Those who follow networking likely know that the long-in-process merger between HP Enterprise (HPE) and Juniper Networks is being challenged by the DoJ. Then, (both issued the same press release on February 28) we heard that Juniper and IBM were cooperating to simplify netops, and the deal would include “joint sales, marketing and product integration efforts.” IBM is quoted, saying “Our collaboration with Juniper illustrates how ecosystem partnerships can help accelerate the adoption of AI for critical business use cases such as network management, demonstrating even greater value to enterprise customers.” Wait! How could a network company proposing to merge with HPE collaborate with an HPE competitor? What’s going on? Essentially, we have three possibilities we need to examine as a hierarchy.
In the top layer, IBM and Juniper may be responding to the DoJ action, presuming the deal with HPE is off, they may be simply creating another Juniper relationship to exploit post-deal, or they might be using the relationship to relieve DoJ concerns. If the former, we have two possibilities. IBM is prepping for an independent Juniper, or IBM may be looking to acquire/merge with Juniper itself. We’ll talk about all of these.
It’s really hard for me to see HPE and IBM both promoting the same Juniper stuff. The press release I quoted says that Juniper and IBM would be cooperating in sales and marketing. Would that work if HPE owned Juniper? We’ll get to that below.
If Juniper/HPE think that the DoJ would kill the deal, then both companies would be exploring their alternatives to the merger. For HPE, they’d need to look at another networking company, or simply forget mergers as a pathway to expanding their network business. If they fear that the DoJ would kill the Juniper deal on competitive grounds, could HPE hope to have any deal approved? I’d suspect not, so they’d walk away.
Juniper is another story. Data center networking drives enterprise network capex, and data center hosting and applications drive data center networking. Juniper would benefit from having HPE pull them through into deals, because HPE sells what runs in the data center. Simple, right? Thus, we could expect to see Juniper doing one of two things. One, they try to do something cooperative with another data-center strategic influence giant, and IBM is the absolute number one there. Two, they could try to get acquired by such a giant, meaning Juniper might replace HPE with IBM as a parent.
If the goal is to try to address DoJ concerns, then the presumption would be that the “IBM Guest Services”, which explicitly links the cooperation to Mist AI and WiFi networks, might be appropriate, given that the whole WiFi thing was the basis of DoJ objections. Don’t worry, regulators, HPE isn’t going to lock up WiFi AI because Juniper is partnering with a major HPE competitor, IBM.
I think that it’s pretty clear that the DoJ means business; they’re not being discouraged by light-hearted comments. Thus, I think we can assume that the Juniper-IBM deal is in fact a reaction to the DoJ. I think it’s a kind of hedging. It may be enough to convince the DoJ that they should approve Juniper/HPE, and if so the deal goes through. IBM and HPE then have to somehow work out how any cooperation would work down the line, but that’s not impossible, given that a large percentage of IBM’s Red Hat customers probably run on HPE gear. If the DoJ is intransigent after the IBM deal, then Juniper and IBM have to decide how far they’d be wanting to take cooperation.
IBM used to be in the network business, but their mainstay was their proprietary (and insightfully excellent) SNA technology. As IP players, they never excelled, and they ended up selling off their network business to rival Cisco. Would IBM now acquire Cisco’s primary competitor? Seems questionable. A cooperative deal with Juniper doesn’t put them in Cisco’s face, and Cisco is after all the largest network player in enterprise accounts. We might expect IBM to do deals with other network vendors too, under this level of cooperation, but it would be hard to do that if they actually bought Juniper.
But…but…but. This doesn’t address the question of Juniper’s viability. Suppose the DoJ squashes the HPE merger. That means that if IBM doesn’t buy Juniper instead, Juniper has to rely on cooperation with IBM to pull up its own sales and revenues, or essentially contend they can go it alone. Their quarters haven’t been bad, after all.
Now, we return to the question of whether IBM and HPE are incompatible bedfellows, or just perhaps a bit strange. The IBM cooperation is valuable to Juniper, period. IBM has the best and smartest position on AI, according to enterprises. It has the largest strategic influence on enterprise planning. Juniper’s AI-native networking push surely aligns with that. With regard to HPE/IBM, remember that IBM doesn’t make servers (other than its mainframes, which are hardly a growth-market offering), so it really isn’t necessarily competing excessively with HPE. Self-hosted AI benefits IBM. If it sells servers, it benefits HPE. If it sells switches it benefits Juniper.
I think an IBM/Juniper deal would be even more interesting, even profound, than an HPE/Juniper deal. However, I think that IBM’s position in the broad market and in AI in particular is a software position. Red Hat and watsonx are both software, and Red Hat is the hook IBM has hung its entire business on as a means of broad-market opportunity harvesting. The question is whether buying Juniper would contaminate this, and that’s a tough one. Would it alienate Cisco? Surely Cisco wouldn’t like it, but what impact might that have. Cisco can’t supply an alternative to Red Hat, nor can they match IBM’s influence with major accounts.
If the DoJ approves the HPE/Juniper deal, then I think IBM’s deal with Juniper probably offers everyone something. IBM gets a showcase example of AI agent operations. HPE gets watsonx and IBM AI options, and Juniper gets another channel to help them sell AI networking.
Can we play out what happens if DoJ does reject the HPE merger with Juniper? If that happens, then IBM is most likely to push hard on the cooperation with Juniper, and HPE and IBM still become more strategic competitors in network equipment for enterprises. HPE is, I think, harmed immediately. Juniper becomes more and more dependent on IBM for strategic influence, even in AI, and eventually it might be prudent for IBM to acquire Juniper rather than leave them hanging in the market, potentially a target for a competitor.
Whatever happens with the DoJ, I think the Juniper/IBM cooperation raises the stakes, making what might have been a sort of yawn decision in the view of many, into what might prove a profound pivot point for IT, networking, and AI.
