Here are some parts from his report. It's quite interesting, if true.
GOOGLE, THE HARDWARE COMPANY. In a recent conversation with renown technology pundit, Robert X. Cringely, we discussed some very intriguing potential hardware moves by Google that could have a profound implication on not only Google, but potentially the Internet and media in general. We particularly find these potential products interesting, as most investors think of Google as a software / programming company, and little discussion has taken place on Wall Street regarding Google as also a hardware company. Should these potential products from Google materialize, this will change.
GOOGLE DATA CENTER CONTAINERS – At Google’s investor conference earlier in the year, we had a chance to ask Eric Schmidt about Google’s job posting for someone to help in purchasing dark fiber. Many investors were unclear as to why Google would need dark fiber, but the obvious answer was Google was looking to connect its data centers cheaply. This made sense, but we were left wondering if there was more to the job search. Since then Google has been very active buying up fiber, and Mr. Cringely has hypothesized (http://www.pbs.org/cringely/pulpit/pulpit20051117.html) what Google may be experimenting with: building a data center that fit with in a standard shipping container. The 20’ x 40’ container could be packed with almost 5000 Opteron processors and over 3.5 petabytes of disk storage. While this packs a lot of power, it wouldn’t outsize any of Google’s current 64 data centers (up from just one only 2 data centers 2 years ago). So what would make these data center containers so special? They could be dropped off literally overnight to any of the 300 Internet peering points globally. Many may wonder what benefit so many access points would bring to Google besides redundancy. The answer is speed – or reduced latency. Google could be closer to the end user than anyone ever was before. There are many applications that would benefit from this: Google Videos and IPTV sitting just a hop or two from the user, virtual office applications with robust features (i.e. running desktop applications on the Internet itself with very fast interactive features like interactive 3D maps), servers would never “go down”, and data would never get “lost”. In essence, the company could build its own Google Net, which would sit right on top of today’s Internet. Google would basically have an access point in every major city which would add two other benefits: 1) it could prioritize its own VOIP, for premier telephony experience; 2) it would reduce Google’s dependence on other 3rd parties for trunking or access. But could Google afford it? To build and operate the containers annually, it would only cost about $2-3B, readily done with over $7B of cash on the balance sheet, (and still plenty of room even if Google invests in AOL for $1B (see below)).
GOOGLE CUBES: THE LAST MILE – In our conversations with Mr. Cringely, it appears that this may not be all that Google has up its sleeve on hardware. While Google is potentially thinking “big”, it may also be thinking “small”, “very small”. Through our conversations, we have learned that Google may be considering developing Google Cubes (we have dubbed them this, as we are not sure what they could be called). Basically the device would be a small box with many connections ports on it, in addition to wireless (Bluetooth / WiFi). Its potential purpose: it could connect to your TV or PC, or PVR, or stereo. As long as one of them is connected to a broadband connection, the Cubes could form a mini mesh home network. What could this enable? Many things – the obvious applications would include: enabling IPTV to easily connect from the web to your TV and enabling the transfer of video downloads from the computer to the TV, enabling easy transferal of MP3s on your computer to your stereo (or vice versa), and VOIP connection to all existing phones in your home. However, the cubes could potentially work with your home security or even your home climate controls (imagine turning your air conditioning on in your home right before you leave work so that your house is cooler when you arrive, and you didn’t waste money cooling it all day). Now imagine that you live in an apartment complex and the neighbors wanted to share 1 broadband connection – all the neighbors cubes could “talk” to each other (if desired). However, should this be in Google’s plans, it would take a lot of network support – enter another role of the widely distributed data center containers. One could even fathom the cubes (or one main cube) talking directly to a nearby data center container, which could eliminate the “last mile” problem completely. The idea sounds intriguing but what would it cost? The cubes would be designed to be as “dumb” as possible (which is the whole point of making the network the computer), and Google would probably subsidize them so that they cost <$20 or maybe even free (like AOL CDs).
HARDWARE CONCLUSION: We underscore to investors that for now, any potential development by Google of data center shipping containers and Google Cubes has not been confirmed. However, one could see how these ideas could prove interesting to Google, should they be feasible. Further, we think that at minimum investors should takeaway from this that Google is not just a software / programming company. In fact, Google could over time become more of a hardware company than anything else. At the very least, we think investors should at least consider these possibilities.
