In Only Ten Years Too Early, Fred Wilson recounts his VC investment in a “cheap internet appliance”:
The idea was that we could build a large user base and make money through advertising, marketing, and e-commerce. It was 1999 of course.
We invested something like $10mm in the business and built a team of talented engineers and business people and launched a device which we partnered with Virgin Entertainment to take to market. This is what the device looked like:
Needless to say, this was not a successful investment. The device worked but it had a number of fatal flaws outlined in this PC World review from 2000.
While this looks like a case of massive wishful thinking taking over sensible viability assessment, we can’t really be sure why the venture failed without inter-related facts and the benefit of hindsight. Fred’s conclusion? “Beware of ‘way too early’.” This does sound like good advice, except perhaps when coming from a VC, especially when contrasted to Peter Drucker’s dictum, “The best way to predict the future is to create it.”
When is it “too early”?
As is the nature of investing, a VC’s job is to bet on the future. VCs and their portfolio companies can either chase the coattails of a trend (usually beyond their control) or, a la Drucker, try to actively shape the future by creating the necessary ingredients if/when necessary:
• Google – We explored how Google created GOOG-411, a free automated directory service, to harvest phoneme data to improve their speech recognition algorithms so that YouTube videos can be indexed (and thus searched and monetized) through speech-to-text conversion. (This will be likely used in the upcoming Google Voice as well.)
• Apple – When Apple wanted to redefine what a cellphone is, it didn’t phone in yet another clunky “smartphone” with an Apple logo on it. For over a decade, the incumbents in that industry were satisfied with existing form factors, UIs, features, carrier relationships, revenue models and infinite small-minded product differentiations. Unlike Nokia or Microsoft, Apple didn’t say, “There are too many industry restrictions and technical hurdles for us to redefine this category. It’s way too early.”
Apple didn’t leave anything to time or fate. In hardware, software and services, it actively created a space for the iPhone to succeed. Where it was needed, it poured in R&D dollars, invented new things and collected hundreds of patents; gained new expertise in radio technology; bought a company specialized in multi-touch; converted a desktop OS into mobile one; created new global distribution channels; negotiated a ground-breaking arrangement with the largest carrier; and so on. In other words, instead of creating, hoping and predicting that the iPhone would somehow become a killer product, it made sure that all the supporting elements were in place to make sure that it would.
• Sun – When Java was first introduced, perhaps the biggest gripe against it was its relative slowness compared to established languages like C++. A lot of people were ready to write off JVM. Sun, it turns out correctly, bet that Moore’s law and software optimization would catch up to bring Java performance up to par, and productivity gains from a better language would outweigh the initial performance concerns. CPUs got much faster and Sun managed to further speed up Java performance. For many years now, Java performance on the server has ceased to be an issue.
• Cloud computing – Perhaps the most “hyped” growth domain today is all the ways in which we can connect our desktops and mobile devices to servers and services in the cloud. Now, any conventional IT executive or a VC in need of an 18-month exit strategy can find excuses to not fund cloud computing projects and products: security, data portability, management, reliability, performance, etc.
For example, it would be easy to claim “too early” by noticing that average consumer upload speeds last year in America via common cable/DSL conduits was only 435 Kbps, much too slow for various entertainment and medical applications, among others. This gets even worse for enterprise computing where terabyte-size datasets need to be routinely uploaded to the cloud for tracking and analysis.
Is it “too early” then to positively consider Nirvanix, Bycast, Mozy, Cleversafe, ParaScale, Dropbox and numerous other mostly consumer oriented companies for cloud storage? How about Amazon, Microsoft and Google offerings that go from simple storage into cloud-based applications and transactions? Should the column-based data warehouse company Vertica not offer an analytic DB for the cloud, because it’s “too early” to transmit gigabytes of data over a typical corporate T-1 connection?
Further, should Nvidia stop its GPU platform because major OS vendors don’t (yet) support it? Should Apple delay its GrandCentral or OpenCL initiatives because the hardware side of the equation isn’t sufficiently complete, pervasive or economic? Examples abound.
Chase or lead?
A decade ago PointCast became a canonical example of a product that was “too early” because we didn’t have pervasive broadband for a push service (though the company would have been a success if it didn’t refuse the $450M buyout offer from News Corp). Since then Apple and Google, among others, have taught us how to create the necessary conditions for their products to succeed by compressing time through strategic invention.
If VCs want to become indispensable to the companies they advise, perhaps they ought to remember that it’s “too early” only if you fail.