30 Comments

Summary:

Baris Karadogan makes a case for five big cloud computers.

Written by Baris Karadogan, a partner at Comventures

“I think there is a world market for maybe five computers.”

This is a famous misquote attributed to Thomas J. Watson Sr., then-president of IBM, in an apparent misjudgment of the PC market’s potential. As the story goes, under Watson’s leadership, IBM — which invented the PC — didn’t have a vision as to how big the market could become and let others, especially Microsoft, get the lion’s share of the value creation.

But maybe five computers are all we need. My friend Ray Conley, of Palo Alto Investors, made a good case over lunch one day that the statement could indeed be true if you think of the computer in the cloud sense. Could five computer “clouds” really allow us to do everything we want? The answer is yes, and they can already be found with Google; Amazon; Salesforce.com; either Sun Microsystems or VMWare; and Akamai.

First up is Google (GOOG), for consumer apps. Google can not only take care of all our email, social networking, sharing and blogging needs, but has the means to offer these apps online. Incidentally, Facebook can offer all of these apps as well, save for search. A 2008 prediction could be that they actually do something big on the search front, but I digress.

Next up is Amazon (AMZN) — not for their ecommerce solutions, but for their web services, which I’ve previously compared to semiconductor fabs. They represent the consumer infrastructure cloud. Any consumer Internet app first needs to be hosted somewhere, whether it be on Google or Facebook. Very few companies have the ability to build a big, global data center, but Amazon has one you can use, and pay for by the byte.

With the consumer apps and infrastructure clouds covered, let’s turn to the enterprise side. After all, anybody in front of a PC looks either like an employee or a consumer.

No. 3, therefore, is Salesforce.com (CRM). Just as Google gives you your consumer apps, Salesforce gives you your enterprise apps. Their cloud takes care of all the things you need to get your job done. And just like Google, they are open to third parties who will develop on their platform.

The fourth slot belongs to the enabler of the enterprise infrastructure via control of servers, storage, etc. The owner of this cloud is nebulous. Sun Microsystems (JAVAD) has products and solutions for backend infrastructure, while VMWare is taking care of servers on the front end. I am leaning towards VMWare (VMW), but there is a good case to be made for Sun as well. You take your pick.

The final spot is reserved for the one that gets all the bits to you — the cloud that has the intelligence to move things around. None of the other four work without it. It’s a CDN, and is best represented by Akamai (AKAM).

There you have it. If these companies could completely dominate their spaces, and kill every competitor, you’d be fine doing what you are doing on only five clouds.

OK, so who cares?

Well, from a venture capital investment perspective, if your company can knock out any one of these five, you have estate-making potential on your hands. Even if your company merely complements one of these clouds, you can still do well — either ride their platform or become one of their takeover targets. If you are trying to create a new kind of cloud, however, you may have a problem.

There is a public company investment angle here, too. These companies — call them the “T.J. Watson Portfolio” — are very well-positioned to win over the long term. Just to show you guys I am willing to put my money where my mouth is, follow my profile on zecco.com (covestor version coming soon), and let’s see if I’m right.

You’re subscribed! If you like, you can update your settings

  1. I don’t think that facebook has the talent nor skill to pull search off. There current infrastructure just isn’t right. Big search player in PHP? Not going to happen. And I just don’t think they’re dumb enough personally. There’s better things they could be doing, and let the “big boys” battle it out.

  2. I am afraid I have to disagree most of your assumptions.

    1. Salesforce.com does not even have complete CRM suite. The AppExchange is a desperate attempt at extending Salesforce applications.

    2 Facebook has no e-mail server, blogging service, calendaring service, web-based office apps, Web and blog search engines. So your assumption that Facebook does all Google apps is not quite accurate.

    1. Amazon.com can become cloud computing solution provider if they address numerous holes in the portfolio. Storage and Queuing services are good start but they are the easiest problems to solve

    Microsoft with Windows, Visual Studio and .NET framework, BizTalk server, E-Commerce servers, Great Plain Business apps and billion dollar investment into data centers might able to morph themselves into leadership position in this space.

    IBM with Linux, app servers with huge stack of development tools from Rational division and 32 data centers they already have is another potential candidate.

