IBM's potential x86 server sale to Lenovo highlights oncoming train
Summary: IBM famously unloaded its PC business to Lenovo well before the post PC era hit. Is Big Blue bailing on commodity servers before the floor falls out?
IBM is reportedly talking to Lenovo about selling its x86-based server business to Lenovo and the move would make a lot of sense.
If the talks, flagged in the Wall Street Journal and CRN, sound familiar that's because Big Blue famously unloaded its PC business to Lenovo in a win-win deal. Lenovo went on to be one of the premier PC makers and IBM focused on software and services and got ahead of trends such as analytics.
To say the IBM's PC situation then and today's server state of affairs rhyme would be an understatement. You could argue the situations are the same thing. When IBM offloaded its PC unit, no one saw tablets coming. All IBM knew is that the margins stunk and it wanted higher value wares. The post-PC era was years away.
Also: Servers, datacenters to see same upheaval as PC industry | IBM's Q1 closes weak as software, mainframe deals slip | IBM reportedly in talks to sell server biz to Lenovo
Fast forward to the server market, which is ripe for disruption. Server sales are doing ok. Companies will have to buy servers right? Of course they will---for about another three to five years. The reality is servers are going in the following directions:
Specialization by workload. Think IBM's PureSystems and Oracle's Exadata efforts.
- Commodity-ville on the x86 front. You can't ignore that companies like Google and Facebook go right to white box makers for servers. That reality isn't so hot for HP, Dell and IBM.
- You need to own the silicon and intellectual property to really work the server business. IBM's Power systems won't go anywhere. Oracle has SPARC. Hewlett-Packard is going processor agnostic with Moonshot, a server line that appears to be innovative.
- Fewer server buyers. As companies move to the cloud, demand for compute will only increase. The problem. Server makers will be selling in bulk to fewer customers and cloud computing farms. There will only be so many cloud providers. Enterprises large enough to roll their own data centers will be few and far between.
Now let's talk timing here. The server market won't unravel tomorrow. It won't unravel in a few years. But Armageddon will occur and the clock starts ticking right about now.
Why? An enterprise that buys a server right now will start a tax depreciation clock that will run about three years. Once those three years are up and those assets depreciate, the CXO in charge will weigh the costs and benefits of the cloud vs. running a data center, server cluster or whatever. I'll bet that in three years the cloud will win by a wide margin. Let's face it---the cloud is already starting to win and all you have to do is show up at one of Amazon Web Services' customer powwows to know the writing is on the server rack.
On Thursday, I caught up with Cycle Computing CEO Jason Stowe. There's a lot to like about Cycle Computing. First, the company is bootstrapped so there's instant respect. Second, Cycle Computing is at the forefront of making high performance computing clusters for the masses. And third, Cycle Computing has top insurance and pharmaceutical companies as customers. Cycle Computing had massive customers from day one. In other words, Cycle Computing is the real deal, hooked up with Amazon Web Services and will enable a lot of science to happen just by democratizing HPC for smaller companies.
Stowe noted that Cycle Computing is starting to land manufacturing and engineering customers now for its HPC management software and cloud connections. In other words, this HPC for the masses is catching on. If you play this out, there will be fewer servers sold because folks will be using Rackspace, AWS or some other former hardware focused vendor.
Today, it's big data and research compute driving Cycle Computing demand. Tomorrow every company will have the mathematic models and horsepower to simulate just about anything. You won't buy your own servers for that computing power.
Stowe said servers will become like wheat fields not things you name. "Today servers are hugged, named and managers know their quirks. There's an attachment. In the future server clusters will be more like wheat fields. You grow the wheat, reap and sow, eat and replant the seeds. There's no attachment to the wheat," said Stowe.
In other words, Stowe's excellent analogy on servers and meeting compute demand translates into cloud farms and fields. Most companies are going to hit the brakes on new server buying as soon as the depreciation ends and new compute demand has to be met. Play this out and the profit margins on servers aren't going to look so hot.
IBM sees all of the servergeddon scenarios developing and that's why it's ditching its commodity server business now. Let Lenovo, which has the scale and ambition to do the commodity server game, carry the ball from here and duke it out with HP and Dell.
Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.
