5 Comments

Summary:

Oracle had another tough quarter in the hardware world but, hey, it could have been worse.

Oracle Exalogic Exadata

In my continuing obsession with Oracle’s hardware business, I took a peek at the software giant’s second quarter earnings. And the hardware portion wasn’t pretty.

Total hardware system revenue, including support, was unchanged year over year, at $1.3 billion; revenue from hardware products alone was down 3 percent to $714 million year over year. To be fair, Oracle execs warned of this in their first quarter call, in September. At that time, Co-president Safra Catz said that for the second quarter: “Hardware product revenue could range from negative 9 percent to positive 1 percent in constant dollars and negative 11 percent to negative 1 percent in reported dollars.”

oracleq2

Given that context things weren’t so bad. In a statement issued  in advance of the earnings call Wednesday, Catz’s Co-president Mark Hurd said the company’s hardware business — including support — actually grew 2 percent in constant currency, a measure that strips out the impact of currency fluctuations on a company’s results. Hurd also cited “double-digit revenue growth” in the various Exa-boxes (Exadata, Exalogic et al.) The SPARC SuperCluster and Oracle’s Big Data Appliance upped the ante, he added, showing “triple-digit growth.”

Still, it’s hard to believe that anyone in Oracle’s executive suite is pleased with these numbers. While Oracle always maintains that it makes good profit on the boxes it does sell, its market share  continues to erode. In IDC’s latest numbers — for the second calendar quarter of 2013 —  Oracle server revenue was off 16 percent year over year. (Silver lining from Oracle’s point of view, IBM’s number was off 19.4 percent.) What may be most worrisome to all the legacy hardware guys is that the no-name white-box vendors in aggregate saw revenue soar a whopping 45.2 percent. Cisco was also up 42.8 percent, but that’s growth off a small (and new) base.

Before the naysayers jump on me for letting IBM off the hook, just remember, IBM has been in the hardware business since Hector was a pup, but Oracle opted to buy into this really tough business by plunking more than $7 billion for Sun Microsystems just a few years ago.

The white box numbers reflect that big web-scale companies are opting more for commodity, inexpensive servers as opposed to high-margin scale up servers (like Oracle’s exa-boxes and HP’s Moonshot servers, etc.) Granted some enterprise workloads that will flow to those scale-up servers, but it’s unclear to me how much of that work there will be going forward.

idc q2 server

  1. Oracle has turned around a failing business (Sun) by getting out of non profitable, commodity (x86) server business, and low cost/high volume SPARC servers and is now focusing on where it can compete, make margin, and offer value, such as with its Engineered Systems (based on both x86 and SPARC) as well as its new SPARC based SPARC T5 and SPARC M6 servers. 



    Revenue gains DOES NOT MEAN success if you can’t make margin from it. Just look at all the commodity based vendors like HP, Dell, IBM, Cisco and even Intel who are all suffering with their latest financials yet they all claim to have leading marketshare and signs of server sales growth?! 



    Back in the Sun days, yes, there was a lot more revenue, but Sun couldn’t make any reasonable margin on that revenue and therefore was a losing business. Without making reasonable margin (profit), theres no way to continue investing in R&D and that’s a losing proposition as seen by Sun’s eventual decline and business nosedive.

    Its taken Oracle 3 years to turn around the Sun HW business, noting that it usually takes 2-3 years just to develop new CPUs (just ask Intel why the next Xeon-EX has taken ~3 years) and new systems based on them.

    So as you can see from latest earnings report, Oracle HW is at an inflection point, where finally, Oracle has declared, for the first time, stabilization, and even growth in its hardware business, and that Engineered Systems now makes up almost 30% of product sales.

    And for the first time ever, Oracle is stating in its guidance, that “Hardware product revenue growth is expected to range from a negative 1% to a positive 9% in constant dollars and negative 2% to a positive 8% in reported dollars”, showing that more growth is expected now.