    All others have huge holes in their stack that including Oracle, SAP, Sun, Amazon and Google – Most significant holes are development tools, Frameworks, Horizontal Apps and credibility with developers.

  3. Thomas J. Watson Sr. died in 1956, long before anyone had any idea there would be such a thing as a PC. When he allegedly made his remark about the world needing only five computers, in 1943, he was referring to mainframes.

  4. I have a question for Baris: where does spatial data platforms fit into this concept of five clouds?

    Let me preface this with the observation, having been the guy behind a few very large spatial data projects for both major governments and very large Internet companies, that there is no one out there today with a general spatial data platform that could serve this role. Even the most celebrated geospatial and virtual world applications are built on incredibly limited back-end technology; there is nothing out there that lets you do with data in a spatial context what Google does with text. It is not for lack of interest either, it is a computer science problem for which there is no solution in literature. The interest in this application space greatly exceeds the ability to deliver at a basic technology level, and to compound the matter, the volume of useful spatial data sources far exceeds most people’s imaginations, easily in the same realm as the text web and we aren’t even trying yet (wait until location-based services for cell phones become ubiquitous…).

    Given that, there are two ways I can see spatial fitting into this conceptual model:

    1.) Spatial data is essentially a separate type of cloud with its own unique characteristics, but is not considered because there is no spatial data platform that can visibly anchor the concept of that space. The “unique characteristics” are very important in fact, as these include socio-political, economic, and technology characteristics that do not have good analogues in data spaces the market is broadly familiar with; this may not be evident unless one has worked in the space a while. Even if you made it part of a broader web services platform (e.g. Amazon), it would likely take on a significant life of its own.

    2.) In theory, spatial data technology is a direct superset of text, and anyone solving the spatial data platform problem in the sense that Google solves it for text could ultimately absorb the market that Google currently occupies (or some other end game that amounts to the same outcome). For this reason, I think it can be reasonably argued that at best a newcomer in this space would merely supplant Google’s role for many purposes. There are even some kinds of text analytics that are currently poor or impractical because implementing them well would require having a general spatial data platform.

    These two models suggest somewhat different market strategies if one was planning to build the backbone of “the” spatial data platform. As someone who just started a company based on a new massive-scale spatial platform that solved the theoretical problems alluded to above, the question of framing how such a platform fits in the market comes up repeatedly when talking to VCs. And it is hard to argue that any particular perspective is invalid in most cases because they are valid even if they reflect the personal interests and biases of the individual. This has not been a problem, but a lot of time gets spent on it because there are so many different views of how spatial data platforms fit within the current and future landscape.

    The article puts a fresh spin on an old subject and I enjoyed it, but it says very little that suggests where spatial data infrastructures fit into this. Spatial is still a little under the radar, but it is likely to become a major “cloud” element over the next couple years. I would be interested in a quick opinion on how spatial data platforms fit into this model over the long-term.

    Thanks!

  5. Interesting hypotheses. The weakest link, in my mind, is Salesforce. They have had such an opportunity to push harder on the SaaS front, while everyone else has been sleeping. With all businesses on wide pipes now, they should have been much further along. At SelectMinds, we saw Salesforce as our biggest potential threat and that was 5 years ago!

  6. I think this time Om was a little bit too visionary. Not very convincing, but funny to read.

  7. Nice try but I think you were very simplistic in your rationale.
    You missed the pipes, the gears and the software at least.

  8. @Nagger, Om did not write this column…

  9. Jonathan Sapir Friday, November 23, 2007

    Absolutely agree! I had a very similar post on “Super Platforms” last month http://www.limitnoneblog.com/2007/09/rise-of-super-platform.html/

  10. Don’t forget about real-time communications!

    The global phone system (the PSTN) is also a computing cloud. It may not seem that way because it isn’t really programmable. But what it lacks in flexibility, it makes up for in reach and reliability.

    Here’s an interesting question – If you had to lose one these clouds tomorrow, which would you miss most?

    The conversion of the PSTN to IP (over the last decade) combined with the current wave of web-telephony integration (aka “Voice 2.0″) is fixing that flexibility gap.

    More here… http://www.shaiberger.com/?p=47

Comments have been disabled for this post