product bookings in fiscal 2025.”
“As we discussed in our June investor day, this momentum is fueled by multiple use cases in production at hyperscalers, several of which are AI related. Additionally, we have won multiple design wins, approximately two-thirds of which are AI related. In addition to the web-scale AI opportunity, we believe we are positioned to be a key beneficiary of the widespread adoption of AI applications in the enterprise.”
According to Cisco, most of its AI-related sales are tied to the back-end Ethernet fabric and optics used in AI training networks, meaning it's seeing increased sales of Silicon One ASICs that support AI workflows, as well as white-box switches and Cisco-designed boxes and associated transceivers.
These hyperscale and cloud builder customers design their own systems and manufacture them under contract. Enterprises don't have the scale to do this. That's one reason Cisco has been able to build a customer base of more than 90,000 over the past 15 years with its Unified Computing System blade and rack servers, which combine computing and networking onto a single virtualized platform. With UCS, customers can get many of the benefits of hyperscale servers while offloading the engineering work to Cisco.
It's been a long time since Cisco has spoken specifically about the financials of its UCS business, but there are signs that companies are pouring money into and switching over to servers in anticipation of adding AI capabilities to their applications. These deals aren't included in the $1 billion in revenue Robbins mentioned above, nor are they included in the $1 billion expected for fiscal year 2025. More specifically, Robbins said some companies are reallocating funds they had earmarked for AI projects to upgrading their core infrastructure, both network and servers, to be AI-ready when they decide to do that. This was the first quarter Cisco has made such a move.
This was also Cisco's last quarter of the year to work through some of the massive equipment backlog that built up due to supply chain disruptions caused by the coronavirus pandemic, which could make comparisons tricky going forward.
Cisco converted $3.5 billion to $4 billion of its backlog into revenue in 1Q24, but that growth is not expected in 1Q25. Even if Splunk's data analytics and security platform is contributing to revenue growth (it closed a $28 billion deal at the end of March and contributed $1.4 billion in sales over the past four months), it won't be enough to close the gap.
In the July quarter, Cisco's revenue fell 10.3% to $13.64 billion. Operating income fell even sharper, 38.4%, to $2.62 billion, and net income fell even sharper, 45.4%, to $2.16 billion. Net income as a percentage of revenue was 15.8%, 7 to 10 percentage points lower than Cisco's average.
So we don't believe them when they say they're not cutting 7% of their workforce to cut costs. Cisco is investing heavily in AI and scaling back its hyperscale and cloud builder business, which is undoubtedly hurting the company's margins. When pressed, Cisco said that what they're actually doing is rebalancing their workforce, moving some work to lower-cost locations, and investing more in security, cloud, and AI. With margins under pressure and a declining backlog in fiscal 2025 expected to make it harder to secure revenue, Cisco is likely looking to cut costs and boost profits while investing. The good news is that companies are starting to spend on infrastructure again, which will help.
Cisco's networking group (which includes switching, routing and serving) had revenue of $6.8 billion, down 28.1% last year against its bubble. Robbins said data center witching orders grew double digits, and enterprise routing, campus switching and wireless orders were also “strong.” The Nvidia GPU-based HyperFabric AI clusters, which were introduced in June, have not yet begun customer trials or contributed to the group's revenue.
As for Cisco's other businesses, Security Group revenue increased 81.1% to $1.79 billion, Collaboration Group revenue decreased 0.4 percentage points to $1.02 billion, Observability Group revenue increased 40.9% to $249 million and Services Group revenue increased 6.5% to $3.78 billion.
In the 15 years that Cisco has been in the systems business, we have sought to understand the company's core data center systems business despite significant organizational and reporting changes, and we continue to do so to gain a consistent view of that core data center business.
Our estimates suggest that the core data center business of servers, switches, and routing within data centers (including service provider, enterprise, hyperscale, and cloud) will post revenue of $6.12 billion in 4Q24, down 28.1%, with operating profit of $1.29 billion, down 50.7%. This 21.1% operating margin is lower than the past five-year average of around 30%.
Sign up for our newsletter
We'll deliver the week's highlights, analysis and stories straight to your inbox, with no middle ground.
Subscribe now
Related articles
