No matter the vertical, nothing is as difficult to assess as the anti-trust implications of M&A. In tech, that’s particularly true because of the pace of evolution and the breadth of influence and symbiosis within the space. Rulings that are based on traditional factors are almost certain to be flawed in reasoning, and thus likely to come to the wrong conclusion. That’s surely true of the decision to file suit to block the HPE merger with Juniper Networks.
I’ve worked with both HPE and Juniper in the past, and I’ve blogged about the deal as well, so you won’t be surprised that I have strong views on the DoJ suit to block it. Let me present my views in as organized a way as I can.
The first comment I’d make is that it seems to me (and many others) that the DoJ’s case is built around the wireless LAN space, and frankly I think that anybody who thinks that modern companies would build an M&A around WLAN is simply crazy. If you want to believe that WiFi isn’t a highly competitive space, with or without the HPE/Juniper deal, I can’t stop you, but I’m sorry to say you’re not thinking this through. Networking is a vast space, and wireless LAN is a pimple on it. In all my years of working with enterprises, talking with hundreds of them every year, I have never had a one tell me that their network strategies were based on wireless LAN. It’s not whether the merger is or isn’t justified based on the wireless LAN space, but that space isn’t what DoJ or regulators in general should even be worrying about.
Second, neither HPE nor Juniper are the leaders in that vast and real enterprise network market, Cisco is. The combination of HPE and Juniper builds a better competitor against the market-leading Cisco, which is surely pro-competitive rather than anti-competitive. Cisco is primarily a network equipment company that also dabbles in servers (UCS) and software (Webex). HPE, absent the merger, is a server company that dabbles in networking. To me, it’s obvious that HPE is building up its network side with the proposed merger, aiming at the broad opportunity in networking and defending against the risk that Cisco would expand its interest in servers and software, which I contend is just what Cisco’s recent reorg suggests it might do.
Third, returning to Cisco’s reorg, is that not a pretty clear signal that networking is commoditizing? Show me a commodity market that doesn’t generate M&A pressure? When margins are low and differentiation is hard to come by, companies look to M&A to create more efficient operation. It’s not something regulators should want to stamp out; if anything it should be encouraged to assure that a market doesn’t stagnate because the meager profits are spread among too many players to allow for R&D and innovation.
Fourth, Mist AI may have its roots in wireless LAN, but it’s now been subsumed into Juniper’s AI-Native Networking and AIops positioning. Anyone, does anyone believe (besides, apparently, some in the DoJ) that AI’s network value proposition is wireless-LAN-specific? Staying with Cisco, isn’t AI a big focus in their reorg? And if there’s pressure on capital spending on network equipment (which we all know to be the case) then doesn’t it make sense to assume that AIops would infiltrate netops? If that’s the case, then how does HPE’s ability to leverage Juniper wireless LAN AI give it any unfair advantage in the market? Who in networking isn’t expecting or already deploying AI in operations?
Fifth, commoditization in network equipment is furthered by generic switching chips, a market in which Broadcom dominates. Broadcom’s merger with VMware was approved, and this gives the company a key position in both networking and servers (via its VMware software). I’d argue that “white-box” devices are the real competitors in the networking space and that they effectively prevent anyone from exercising monopoly power in the market.
Sixth, the real issue in enterprise network equipment, and the only issue that really matters, is data center networking. For decades, this space has been influenced largely by the data center side, meaning in the end the policies on application software and enterprise data deployment. The greatest strategic influence in that space is IBM, who doesn’t even have network products these days. Could the HPE deal with Juniper make it more competitive in the data center? Sure; HPE is number two or three (Oracle and HPE juggle for second) in the data center strategic influence race. But who does M&A to make themselves less competitive? Does HPE/Juniper make that market less competitive? No; it makes it more competitive.
I think DoJ is suffering from a combination of being locked in the past and the classic forest-for-the-trees problem. Wireless LAN isn’t a key market; it’s been a decade since it has been. In any event, even the data center network market is more influenced by outside factors, as IBM’s influence shows, than by the juggling of equipment issues or competitors.
The bigger picture, though, is that both networking and IT are in need of a new set of business cases to drive up both spending and innovation, and I’ve long said that the biggest barrier to getting these business cases is their breadth of impact. Cisco, as I’ve noted before, was arguably launched by a single product—a NIC for a DEC minicomputer. The days when a tech revolution could be kicked off by something that simple and localized are gone forever. What we need now is a parallel push across many different technologies, and that’s not something that a marketplace that specializes in pieces of this and that will easily create.
We have systemic challenges in IT today, not little areas that when fixed relieve market pressures on a broad front. Only systemic players can meet these challenges, so for the DoJ to cripple a merger that creates one is a very wrong move, one that protects nobody and risks everything. The suit to block the merger, in my view, has no justification and makes no sense when examined through the lens of market reality.
