HPE and Juniper: Why?

Very few enterprises I’ve chatted with can answer the question “Why would HPE buy Juniper?” Very few competitors of either company, very few on Wall Street, and even very few employees of either company are convinced they can answer it either.

Attempted justifications of the deal tend to expose three interpretations of one old adage. One take might mean that “the whole is greater than the sum of its parts,” as the original adage says. Or that “the whole can be run cheaper than the sum of its parts,” which is the typical Wall Street perspective. Or maybe “the whole needs another part,” which is what the media seems to think. Everything, for both companies, depends on which view we select.

The second of the two views is the most popular. A good M&A is based on the idea that there’s a lot of staff redundancy in the combination of two companies, and you can save a boatload of costs by merging and laying off employees. No revolutionary change in strategy, or in product plans. Economy of scale is the game. But consolidation is usually a response to commoditization, which means that at best, you’re circling the drain in a more favorable position (if that’s possible). Weak.

All three options for the “why” of the deal share a common underpinning. They all reject simple operational efficiency in favor of some form of symbiosis. Juniper stuff will make HPE better, HPE stuff will make Juniper better, or the combination of stuffs will make everyone better. These options all depend on the “stuff” that’s being added. What is that? That’s the real question, and there are only two answers.

The first is the simplest. Juniper has AI and HPE wants it. On the surface that seems a bit of a stretch, because Juniper has a lot more than AI. Do you buy a hardware store because you need a nut or bolt? If so, I’d like to chat with you about opportunities. But there is a deeper slant on this point we could take, which goes back to that almost-age-old problem of complexity. Operations, both network and data center, are plagued by human error, security risks, excessive costs, inertia, and just about every other problem we can have. Juniper’s AI, from Mist AI to Marvis Virtual Network Assistant, gets pretty good ratings from enterprises as a tool to improve operations. Juniper’s Apstra is a virtual-data-center concept that extends AI to the data center, so HPE could tie their stuff to Juniper stuff through Apstra. Could that be what HPE wants?

Maybe, but we still have to ask why HPE wouldn’t have just bought the AI. Answering that raises our second answer: They’d have to have a reason they would want the network gear too. One possibility is that HPE believes that they can wring more value out of Juniper than Juniper itself can. Enterprises offer a reason HPE might be able to do that. In 4Q23, the number of enterprises who said that their network planning and procurement was most influenced by their data center vendor reached almost 90%. Application software, platform software and middleware, and computing all have more influence than network issues themselves. HPE could, in fact, peddle Juniper gear more effectively than Juniper…if they understand this point and address it in their sales/marketing programs.

The problem for HPE here is that despite their greater strategic influence as IT vendors, all the data center elements are commoditizing just like networks are. Apparently having influence isn’t the same as making a business case. HPE needs to green up some of the brown fields of IT. They need a set of new business cases that will drive the new applications that will justify the new computing and demand more networking. That’s a long value chain to follow, and we don’t have any clear signs that HPE or Juniper can avoid kinking the chain along the way. We do, though, have some signs about what the business case will have to look like, and maybe that can lead us to what HPE has in mind.

There have been three times in the last 70 years when enterprise spending on information technology grew faster than GDP. The first was the mainframe era, the second the minicomputer period, and the last the PC. The benefit transformation was batch computing to online transaction processing to personal computing, and every step brought IT closer to workers. Then it stopped, so logically a hot business case would get it started again, but what? Answer: Point of activity empowerment, through real-world, real-time computing.

Suppose we imagine an enormous virtual-metaverse digital twin of the real world. Suppose it’s made up of many smaller digital twins—twins of rooms, buildings, towns and cities, all connected by networks. Suppose that consumers and workers have a foot in both worlds, and that their footing in the virtual world lets us give them all the information they need to do their jobs or live their lives, right at the point and time they need it. How much would that be worth in productivity, in quality of life? Network and IT operations have gotten more complicated as workers went from getting sales slips keypunched onto cards to an age where they entered things in real time and their computer talked to their company’s computer. Imagine how complicated it would be to make this real-world-connected-digital-twin thing work? Failing globally at the speed of light is not an option. We need a superintelligence on our side.

Like what AI provides, and like Juniper has just announced with its AI-Native Networking. Juniper also has a foot in the compute door, with their universal data center virtualization tool, Apstra, and it’s also now included in its AI-Native Networking. It’s not clear that AI-Native Networking is so critical to current network/IT operations that it would make the HPE acquisition a big success under current conditions, but it very likely would be if HPE pushes point-of-activity empowerment. The combination of HPE and Juniper could then be essential in realizing the dream. It may be that a fear of what AI could mean in this application has driven Cisco’s recent relationship with Nvidia to bring AI into Cisco networks, too.

The obvious risk here is that the HPE/Juniper deal might not close for a year. During that time, every competitor will have an opportunity to assess what the combined company might do, and to work out countermeasures. But HPE and Juniper could take that same year to do creative things of their own, dream of things others couldn’t easily counter.

The risk in all of this is that you can’t realize dreams you don’t have, and we don’t know what HPE is dreaming here and now. Just because most of the justifications offered for acquiring Juniper are weak doesn’t mean that HPE didn’t pick one of them. The real-world-real-time opportunity hasn’t emerged this year, even if IBM’s threat to HPE has. HPE didn’t recognize the real-time opportunity earlier, nor did Juniper. Can they emerge from tactics and quarterly reports and dream? Robert Kennedy had a saying that’s appropriate here. “Some men see things as they are and ask ‘Why?’ I dream of things that never were and ask ‘Why not?’” Well, HPE, why not?