Talkback
Capital Expenditure vs Monthly Expense
Cloud computing changes how businesses acquire hardware and software and how it's charged on the balance sheet. Cloud computing is a monthly expense, like a lease, rather than a capital expenditure and associated depreciation of that asset. Like telecoms, computer hardware and software makers are moving from the buy a product to buy a service for that monthly income/revenue stream. Computing thus become like Cable TV, or your Cell Phone. You pay for it every month, rather than buying it once and using it for years.
Even BYOD is a move by businesses to reduce expenses by shifting the buying of end user equipment to users.
The real issue for Cloud Computing and BYOD is affordability by end users, small & medium sized businesses who's operating margins are thin. Will small and medium sized businesses and the workers able to afford an added monthly expense, and continue to be able to pay for that added monthly expense, and if not what happens to their "data" and services if they can't.
While the shift might go to the cloud, as the next "latest and greatest" technology, the realities of the monthly costs of it might shift businesses, especially small & medium sized businesses back to the old model of capital expenditure as they find that buy once and use for years is easier on the balance sheet than adding another monthly expense that they may or may not be able to pay for, and if not, their data and ability to conduct business, is hampered by the "Cloud" and BYOD.
The problem with cloud...
A lot of CIOs and CTOs, instead of looking at external providers are starting to look at doing cloud internally (private clouds). This way they get the advantage of cloud, but also get the advantage of being able to control their data. So, I don't see servers disappearing that quickly until cloud providers can truly show that they'll protect your data as well as you would protect your data (most IT professionals believe in having a copy of a copy of the data, ie, 3 places where the data is stored for recovery purposes).
So, the only businesses I truly see going to cloud providers are going to be small businesses (less than 25 employees), medium size and larger companies are going to likely go private cloud until the cloud providers get things fixed up and can give businesses the peace of mind that their data will be well taken care of and easily restored if something happens.
No cloud for us
...uh, yes you can
U.S. Citizen access or clouds with HIPAA in mind. I too am not a fan of
cloud-everything; at minimum I want to be able to hobble along if the cloud vendor
is down for a week, but the vendors are knocking down many objections
one by one and if you aren't paying attention that can make one look uninformed
and that is a bad position to be in during decision times.
All of the above
Truth
you need to read the fine print
True but, in the end, the collective cost of BYOD ...
On the other side of that, the employer may find that they incur a greater expense because they have a much wider range of differing devices to support.
Collective costs makes no difference
I am big iron IBM retired with many customer discussions comparing the cost of mid-range and mainframe servers. IBM wanted us to be hardware agnostic and I often compared expense to run networked PCs vs bigger boxes to serve similar function for users. The thresh hold was very low before big iron was more cost-effective, sometimes by a wide margin. I recall one case where the cost break point was six PCs compared to the old AS/400. With the seventh PC, the AS/400 was cheaper to support and the difference grew larger with every additional PC.
It rarely made a difference because the decision-maker was often outside the glass house. Consultants, investment bankers and, in one case, the children of the chairman had more say than the CIO.
If Clouds are cool and employees buy their own device, dollars and sense can go out the window.
AS/400
But is this the case for a Win, or *IX system?
Not an OS question
Clouds are after my time. However, I suspect there is a similar argument to be made. I just don't know what it is. ;-)
HP Moonshot "appears to be innovative"
Not saying much ...
HAHAHAHAHA No, nothing innovative or game changing.
IBM gave up on the low end x86 market years ago
For many years now, the product simply hasn't been available through distribution. We'd work hard to get a customer to issue a purchase order for a standard IBM server config and it would take us weeks to get the necessary components. Hard drives? Not available. Memory? Not available. RAID card? Maybe 6 weeks. And, despite promises of improvement, product availability only got worse.
When one of our techs mentioned that IBM xSeries support calls are now being answered in Cairo, Egypt I knew the end was near.
On the bright side, we love what Lenovo has done on the PC side. Innovative products, good availability, decent margins for us and support that is better than most (HP=UGH). So, we look forward to improvements once Lenovo takes over the xSeries.
Support has been IBM's biggest positive factor.
First PCs, Now Servers
I'm sure Lenovo begs to differ
Seeing that Lenovo's is far from stupid, why would they buy a business if they felt there "simply is no money to be made from selling any kind of x86-based hardware"?
Because they dropped the US employees for one.
So ldo17 was incorrect, as apparently you can
but then inst that like saying that you also couldn't make any money off of iPads if they were manufactured in the US.
Otherwise, they would be.