    And fyi, If you were to look at the latest Gartner or IDC server sales statistics, you will see that Oracle is showing positive marketshare gains in the markets, and price points that it sells to, which is predominantly the $100K-$500K price band (SPARC T5 and SPARC M6 mid range servers). According to the analysts, Oracle’s share was ~19.8% in CY13Q3, compared to 13.9% in the year ago period. Oracle’s growth depending on which analyst you look at, is from 23.7% to 48.6% y/y, recording strong sales of SPARC T5 and SPARC M6 servers that just started shipping only 6 months ago. IBM meanwhile shrank by 10.9% Y/Y in same price band and HP dropped a whopping -44.3%.

    

And whats interesting, is that while “white-box” vendors grew in revenue & marketshare, who are these vendors and how successful are their businesses? From what I understand, most are Taiwanese, Korean or Chinese vendors, all grouped up in the “others” category. Why others? Good question! Should the tier 1 vendors be worried? yes, if they are competing against these whitebox vendors with the same razor thin margins. From what I can see, it’s a losing proposition that is hard to survive on long term. Just look at how long these companies have existed and you’ll see what I mean!

    White box vendors come and go like the wind.  Do Enterprises really want to make investments in these companies products and the long term risks associated? I wouldn’t. The perception of buying “lower cost” servers up front is foolish when you look at the longer term costs of managing and maintaining these systems . After all, most companies have an IT OPEX problem today, and buying loads of cheap whitebox servers doesn’t reduce OPEX, it increases it. Just ask the Tier 1 analysts.

    Share
  2. Oracle last quarter results were not up to the mark of the analysts. http://bit.ly/OracleAnalysts

    Share
  3. Phil,

    You work for Oracle in their Business Development/Competitive analyst and pre-sales support, which explains the rose colored glasses.

    The reality is that SPARC and UNIX are done. Look at the latest IDC and Gartner numbers, or better yet look at who has eroded the SPARC business. Cloud/Linux based systems, Dell and Cisco were the two emerging competitors that were eating SPARC sales, which utterly shocked management at Oracle. The reality is that Oracle’s daily IT, their Cloud and their engineered systems are all on x86 systems, although the Engineering execs (all legacy Sun) are desperately trying to hold onto SPARC.

    You also left off key other facts, Oracle spent billions on the development of the new SPARC chip, and based on the massive free-fall of SPARC sales, plus the millions spent on marketing and sales pushes, I’d say Oracle has a negative return on the new SPARC systems vs 90% of their hardware revenue coming from the x86 based Engineered systems.

    You also showed how little you know of the overall market, which is not uncommon for Sun-Lifers. Intel’s EX refresh took longer than anticipated because their E5-2600 v3 refresh which makes up the vast majority of the entire x86 market (2P Systems) is their priority. 4 socket systems are slowing, and anything above that has been eroded by 2P clusters…see your SPARC sales as proof. So its not that Intel took forever to launch EX, its that it wasnt a priority. And yes, Oracle has gained market share in the very high-end, except the TAM as a whole is virtually flat, so again its Oracle creatively marketing “success”, and it was almost entirely Exadata.

    While revenue isnt the only metric, one could compare it to being present at work. Does NOT being physically present at work mean you’re not working, does being physically present at work mean you are? No and no, but at least if you’re there, no one can question…and the same goes for Oracle’s hardware business. Declining revenue quarter, after quarter, after quarter doesnt mean a lack of business solvency, however it certainly does mean there are other issues behind closed doors, and I saw these issues with my own eyes.

    Keep telling yourself and your fewer and fewer customers all is ok, as big customer after big customer replaces their legacy SPARC systems with Rackspace Open Stack solutions (which is why IBM bought Softlayer). The reality is that customers can get Oracle software as a service cheaper from other cloud providers, as they can also bundle in hardware that doesnt lock them into a closed IT stack (IaaS, SIaaS). Secondly, your service sales revenues are predicted on volume, which continues to go down. Thirdly, of the top SaaS platforms Oracle Eloqua was the one of two software solutions that Oracle had in the top sellers, and Eloqua was an acquisition, not an organically developed solution.

    The truth is that Oracle missed the market winds of change with their closed, proprietary and non-customer centric business model, and the only silver linings are – Exadata and the ability to acquire firms like Xsigo/Eloqua/ACME Packet to bolster the bottom line. I guess having a lot of cash can make up for a poor roadmap and business strategy, so hey, you got that going for you, and that means you can keep that huge legacy Sun 401K growing while watching the water flooding the boat…congrats! Thank Larry for bailing Sun out!

