I don't understand HPE: they bought Aruba, already had a strong wired switches division. They also bought Cray, which obviously has their own advanced networking expertise. And now Juniper?
It is not like they lack in-house talent, they have it!
HPE consistently struggles to retain talent and foster growth post-acquisition. Integration is hampered by decades of IT cruft, leading to a sluggish tech stack and a deeply bureaucratic culture. Like many public companies, they also suffer from a lack of long-term focus.
My employer used to have some HP switches in one of our campus offices when we were smaller, all of that gear has, to my knowledge, been pulled out and replaced with Cisco gear over the last 6 or so years. I really wouldn't call them "strong" by any means, the SMB niche they historically filled is filled with vendors competing for smaller business networks (< 5-10K sqft offices that need some fairly basic managed switches, their IT gear can probably fit in less than 24U of rack space.) Cisco SMB, Dell Force10, Netgear, Chinese brands like TP-Link, and HP's Aruba line all handle these deployments without much issue; meanwhile, large enterprise and carrier-grade deployments are almost entirely Cisco, Juniper, Arista, and some specialized players like Nokia, despite HPE's best efforts to enter this space.
I trust HP/HPE as far as I can throw them to not botch another acquisition, because regardless of any talent they have (or acquire), their management repeatedly manages to torpedo it. Out of any large player in the networking space, Juniper's gear is such a breath of fresh air because of their long-term focus on software, JunOS and supporting products are what sells their hardware instead of the other way around, so here's hoping they don't find a way to fuck it up (don't hold your breath).
They’ve been trying to compete in the controller-less WiFi market for a while and it’s not going well. Tons of larger (higher ed, which have complex environments with very many end points) campuses have tried to move to Aruba Central, and subsequently moved back because it isn’t working well.
And there’s huge operational benefits to Mist’s architecture even if you leave out the AI part. My best guess for why Aruba hasn’t been able to replicate at least the architecture is that they’re carrying around too much legacy they can’t break with to make this feasible.
Since they’re nearly explicitly saying “we are buying Juniper for Mist”, and at least in my opinion aren’t doing that only for further developing that portfolio towards the data center, I guess they’ve just decided to buy someone who was able to execute on that architecture shift.
Aruba Networks (acquired by HP in 2015, right before HP split into HP/HPE) has approximately 5% of the market for switching outside of the datacenter segment. They aren't a big player in the DC, you're right about that.
Google used to use HP (pre Aruba) switches in the DC in the early days. Not many though. Mostly for OOB access. I doubt they still do.
It’s interesting that this requires HPE to divest from Instant On, which is their SMB brand. It’s unclear how that alone maintains a competitive landscape for enterprise customers that largely would have chosen Aruba’s more traditional offering.
That said, it’ll be interesting to see what use cases for data centers they come up with. Mist is one of very few infrastructure providers I’ve seen where AI/ML features actually improve operations. Surprisingly often in ways that I’m puzzled competitors have not been able to duplicate: something as simple as sending all access point logs through outlier detection and automated triage to proactively send RMAs for broken hardware before the customer even complains.
I’m also puzzled by the Instant On divestiture. In the original filing and press release the DOJ does call out SMB, even though most of the discussion is about enterprise markets:
My guess is that by forcing HPE to spin off Instant On, the DOJ hopes to score a modest PR win by claiming it protected small businesses, by preserving one of the few turnkey, cloud-native wireless solutions for SMBs. Since Instant On represents only a tiny portion of the deal, though, it seems odd that it was singled out.
I don't understand HPE: they bought Aruba, already had a strong wired switches division. They also bought Cray, which obviously has their own advanced networking expertise. And now Juniper?
It is not like they lack in-house talent, they have it!
Acquisitions have been central to Hewlett-Packard’s strategy since the mid-1980s. Just look at this list:
https://en.wikipedia.org/wiki/List_of_acquisitions_by_Hewlet...
