What do people see as the future around Oracle hardware and systems business? Will they get out of selling to customers and only build hardware for their cloud infrastructure? Or will they exit entirely and second source hardware for the cloud infrastructure? Will it be Intel, AMD, and/or ARM? Timeframe?
20 replies (most recent on top)
People can criticize the acquisition all they want but Oracle made more money than the Sun acquisition cost them.
1mzo : I'm not in LE's mind, so I cannot know why he bought Sun, offering more than what IBM was offering.
what I know though is that Sun in 2009 (the year before the acquisition) had 3B in cash with more than 11B in total assets, and the net revenues where north of 11B. not a bad deal for a 7.4B investment.
from 2010 to 1016 Oracle has invested more or less 1B/y in R&D only to maintain a planned YoY sales decrease, basically all the cpu presented during 2010-2016 where already in Sun's roadmap (I still have a 2009 roadmap showing those cpu until 2019). the only "new thing" was Supercluster, a total failure from a selling pov, and just a bunch of pieces without any "secret sauce" but with a huge amount of marketing.
moreover, Oracle has heavily decimated sales and presales orgs, and then finally closed the eng org. I can expect the total operating cost in this period to be even less than what Sun was facing.
on top of that, Sun had some potentially interesting assets, like Java, to be monetized. but Oracle has chosen the wrongest way, and actually did not pan out. same as SunCloud, the very first cloud IaaS offering in the world, even some months before the first AWS service. Oracle killed SunCloud immediately because LE thought that it was just another stupid thing done by Sun like Java, like multithreading, like multicore, like vertical scalable OS, like OS containers, like HW virtualization, etc..
so, again I'm not in LE's mind, but I think Sun aquisition has been just another financial engineering move from LE+SC in which Oracle has gained some money.
Oracle was and is not interested at all in selling HW, be it SPARC or x86.
and I'm quite sure that Oracle will not sell what's remaining of Sun, it will just be closed in FY21, and absorbed as just another coretech org dept.
it was all about monetizing Java, including some potential high-profile litigation.
Larry gambled $7B on that potential and lost.
The Sun acquisition had nothing to do with cloud or IaaS, it was all about monetizing Java, including some potential high-profile litigation. Sun actually had a emerging cloud buildout called SunCloud but Oracle cancelled it. Remember that in late 2009 when the Sun acquisition was announced, LE was still on record saying the cloud trend is a joke.
No need to "outsource" anything to Dell, HP, Supermicro, etc. or to have hardware on the price books anymore. Just qualify a few third-party systems periodically for those customers that want on-prem equipment. And gladly take kickbacks from hardware vendors wanting to get on the list.
Could oracle have tried to build cloud offerings like IaaS, without the Sun acquisition? Not worth the money given the paltry results, but could that have been the thought behind it?
So tell me what did Oracle actually gain by spending $7B to buy Sun?
HW business: on its last legs, as you point out.
SW business: most of Sun's portfolio was freeware - no revenue in it and one of the reasons Sun was itself going bankrupt. Java support business? Can't see it being a big factor long term, especially with developers moving away from it.
The only other thing I can think of is their hopes for a big settlement from the lawsuit against Google. We all know that was a dud.
@WbLPqiB-1qir No, they know that. They leverage those $ from hardware revenue to hire young kids from AWS with 3x salaries.
Some stupid kids and managers from acquired companies don't even understand what is hardware. They don't even understand what is atomic operation or endianness etc. They all think those OS and hardware are done deal to them. They have been playing LEGO since they were a 5 year old kid, and still doing that.
1vod : my dear friend, if you want to say something so stupid, you should at least check it is not so easily figured out.......
this is Q1FY19
https://investor.oracle.com/financial-news/financial-news-details/2018/Q1-FY19-GAAP-EPS-UP-13-TO-057-and-NON-GAAP-EPS-UP-18-TO-071/default.aspx
more than 900M$ only in HW without services:, 10% of total revenues, more than on-prem license that is at 9% ... and Q1 is not known to be the best quarter.....
and this is the consolidated result for FY18:
https://investor.oracle.com/financial-news/financial-news-details/2018/Q4-FY18-GAAP-EPS-UP-8-TO-082-and-NON-GAAP-EPS-UP-11-TO-099/default.aspx
3.99B$ in HW without services
I honestly don't know where you got that 500M$ figure, but it's definitely, completely, totally wrong.