    Share
    1. Well Raul, you have some good arguments and although I agree with some of them, others don’t make sense or just wrong. Heres my POV.

      “You work for Oracle in their Business Development/Competitive analyst and pre-sales support, which explains the rose colored glasses.”

      Yeah sure, you know who I am and that’s great. And if you look further into my background, you’ll see that Ive been focused on competitive analysis for the last 10+ years and have been tracking the IT industry for over 20 years. Not sure where you are coming from though…

      “The reality is that SPARC and UNIX are done. Look at the latest IDC and Gartner numbers, or better yet look at who has eroded the SPARC business. Cloud/Linux based systems, Dell and Cisco were the two emerging competitors that were eating SPARC sales, which utterly shocked management at Oracle. The reality is that Oracle’s daily IT, their Cloud and their engineered systems are all on x86 systems, although the Engineering execs (all legacy Sun) are desperately trying to hold onto SPARC.”

      Let me re-phrase-SPARC was done before Oracle took over and its taken Oracle ~3 years to turn the architecture & business around. 5 SPARC generations later, SPARC is actually leading in all key metrics from performance, scalability, throughput, response times and price/performance. You can search all the published benchmarks for proof points. Theres actually no processor on market today that can beat SPARC on these metrics. Sure, there are cheaper CPUs out there, but cheaper doesn’t mean better-especially in the enterprise, where SPARC is focused. But this is all very recent so clearly takes time to shift markets.
      And furthermore, its not just Oracle investing in SPARC, so is Fujitsu. If SPARC was dead, why would both companies be investing so heavily?

      http://www.oracle.com/us/products/servers-storage/servers/sparc-enterprise/public-sparc-roadmap-421264.pdf

      http://www.fujitsu.com/global/services/computing/server/sparc/key-reports/featurestory/fs-20130806.html

      And actually, in the last few quarters, now that Oracle has released its most competitive SPARC systems to date (SPARC T5 and SPARC M6), and Fujitsu with the SPARC M10, both IDC and Gartner are showing marketshare gains for SPARC that are in the double-digits. You can even see in latest Oracle earnings reports that hardware revenues are showing growth for first time since the acquisition. Look at IBM’s dismal hardware earnings these past 3-4 qtrs and its mostly due to Oracle gains.

      And regarding Oracles Engineered systems, you need to realize that Oracle was shipping Exadata x86 before Oracle acquired Sun so theres more history of x86 @ Oracle than SPARC, However today, with SPARC T5 and SPARC M6 being released just 6 months ago, Oracle is reporting a shift to SPARC based engineered systems. Oracle SuperCluster, which is Exadata and Exalogic combined, saw 100% growth these last 2 quarters. And while Oracle will always offer choice of architectures, Oracle clearly can’t depend on Intel continuing to deliver competitive Xeon CPUs when they are getting hammered by AMD and mobile market, so clearly Intel has to switch its focus to mobile and so defocusing on other businesses. Just look at whats happened to Itanium. Its dead. Look at 4-socket and 8-socket Xeon, its late-by like a year. Look at the E5-v2 generation and its basically more cores/die with slight GHZ bump-nothing major. Problem is that with SW that’s licensed per core, Intel is going in wrong direction with worse per/core performance. Of course Oracle, Microsoft, IBM and anyone licensing per core are happy, but customers clearly aren’t. Performance/license matters.

      “You also left off key other facts, Oracle spent billions on the development of the new SPARC chip, and based on the massive free-fall of SPARC sales, plus the millions spent on marketing and sales pushes, I’d say Oracle has a negative return on the new SPARC systems vs 90% of their hardware revenue coming from the x86 based Engineered systems.”

