Microsoft should buy Rackable instead of building custom computers

Comment We're in the midst of some very strange times. An online book seller owns the leading utility computing service. An advertising company manufactures its own servers and switches. And spots in rural America best known for being, well, rural are turning into technology heavyweights because they have access to cheap power and generous taxpayers.

Given such circumstances, it seems only sane to expect a few peculiar acquisitions to take place in the coming months. After all, the characteristics that once separated a software company from a hardware company or an online store from a computing provider have started to vanish.

Were Microsoft a hard-charging, forward-thinking company rather than a reactive laggard, it might choose to address this new reality by purchasing Rackable Systems of all things. It may sound crazy, but here's why it's not.

Last month, Microsoft's infrastructure chief Debra Chrapaty gave a speech which failed to attract the attention it deserved. Speaking at the Microsoft Management Summit in Las Vegas, she outlined the company's aggressive plans to build out its computing systems. Chrapaty, during the speech, went so far as to confirm that Microsoft will start building its own systems a la Google.

"We're also doing some other innovations," she said. "You know, when you think about the servers, you can just design servers that use less power, and so we're doing that. We're working on our own server design, our own motherboard design, chip design."

It's hard to tell how seriously we're meant to take that statement.

For one, it's more than vague, since Chrapaty is claiming that Microsoft will do its own chip designs. We doubt that Redmond plans to build a chip fabrication plant anytime soon, so we think she just meant that Microsoft is incorporating low-power processors into its custom motherboards.

Beyond that, however, there's just no indication that Microsoft is actually building its own gear. The most radical infrastructure thing Microsoft has done to date is to go the whole hog with the data center-in- a-container idea. Microsoft is in the middle of building a massive data center in Chicago, which will feature hundreds of the containers. Our sources indicate that Microsoft has no intention of filling these containers with its own systems. Instead, it's asking the likes of Rackable and Dell to craft very demanding boxes.

These statements seem to reflect Microsoft's tendency to talk big and deliver small. (Do we really need to point you at Vista?)

So, while it has hopes of building its own systems to compete with Google, it would likely take the company years and years to establish the necessary internal systems to produce hardware.

Microsoft could, however, speed up this whole process by acquiring Rackable, which stands as the leading cloud/utility hardware maker around and which already sells tens of thousands of servers to Microsoft.

Rack 'Em Up

Under previous management, Rackable received and then turned down an acquisition offer from Dell. That was a huge mistake since Rackable's share price was almost triple its current level. In addition, Dell along with all of the major hardware makers have since moved on with their own cloud/utility computing designs. This leaves Rackable in an awkward position. Its natural acquirers have disappeared. It will need an unnatural act to strike it rich.

Rackable is expected to release new systems any day now, and our sources indicate that the hardware will put the stuff being pimped by the Tier 1s to shame. The Silicon Valley-based company offers the most energy efficient, balanced servers of anyone and has, by most accounts, the top data center in a container design, which should only improve with the new servers.

In addition, Rackable has always run on a build-to-order model. The company is flexible enough and smart enough to cater to a customer's individual needs, unlike the big boys which must make drastic compromises to serve as broad a set of customers as possible.

So why not have Microsoft commit that unnatural act and buy the server house? Rackable's market capitalization stands at just $417m, as of this writing. Microsoft could afford to pay a 100 per cent premium and grab Rackable for $1bn. That's just two data centers' worth of money at the going $500m per center rate.

With Rackable on its side, Microsoft would have cutting-edge server designs, potentially capable of saving the company millions of dollars in power costs every year. It would also have the best container team going. And, most importantly, Microsoft would gain top data center talent with the skills to help it keep pace with Google.