There is no $4B in hardware sales at Oracle, that is just fantasy. They do less than $500m each year.
So what it the timeline for outsourcing x86 development to Dell? I saw rumors on another thread that this could begin as early as CY 2019.
It's too late to keep any of the Sun systems group (SPARC, platform, and Solaris), they're all gone. Oracle cheerleaders will still tell you that they've got thousands working on Solaris but those numbers are contrived and dishonest. You can't build and feed an ecosystem in this century without a strong open-source foundation. Red Hat got that right and the payoff was $34B. Oracle went the other way with their 1990's model and there are now tumbleweeds blowing across their Santa Clara campus, between the FOR LEASE signs.
Oracle should keep its own hardware design team, as well as OS. Sparc and Solaris are well engineered system. Oracle should feed those products with eco-system to survive. Especially, the RedHat is in IBM's claw now. Other ambiguity products which never ever earn 1$ for oracle should go.
Oracle will continue to have a business supporting Solaris for another decade or two, so why would they open source anything? SC may not need her Solaris support people in Romania, India, Nigeria, and Papua New Guinea by 2030 because AI and supportbots from the Oracle Cloud may be ready by then.
Future, oracle, really? Admire OP’s sense of humor.
So they bought Sun and it’s assets including the brain trust and Solaris, customers still want/need the high end Unix for certain workloads or at least the hardware. Yet it will be sunsetted. Tossed in the trash bin of technology. Oracle is such as d--k move company they would probably hold stuff back and not open source it.
They should sell the HW and Solaris to Fujitsu. If they don’t want to sell it with Solaris then open source Solaris.
HW business is stil at an healthy (i.e. highest margin on the market) 4B$/year, more or less 2B$ for x86 (almost all internal sales to OCI , including ExaCS and ExaCC) + Exa onprem and 2B$ for SPARC (almost all servers, the dmw Supercluster and Minicluster have always been marginal at best). Storage onprem (i.e. ZFS) and Tape are almost irrelevant.
Given what management has decided, SPARC business will evaporate during this FY19, expect only marginal sales in FY20 (small IB refresh in those customers with longer transition time to x86), and it is something done on purpose to steer customers to cloud or more likely Exa (in all regions but US it is highly unlikely that a customer running a SPARC IB will move directly to cloud).
The current 4B$ is expected to decrease to something like 3B$ in FY20 (almost all Exa, we expect to sell between 600M and 700M of SPARC) and the 1B$ lost is expected to generate cloud revenues in the order of 2B$ (wishful thinking at best, but hey I'm not genius like the 3 elders).
There will be a new reorg in the sales/presales field org, mainly in EMEA and APAC, in Q4FY19 to reflect the lower sales. Numbers are sill not there, but based on what Q3 wil be, should be something like a 50% reduction in SSO (not saying it will be a rif, in many cases SSO people will be moved in the ES&Cloud org), and more EMEA countries will no more have an SSO field org (full ODP org, like Spain, France and some others today).
ARM is just the new toy of the WC team, from an HW pov it is nothing more than a very far future possibility, there is no HW project actually running.
The HW strategy is very clear, and very well explained in AA management line since the reorg in 2017: Oracle is not interested in being yet another HW provider, we will sell only high margin x86 HW (Exa) using x86 systems (Intel or AMD) engineered for OCI, until we manage to outsource all HW development to a 3rd party (Dell or Fujitsu).
So on the mid term, the HW future is just Exa. On the long term, no future.
There is no external hardware business anymore except for maybe selling some remaining inventory. A few small internal groups are still around making customized HW for the cloud (X64, Pillar storage, etc.), that's it. They are dipping their toes in the ARM server waters but just slightly. There are some ARM fans in here that think the story is stronger than that but we'll see.