      Heres where I agree (on first part). Yes, Oracle has invested over $1BN a year in turning around SPARC/Solaris and now 3 years later, is showing success, leadership and positive growth. If Oracle were to have a negative ROI growth, it would show in the financial reports and it doesn’t. Oracle continues to show revenue gains Qtr after Qtr and its actually Oracles HW business that’s driving a lot of software sales, as software updates sales has increased. Heres a great article discussing this: http://seekingalpha.com/article/1722832-oracles-sun-acquisition-is-paying-off-in-spades

      ” Intel’s EX refresh took longer than anticipated because their E5-2600 v3 refresh which makes up the vast majority of the entire x86 market (2P Systems) is their priority. 4 socket systems are slowing, and anything above that has been eroded by 2P clusters…see your SPARC sales as proof. So its not that Intel took forever to launch EX, its that it wasn’t a priority. And yes, Oracle has gained market share in the very high-end, except the TAM as a whole is virtually flat, so again its Oracle creatively marketing “success”, and it was almost entirely Exadata.”

      You mean E5-2600 v2? Why is Intel only focusing on 2P systems? Because its easier to design, develop and test and scaling to 4 & 8-sockets is a lot more complex and costly (for them). The problem is that with only 2-socket being price/performance competitive, it may drive up more sales volumes, but its creating other problems, OPEX problems, and the key analysts have discovered this over the last few years. A cluster of 2-socket systems is more costly to manage, more SW licensing and more to maintain that a larger system, due to more cabling, networking, additional SW, etc. Look at what IDC’s been saying:

      http://www.oracle.com/us/products/servers-storage/servers/sparc-enterprise/idc-dce-2012-1612359.pdf

      “Another step IT organizations should consider to save on management and administration costs is represented in Figure 1. Without a reduction in those costs, which have grown dramatically in recent years, the growth in operational costs (opex) will continue to outpace the gains (savings) made in reducing capex for acquired systems.”

      And plus, look at the health of all the x86 vendors. They’re all reporting massive revenue & margin declines these last 2+ yrs. Selling commodity eventually “commoditizes” the vendor. Dell had to go private, HP is suffering bad, IBM is trying to sell off its x86 business and Cisco has been reporting losses too. Even Intel has been hurting these last several quarters. And where is AMD these days? Ouch!

      “The truth is that Oracle missed the market winds of change with their closed, proprietary and non-customer centric business model, and the only silver linings are – Exadata and the ability to acquire firms like Xsigo/Eloqua/ACME Packet to bolster the bottom line.”

      Well, this is completely false and clearly FUD. Wheres your proof/subitationation? Theres no products at Oracle that force lock-in, nor are proprietary and non-customer centric. Every “layer” of the Oracle stack of products, from Oracle DB, Oracle Linux, Solaris, SPARC, Oracle VM, Engineered Systems, etc, offer choice, flexibility and are even offered/supported/sold by multiple-vendors. Oracle DB runs on almost any platform out there, SPARC, an IEEE standard, is being developed and sold by Oracle and Fujitsu, Solaris runs on SPARC or any x86 system out there (theres even Solaris engineers at Intel), etc. So wheres the lock-in? Whats proprietary?

      What I believe is the real “lock-in” in the industry, especially in data centers, is “complexity”. From what Ive read, the analysts are claiming over 70% of IT budgets are going to “keeping the lights on”, meaning managing those racks and stacks of 2-socket servers running open source. Sure it was low cost to acquire but managing all the layers, all the vendors, etc is a nightmare in managing over its lifecycle. Why do you think Cloud is taking off so fast? Because companies are looking at ways to reduce OPEX and shift the costs into the cloud, and the cloud provider. Problem is, I don’t believe will solve the problem as folks will still have to manage whats going into/out of the cloud. Complexity needs to be reduced.

      So the silver lining at Oracle is that today, Oracle offers a complete end to end SW & HW stack to run IT and just like Apple has their stack and why its been so successful with iPhone and complete integration of technologies from SW to HW (including controlling the OS and CPU developments), Oracle is focused on doing same for IT. Sure, theres Samsung competing with an alternative stack, but the elements are the same. You get it all from one company, designing to make it all work together. I believe that’s where the industry is headed and why Oracle is showing great growth with its engineered systems. You can see this in IDC’s latest report: http://www.idc.com/getdoc.jsp?containerId=prUS24367013

      Share
  4. Correction on text above: “AMD” should be “ARM” in “when they are getting hammered by AMD and mobile market”

    Share

Comments have been disabled for this post