The Voodoo Economics Behind Utility Computing

Before you sign up for pay-as-you-go, you need to understand how utilities can manipulate and hide the true costs of plugging in

The whole marketing idea behind information utilities is that their data and transactions - much like water and electricity - are available whenever you need a sip or want a jolt.

The frustrated General Manager of a fast-growing division of a Fortune 100 pharmaceutical company decided to game his corporation's IT budget rules. On one hand, he couldn't afford to self-fund a supply chain initiative he thought essential to his group's growth. On the other, corporate wouldn't fund division-driven apps unless the group committed itself to an unrealistic ROI.

The business manager sat down with his IT guru and crafted a cunning third option: Transform the supply chain initiative into a "value-added" e-mail initiative. Why? Because e-mail-oriented IT initiatives were funded by corporate as "infrastructure". The manager got his infrastructure proposal approved.

So what's the real difference between an app and an infrastructure? That's easy: Don't look at who uses it; look at who pays for it. Management - not technology - determines when an app is an infrastructure and an infrastructure is an app.

The current incarnation of the apps versus infrastructure debate can be found in the promises of the pay-as-you-go "information utility" metaphor being marketed by such vendors as Hewlett-Packard, IBM and Sun Microsystems. "If you can make [computing] a utility," HP senior technical adviser Joel Birnbaum observes, "that means your network is on all the time, and you'll use special services only when you need them. If you do one day's supercomputing a month, you don't need to own it."

Indeed. The whole marketing idea behind information utilities is that their data and transactions - much like water and electricity - are available whenever you need a sip or want a jolt. Utilities are infrastructures that facilitate and enable apps. Quality and reliability standards exist. Costs are more or less predictable. Information-intensive companies such as JP Morgan Chase and American Express seem increasingly convinced that the utility analogy is the smart way to manage their businesses.

There's legitimate logic to this. But CIOs investing in the utility paradigm need to understand what they're really implementing. Any serious discussion about utilities requires a brief appreciation of their economics. The fact is that the history of utilities in every industry is a history of regulation, politically driven cross-subsidies and monopoly pricing - which are not necessarily bad things. But let's not kid ourselves that an enterprisewide information utility would be anything but a creature of internal and external regulation, cross-subsidies and monopoly, no matter who runs it. Numerous empirical studies assert that regulators invariably fall captive to the utilities they oversee. Executive operating committees supervising their information utilities may share that same fate.

So any company implementing an information utility isn't implementing a cost-effective ensemble of digital networks as much as creating a regulated monopoly destined to battle over cost allocations for bits, bytes and bandwidth. Why? Because utility economics vastly favours cost recovery over value creation.

When a utility incorporates a feature or a function, one way or the other, it seeks subsidy to guarantee a return on its investment. A utility is a social construct that uses cross-subsidies to assure that everyone has access to the desired resource at an "equitable" price.

To be sure, a utility's seeming ability to exploit economies of scale, standards and interoperability makes good business sense. In fact, these arguments for corporate information utilities sound familiar. They're just like the arguments used by the old Bell system to justify its monopoly.Ain't nostalgia grand?

Page Break

The best reason why information utilities may be hazardous to corporate CIO health is that they inherently trivialise the hard lessons we've learned from deregulation in so many industries during the past several decades. Energy utilities and telcos developed increasingly complex networks of financial cross-subsidies that far exceeded the technical sophistication of their physical networks. Large customers subsidised consumers and vice versa. No one - not even the regulators - could get a real grasp of costs. The clever accountant had greater impact on a utility's fortunes than a brilliant engineer.

To be sure, deregulation has its debacles. Enron's frauds immediately come to mind. The hideously mismanaged deregulation of California's power grid. The savings and loan scandals of the 1980s. But these examples only illuminate the larger point: Ill-conceived regulations create market distortions that pervert economic efficiencies and undermine business effectiveness. Utility economics are predicated on the fundamental notion that a regulated monopoly will allocate resources more efficiently than a more competitive marketplace.

Now, I'd be the first person to agree that internal competition for IT resources is not likely to be cost-effective. But I'm the last person to believe that a dominant information utility is the most economical, responsive and cost-effective approach to IT management in either the short or long term.

The classic spiel supporting information utilities is that corporate customers will plug in apps just like electric utility customers plug in appliances. The sad fact is that not all plugs are compatible. Not all utilities understand how to manage peak and off-peak pricing. Monopolists tend to be lousy collaborators. Indeed, many large customers annoyed with utility pricing actually go off the grid. They explore alternative energy sources.

Utility Computing's Siren Call

The current interest in information utilities reflects that corporations are sick and tired of the uncertainties, risks and costs of enterprise computing. Vendors recognise this. That's why the lure of an outsourced utility is so tempting (read "Plug and Play", May CIO). Then again, if pay-as-you-go info-utilities were really the way to go, you'd think more businesses would use chargebacks. They don't. What we have is a wilful ignorance of real economics and true costs.

There is no point in trying to implement an information utility until the CIO sits down with the CFO and COO and explains that a CRM system or the e-mail network can either be infrastructure or apps. Implementing a utility means using a cost structure - nothing more, nothing less. Allocating costs for shared services is an accounting game, not technology management. That's equally true for recovering costs from that information utility investment.

To put the question harshly, how do we know we're being cost-effective if we don't know what our costs really are? The CIO as "Information Utility CEO" is appealing because utilities are so good at concealing, manipulating and cleverly reallocating their costs - at least until some serious competition comes along. That's the cynical interpretation. Here's a kinder one: CIOs should encourage top management to carefully examine the information utility idea to better appreciate how their internal marketplaces distort, pervert and misalign IT investments and implementations. The accounting costs of implementation can't be divorced from the implementation of accounting costs. Tomorrow's IT infrastructure investments should be determined by the business value they can create, not the accounting loopholes they can exploit.

Michael Schrage is codirector of the MIT Media Lab's eMarkets Initiative. He can be reached via e-mail at