HPE to acquire Juniper Networks - First Take, Thoughts, and Perspectives

Background : Juniper Networks has agreed to an offer by HPE. Here are some 360 degree thinking on why and what happens next.

https://blogs.juniper.net/en-us/driven-by-experience/juniper-networks-to-combine-with-hpe-accelerating-ai-native-networking-leadership

https://newsroom.juniper.net/news/news-details/2024/HPE-to-Acquire-Juniper-Networks-to-Accelerate-AI-Driven-Innovation/

Note that the deal will close in late 2024 but looks to be certain at this stage.

Why ?

In my view, HPE buying Juniper Networks primarily to position for Telco/Mobileco business. Paragon for core network operations plus JunOS hardware & security can be a turnkey subscription service aka Emerald/Greenlake. Think of all those 5G POP/edge/OpenRAN deployments that will never be ‘in the cloud‘. The high end routers, security, and network orchestation complete the existing HPE telco portfolio of servers and openRAN/OSS/BSS software.

I believe that the Greenlake subscription business is the core focus here. Why ? Modern on-prem cloud stacks are hideously complex and customers don't want to spend money on headcount to perform the integration. Costs are nearly impossible to manage and "DevOps" means tripling/quadrupling IT headcount by hiring top notch experts and paying them accordingly. Finding managers who can run those teams is roughly equivalent to finding rocking horse poop. Also boomer execs can't stand having people smarter than themeselves in critical roles.

AI. Its an AI Deal. Not Really.

Its not really about AI but it will be pitched to shareholders that way. So much coolness and fashion!!! The MIST AI campus platform is amazing and has been winning deals everywhere, its now on a 1Billion run rate. I believe MIST can grow to be more central to IT operations but mostly its an amazing service that can be added to Emerald/Greenlake so that customers get better operations but also HPE gets to keep costs down on those bundles.

OpenRAN Bundles

HPE has worked hard to grow its Telco/Mobileco with limited success. Flogging servers to telcos with some small part of OSS/BSS stack software misses most of the revenue. There is a lot of money in OpenRAN coming, literal billions of captial spending as 5G rolls out over the next decade. Recent SP/Telco/Mobileco deals suggest full stack bundles from sole suppliers is the fashion e.g. Ericsson just won $15B/10year deal at ATT as primary integrator. HPE & JNPR are suppliers to that deal.

The Business

I believe that HPE really sees the service providers as growth market with subscription bundles (aka greenlake/emerald). It needs to do that alone to capture the profits, instead of 'sharing' accounts with Cisco, Ericsson, CIENA, or Nokia.

HPE knows that they are missing deals, especially against Dell, because Aruba Networking is mostly campus/branch. Attempts to grow into the Enterprise DC with Pensando failed (but good for SP/telco market!) and developing DC switches has been slow - I don't think Aruba invested enough and it was seen internally as a not-very-important. Networking is now important.

Enterprise

Juniper Service Provider hardware is unique while Enterprise has overlaps but not total. Mist can drive Aruba Campus (Wired/Wireless) and SDWAN/Branch, Apstra offers SDN for DC/POP/RAN.

HPE needs market leading enterprise DC and WAN networking to compete its Enterprise portfolio. Obviously there could be rationalisation of enterprise hardware - too many overlapping products but much of the Juniper enterprise hardware isn’t unique, it’s OEM. The enterprise value is in Apstra/Mist SDN tooling which is selling very well and winning against everyone.

For enterprise, you can use third party silicon from Broadcom, Marvell, Centec, Intel - its doesn't really matter in business terms. The main focus is the software operation!

CyberSecurity!!

HPE finally gets a serious security portfolio with JNPR SASE/NGFW etc for both SP and Enterprise instead of reselling/partners. HPE should have had security 20 years ago instead of reselling someone else's gear. Thats an especially big deal for their Aruba SDWAN/SASE product which is underinvested and late.

I would highlight that Juniper SRX is "telco grade" while Aruba's existing SASE more a mid-market solution.

Patents

Also, patents. Juniper has a lot of patents and this will reduce internal accounting cost by cross-licensing of portfolio.

What Dies, What Lives

For me, its possible that Aruba becomes a Meraki-style mid-market fully managed solution while Juniper positions as premium enterprose product. Arista and Cisco has proven that model works and could be worth copying. Depends on available investment, cost cutting needs and ofc volume and profitability.

I have questions whether HPE can sustain and spend on ASIC development but it might not matter in five years. Broadcom Jericho and Marvell Prestera are really serious chips. Maybe Intel finally gets a router ASIC out the door.Also JNPR has silicon photonics for IPoDWDM in its portfolio.

Podcast

We will be discussing this in detail on the Network Break podcast next week (15 Jan 2024) if you want more. https://packetpushers.net/podcast/network-break/

Services

With mountains (literally multi-billions/yr) of guaranteed subscription revenue, vendors can hire their own engineers at scale to manage and operate the bundled infrastructure offerings. I predict that customers are returning to the vendor services model of old to get away from owning and operating their infrastructure systems. This pattern will last for a decade or two, before the vendors lose the plot/overcharge/perform poorly or another technology transition comes through to change the model.

Resellers don't make enough from professional services, won't spent on training and won't retain good staff. At best, resellers will be adjunct to vendors while they find niches to exist in.

Evidence ? Look at Oracle, SAP et al for their service model, product portfolio and profitability. On-prem integration and operation is fading quickly.

Conclusion

• This is a good deal for both HPE and JNPR shareholders. The price is reasonable and HPE can leverage the products.
• Juniper Networks has great technology but struggled with sales and marketing. HPE sales and subscription bundles are a good fit.
• HPE should get more traction in the 5G/Telco/POP/Edge/OpenRAN megadeal markets having with world class telco networking combined with its servers/storage/HCI and OSS/BSS products and partners.