You may recall one of my first blog entries assessed the fitness of the word “commodity” for the computing marketplace. Distilled to a single sentence, my conclusion was that despite the self-interested rhetoric of some vendors (and gullibility of a few pundits), computers weren’t the commodity – computing (and bandwidth) was. Just as power generators built by my friends at GE aren’t the commodity, electricity is. It’s not even close to a subtle distinction.
In looking at the evolution of the commodity called computing, history provides an extraordinary parallel to the evolution of electricity. In fact, if you haven’t read it, I’d highly recommend “Empires of Light,” by Jill Jonnes. It’s a very entertaining historical examination of how electricity was first discovered (rubbing amber produced mysterious sparks), reliably generated, and ultimately distributed across the world.
One of my favorite anecdotes from the book describes the financier JP Morgan’s decision, as the primary backer of Thomas Edison’s Menlo Park electrical inventions, to wire his house for electricity. He elected to dispense with the “vile poison” of gas lighting, and place lightbulbs throughout his Madison Avenue mansion.
Not only did he have a customized generator placed in his stable, but given the fragility of the system, he had a professional electrical engineer staffed to manage it. So at 11pm every night, the lights went out, because the engineer went home. It makes the point – early in its evolution, only the wealthy could afford electricity (along with the requisite generator and electrician), and the technologies were fragile. Businesses that wanted power generation facilities were similarly wealthy enough to afford large-scale versions of the same, staffed with a “chief electricity officer” and teams of electricians. They didn’t exactly experience 5 9’s availability (and people actually died regularly, talk about MTBF).
It took about a decade for those deploying electricity to settle on a few standards that ultimately accelerated consolidation. From voltage to cycle to plug configuration. (The processes used to get there, although they involved far more violence and loss of animal life, bear a remarkable resemblance to standard setting in the computing industry.) Spooling forward, once the standards existed, businesses could plug into a grid – labor markets went through a fairly sizable dislocation (all those engineers and “CEO’s” had to find other work), but electricity was firmly established as a ubiquitous service. Scale efficiencies and the resulting massive decrease in price allowed the government to bridge the power divide through rural electrification. Electricity that started out 20 times the price of gas lighting – obviously got a lot cheaper.
What’s most interesting to me is that once the standards were set, and the grid powered up, electricity finally established a transparent price – the hallmark of a true commodity. If pricing isn’t transparent, products can’t be deemed a commodity – by transparent, I mean equivalently defined for a standard unit of measurement. Here are a few examples, “5 cents per kilowatt hour,” “2 dollars per gallon.” It’s either a standardized physical delivery (gallon, barrel, ton), or unit of consumption (typically time based, 100 megabit hours, megawatt hours, etc.) – but it’s the same across the industry.
Here are some examples of things that aren’t commodities: a 4-way x86 dual core Opteron server running an open source indemnified Solaris OS with a J2EE 1.4 compatible Java Enterprise System web services stack, optimized JVM and 256M of RAM. Should be obvious, but I’m amazed that some folks still think that’s the commodity – vs. a unit of computing on such a device, or moreover, on a very large scale grid of such devices.
And so it’s to that very issue you’re going to see Sun address our upcoming quarterly announcement event this week – the evolution of a true computing grid, priced at $1/cpu-hr; the business and technology models underlying such a transformation; and moreover, the impending impact on the marketplace for computing power and value. (Whereas power operators can’t add a lot of value to a powerline, things are a tad more hopeful for the network operators.)
As I pointed out in some of my earlier musings, there’s a sense of dysphoria among those that see computers as commoditizing, a belief that no one can make any money if the technology’s interchangeable. In my view, the great thing about commodities, whether financial services, telecommunications, oil and gas, and now computing – is that the companies whose business it is to monetize those commodities, along with the businesses that supply the technologies necessary to compete in a commodity market, are among the largest on earth.
Vodafone, Citigroup, Exxon Mobil. What do they have in common?
1) They’re among the most valuable businesses on earth.
2) They’re primarily technology companies.
3) They’re among the largest buyers of technology in the world. And
4) They’re all in commodity businesses.
And how do they all thrive?
ps. You’ve got to love IBM’s ability to play the community. Going through some of the patents they “donated” to the open source community a few weeks back, it looks as if they all, curiously, seem to be due for payment – and thus potential expiration – this year. Were they destined for the bit bucket (turns out IBM is among the largest patent expirers in the world, along with its largest issuer).
And some of the patents have nothing to do with open source software – my favorite in the heap is this one. Not sure that’s going to be quite the comfort the community’s looking for. Here are a few others – for those working in gel embodiments; and for the open source doctors in the crowd.
We know we need to help the community understand how to take advantage of our grant – but at least all 1600 of the patents we’ve granted to the world were for operating systems and software (USPTO 700, for the wonks in the crowd).