HPE consistently struggles to retain talent and foster growth post-acquisition. Integration is hampered by decades of IT cruft, leading to a sluggish tech stack and a deeply bureaucratic culture. Like many public companies, they also suffer from a lack of long-term focus.
My employer used to have some HP switches in one of our campus offices when we were smaller, all of that gear has, to my knowledge, been pulled out and replaced with Cisco gear over the last 6 or so years. I really wouldn't call them "strong" by any means, the SMB niche they historically filled is filled with vendors competing for smaller business networks (< 5-10K sqft offices that need some fairly basic managed switches, their IT gear can probably fit in less than 24U of rack space.) Cisco SMB, Dell Force10, Netgear, Chinese brands like TP-Link, and HP's Aruba line all handle these deployments without much issue; meanwhile, large enterprise and carrier-grade deployments are almost entirely Cisco, Juniper, Arista, and some specialized players like Nokia, despite HPE's best efforts to enter this space.
I trust HP/HPE as far as I can throw them to not botch another acquisition, because regardless of any talent they have (or acquire), their management repeatedly manages to torpedo it. Out of any large player in the networking space, Juniper's gear is such a breath of fresh air because of their long-term focus on software, JunOS and supporting products are what sells their hardware instead of the other way around, so here's hoping they don't find a way to fuck it up (don't hold your breath).
They’ve been trying to compete in the controller-less WiFi market for a while and it’s not going well. Tons of larger (higher ed, which have complex environments with very many end points) campuses have tried to move to Aruba Central, and subsequently moved back because it isn’t working well.
And there’s huge operational benefits to Mist’s architecture even if you leave out the AI part. My best guess for why Aruba hasn’t been able to replicate at least the architecture is that they’re carrying around too much legacy they can’t break with to make this feasible.
Since they’re nearly explicitly saying “we are buying Juniper for Mist”, and at least in my opinion aren’t doing that only for further developing that portfolio towards the data center, I guess they’ve just decided to buy someone who was able to execute on that architecture shift.
Aruba probably declined after HPE bought it (just like every acquisition) so now they need another one to juice revenue.
Cray networking is HPC specific and isn't used in the campus or general data center markets.
> already had a strong wired switches division
Huh? They had basically 0 market share outside of SMB in the switching space.
Cisco, Arista, Juniper, Force10 (Dell).
Foundry was the bottom dollar alternative before being acquired by brocade, which Broadcom eventually ruined. And mellanox in the HPC space.
I’ve literally never run across HP switches in any enterprise datacenter, including shops that were “all in” with HP.
Aruba Networks (acquired by HP in 2015, right before HP split into HP/HPE) has approximately 5% of the market for switching outside of the datacenter segment. They aren't a big player in the DC, you're right about that.
Google used to use HP (pre Aruba) switches in the DC in the early days. Not many though. Mostly for OOB access. I doubt they still do.
https://www.businesswire.com/news/home/20250628599766/en/HPE...
Press release.
It’s interesting that this requires HPE to divest from Instant On, which is their SMB brand. It’s unclear how that alone maintains a competitive landscape for enterprise customers that largely would have chosen Aruba’s more traditional offering.
That said, it’ll be interesting to see what use cases for data centers they come up with. Mist is one of very few infrastructure providers I’ve seen where AI/ML features actually improve operations. Surprisingly often in ways that I’m puzzled competitors have not been able to duplicate: something as simple as sending all access point logs through outlier detection and automated triage to proactively send RMAs for broken hardware before the customer even complains.
I’m also puzzled by the Instant On divestiture. In the original filing and press release the DOJ does call out SMB, even though most of the discussion is about enterprise markets:
https://www.justice.gov/opa/pr/justice-department-sues-block...
My guess is that by forcing HPE to spin off Instant On, the DOJ hopes to score a modest PR win by claiming it protected small businesses, by preserving one of the few turnkey, cloud-native wireless solutions for SMBs. Since Instant On represents only a tiny portion of the deal, though, it seems odd that it was singled out.
https://archive.ph/QAHjL