Advanced Computing in the Age of AI | Saturday, September 23, 2023

Cisco CEO Predicts ‘Brutal, Brutal’ IT Consolidation 

John Chambers, CEO at Cisco Systems for the past 25 years, has hosted a CiscoLive customer and partner events each year he has steered the company, moving it from routing into switching, telephony, collaboration, and most recently systems. Like other IT industry titans, Cisco has made its share of mistakes, and the fact that it has adapted well over the past two and a half decades is no guarantee that it will do so over the coming decades. The same holds true for all other technology companies, and indeed, all of the private sector.

That was the blunt message from Chambers in his opening keynote for CiscoLive in San Francisco, an annual shindig to preach about embracing change and product transitions in the IT sector. Chambers said that the pace of change in IT was accelerating.

"You are going to see brutal, brutal consolidation in the IT industry, where out of the top five players, only two or three of us will be meaningful in as quick as five years," Chambers said. "You will see this disruption not so much in consolidations, but almost like musical chairs. And by the way, the same thing is going to happen in each of your industries in the private sector."

Companies come and go on the Fortune 500 or S&P 500 lists, or indeed on the major stock indexes that representing the cream of the crop of corporate entities. Of the companies that were on the Fortune 500 25 years ago when Chambers first took over Cisco, only 24 percent of them are on the list today. (Cisco was not one of them at the time; it was number 64 on the 2013 Fortune 500 list, and is climbing.) Of the major corporations in the world today, only about a third of them will survive the next 25 years, Chambers predicted, and lest you think somehow smaller companies might be spared, he said that 87 percent of all private companies would have some kind of financial stall in the next 25 years and only 11 percent of them would recover.

The idea that Chambers wanted to instill with such statistics is that embracing change is no longer enough, you have to relish accelerated change and figure out how to create sustainable competitive advantage. Chambers used Cisco's evolution in networking and collaboration over the decades as well as its rise in the converged systems arena as examples of how Cisco remains relevant.


Of the 50 companies that were in the networking space when Cisco entered the market two and a half decades ago, all of them are either gone or acquired. "And if you look at three years ago, Huawei, Avaya, and Juniper were the worries, and they were going to eat our lunch. Nobody eats our lunch," Chambers said emphatically, and then added: "If you look at most of the companies in the industry, most of them will not exist in ten years."

Everybody will have to change, and the good news for Cisco is that the network is at the center of everything, said Chambers. (This is a mantra that Cisco has espoused for years, and so did Sun Microsystems – the Network is the Computer – many years before it was eaten by Oracle.) That includes, by the way, systems – a market that Cisco only entered five years ago, causing upheavals in networking and servers alike. And basically, Chambers said that it would come down to all of the major players being caught been Cisco and so-called "white label" systems.

To illustrate his point, Chambers flashed up a slide showing revenue growth for the top X86 system makers:


As EnterpriseTech has discussed many times, the incumbent server makers are having a hard time competing against original design manufacturers (ODMs) among the hyperscale datacenter operators. Of the largest cloud builders, only Microsoft buys servers from a tier one maker anymore (either Dell or Hewlett-Packard, depending on the quarter) and that could very well change now that Microsoft has designed its own server chassis and nodes and contributed them to the Open Compute Project. Cisco and its various partners peddling its Unified Computing System machines pretty much own the market for converged systems, which weld together servers, switching, and storage and provide a consistent management framework across physical and virtual machines. Cisco had a 65.8 percent revenue share of this $4 billion market.

"Individual companies will get squeezed between Cisco and white label if we do our job right," said Chambers.

This assumes, of course, that the markets for minimalist machines tuned for hyperscale datacenters and converged systems aimed at enterprises for cloud and other virtualized applications continue to grow at current rates. It is not tough to imagine the hyperscale market slice doubling its share of shipments and doing slightly less than that on revenues; it is harder to believe that converged systems can keep growing at around 50 percent per year (including both raw machines and those tuned up with application software). But it is possible. The overall server market grew only 4 percent in terms of aggregate revenues last year, and if you take out converged systems and hyperscale machines, the overall market is in decline.

Chambers said that Cisco was similarly prepared for the white label movement in networking, where companies want an open switch and an open source network operating system instead of the kind of sealed boxes that Cisco provides.


"We saw white label coming three years ago – I talked about it at a financial conference," he said, "Standalone products will be in real trouble. Your differentiation if you take merchant silicon and bare metal and on top of that open source, that is going to be an interesting pricing capability. But if you look at the hidden costs, we are about 27 percent less expensive."

Past success is never a guarantee of future success, as Wall Street and, the history of the Fortune 500, the Dow Jones Industrial Average, or the S&P 500 show you. To be sure, Cisco has up and down times and makes some bad acquisitions – Chambers himself quipped that one out of three acquisitions would fail, but you can't tell on the front end which one it will be – but it has a pretty good track record of entering new markets.


Cisco doesn't enter new markets if it can attain at least a 40 percent market share over the long haul, and it only enters markets where it can provide some kind of sustainable differentiation and margins. Of the eighteen major product lines that Cisco has, it is number one in a dozen of them, number two in four of them, and number three in two of them. This is not a bad track record.

"I don't want to mislead you," said Chambers. "We are not moving fast enough. We are pulling away from our peers, but we have to accelerate that."

As an example of the acceleration that Cisco is bringing to bear, Chambers held up a ASIC at the heart of its CRS-X carrier-grade core routers, which has 4 billion transistors and 336 cores. This chip he said, cost about $250 million to develop and was brought to market much faster than prior generations of ASICs. On the top-end Nexus 9000 switches, which were announced last fall and which are ramping now, Cisco went from idea to finished product in 21 months (of course, it had a little help through the acquisition of Insieme Networks, which was spun out of and then spun back into the company). The point is that the ASICs at the heart of these systems took 14 and 16 months, respectively, to develop, and that was twice as fast as what Cisco could do even three years ago.

Speed will matter, but the sales pitch is going to matter more, and even Chambers said that the Cisco sales force would have to shift from peddling switches and routers to selling architectures and business outcomes.

That is the big bet with the Nexus 9000 switches and their Application Centric Infrastructure (ACI) approach, which bakes software-defined networking policies and security that span from the core to the edge. This approach to SDN requires Cisco iron, and even if it does have open APIs for integration, there are some companies that want open source networking stacks. And Cisco knows there is not much it can do about that. But for the enterprise customers – the same shops who have been coming back again and again for UCS iron – the Nexus switch and ACI story will no doubt resonate. Chambers threw up some stats showing that provisioning applications on the network would drop from weeks to minutes and that capital expenses would drop a little and operational expenses would drop a lot, resulting in a 41 percent cost savings over Cisco networks before ACI came along.

Some companies will go the ACI route because Cisco is their switching vendor. And others, particularly hyperscale datacenter operators who are comfortable designing their own switches already or are eager to start, will probably go the white label route and there wasn't much Cisco could do to change that. The big group in the middle looking at their SDN choices is a little bit harder to predict.