Sun's COO Jonathan Schwartz wrote has a couple of interesting blog entries on software pricing and business models. His final comment in the most recent entry is interesting:

“It’s far easier to convince a customer to pay for a product they're already using, than one they haven't even tried."

Indeed!

I'm sure Jonathan gets a chance to meet with some cool and senior level people; so do I and it does make the job a lot of fun. However, I've yet to find a CEO, CIO or CTO let alone a president or prime minister who can't see a "Loss Leader" from a mile away. Maybe I just misunderstood what Jonathan meant but I don't think so. Sun's strategy is evolving to a business model where you give software away for free and you generate revenue from other sources like services and hardware sales. Sun has always been a hardware company so that’s nothing new. Sun has not been a services company in any meaningful way so that is new and worth discussing.

In my view Jonathan’s equating of consumer and enterprise service models is problematic.  In the consumer space you can get away with defining a half dozen aggregate service models which map to your customer demographics. That’s why when you walk into your local cell phone store you'll see only a limited number of service plans. Unfortunately companies of all sizes are big hairy complex entities which in most cases have unique business models, processes, resources, customer strategies etc. No two businesses are the same. In fact if you think about it you will realize that a priori businesses must be unique. If they were not then how would they differentiate themselves in a competitive market? Aggregate service models which work in the consumer space do not work for companies.

In the enterprise you need to offer highly customized services because each business is unique. As I discussed before the systems integration service model does not motivate behavior which is attractive to customers i.e. it does not promote a focus on reducing complexity over time. It can not. If it were to do so and you were to follow that to its logical conclusion where complexity is reduced to zero impact then any service company following this model would have put itself out of business. There are also some very large players in this space who out-scale Sun.

There’s always the utility computing service model. That could be the one Sun is going to focus on. Unfortunately that model has its own problems. Businesses customers require highly customized and unique services to match their unique business models. This is not like the consumer business.

Sun adopting a utility services model would be entering a market dominated by existing players such as IBM and HP. Playing number three in any market is unlikely to be a profitable proposition. It’s not clear what advantage Sun would have over either of these companies. There are also some fairly systemic problems with the whole concept of utility computing.

I get utility service from my electricity provider today. They deliver 220 volts at 50hz at a lower cost than I could probably generate it myself. I can not order 190 volts at 65hz because the economics of a utility service require scale and standardization. Today’s state of the art in systems and software mean that the same scale and standardization economics apply to utility computing service models as they do to utility energy models. Automated mass customization of computing service delivery may be economically viable ten years from now but that is certainly not the case today. If you want utility computing tailored to the unique needs of your business that means lots of consultants and increased systems development, deployment and operational costs for your utility service provider. Those costs inevitably get passed on to the customer. Only through massive scale and agreggation can these costs structures be kept under control. If you are not one of the very biggest players in the utility computing space your ability to compete on price will be very limited.

The reality is that "Utility Computing" today is just the outsourcing wolf in virtual sheeps clothing. IBM’s “On Demand” (An IBM trade mark) branding is just that; branding. IBM is working on some cool software and advanced technologies for virtualization which will help to deliver autonomic computing at some point in the future. The ugly reality is, however, that the state of the art in software today means that “On Demand” in any practical sense is just re-branding of IBM’s good old outsourcing service model with the all the labor and resources intensive processes and costs that implies.

The economics of outsourcing are awful. If I was investing in an IT services business today then one based on infrastructure and operations outsourcing would be below last on my list. The razor thin margins in the classical outsourcing business mean that only IBM and the few big players in this business can survive because of their huge scale and abilities to drive cost savings based on volume and aggregation. It is very common to find that outsourcing deals only become profitable in the fourth and fifth year of a five year contract. Outsourcing also has some unnerving similarities to pyramid selling schemes. New customer need to be continually signed up to feed ever greater levels of scaling which enable continuing aggregation and scale based cost reductions. If that input of new customer slows down then the foundations of this business model start to look a little creaky.

The irony is that the very breakthroughs in software that are required to truly deliver profitable outsourced/utility computing service models are the same breakthrough that will deliver value directly into the hands of customers. In my view it is only software companies that will be primarily motivated to invest the R&D to deliver these breakthroughs. If software complexity costs trend to zero over time then outsource service providers will surely benefit but then again so will end customers. If a customer has to choose between handing their mission critical IT infrastructure over to a services provider, with the loss of control this implies, or can choose to run an autonomic computing environment in house with little or no cost penalty then which option do you think they will find most attractive?

If Sun is moving down a path to give all its software away for free and reinventing itself as a services company the critical question is which services model does it think it can both compete in while delivering enough revenue and margin to keep shareholders happy. If business models do matter and they ultimately determine how a company will deliver value then this might be all you need to know about Sun's strategy and what this means for its future.