Thursday, August 14, 2008

Are We Overselling the Cloud to Ourselves?

I was doing some casual reading tonight (which is all I have time to do lately, it seems), when I came across this post from Thomas Wailgum of on InfoWorld. (Ain't syndication grand?) The majority of the post is commentary from Gartner about the relative infancy of SaaS ERP solutions relative to their on-premises bretheren. Interesting in and of itself, but not normally worthy of a post here.

However, on the second page, I came across the following quote:
Other inhibitors to more widespread SaaS ERP adoption, Ganly contends, relate to total cost of ownership (TCO). TCO of "SaaS ERP suites likely will be significant and may not compare favorably with on-premises solutions," she adds. This problem applies to vendors as well. SaaS vendors "often have unrealistic expectations of their operating costs," she writes. "The multitenant architecture needed for SaaS ERP suites results in high internal efforts and costs for the initial setup and the ongoing maintenance and upgrade of the system."

Security has also been an issue with SaaS ERP offerings, "especially with regard to financial data and privacy concerns," Ganly writes. "Vendors must prove to organizations that are considering SaaS ERP adoption that their security and privacy concerns are unfounded through super low-cost or no-cost, proof-of-concept trials, encouraging early adoption through value pricing and getting early adopters to share their success stories."

[Emphasis mine.]

It occurs to me that this is a really good point to consider when looking at the economics of the cloud computing market. For SaaS vendors, cost-of-sales is still high, as the sale is (and always will be) a hybrid of the traditional enterprise sales model: high investment in building customer relationships, proving technical and business feasibility, and navigating corporate politics, though likely with fewer of the "big meeting" costs found in traditional relationship sales.

Thus, the "economies of scale" from data center operations may be vastly overshadowed by cost of acquiring customers.

However, a "pure infrastructure" play (such as poster child Amazon), eliminates most of the cost-of-sales if they can prove low barrier to entry and significant flexibility of use. Most customers discover Amazon, get set up for free, then pay nominal charges to figure out for themselves how to use the platform. There is no real data lockin, as the storage services are essentially device storage (as opposed to specific data schemas), so the cost of choosing not to move forward with a pure IaaS vendor is relatively low.

There are few CxO level relationships between AWS and their customers (though I don't doubt there are several with, say, financial services megoliths with deep pockets and an interest in influencing AWS).

The point is, when most technologists think of the cloud, they think of something like Amazon, not something like DemandERP. But much of the value of the cloud comes from getting the resources you need in (usually) an on demand model. If price and experience can't be both superior to on-premises ERP for the customer, and profitable for the vendor...well, as the kids say these days, "fail".

I worry that many of the boutique IaaS vendors are also going to fall into the same trap--not understanding how the cost of acquiring customers to a specialized platform or service will wipe out the economies of scale savings of multitennancy. There will be a lot of churn out there in the coming years, and a lot of wispy corpses floating in the clouds. Caveat emptor.

Oh, and to the point of the efficiency of Amazon's model: Jeff Barr notes that he can't find a failed startup that used EC2/S3 as its core infrastructure:
"One of the major value propositions of Amazon Web Services is the utility pricing plan. That is, you only pay for what you use, and the cost is very low. Sometimes it feels like I am just saying that: not because there is any doubt that it’s true; rather because it’s difficult to produce metrics to back up assertions that “low cost utility pricing” is truly a game changer.

Then it hit me… Looking at the list of Start-Up Project presentations on Slideshare’s site, I realized that not a single one of these companies is “off the air”; that is, they all are still in business. In the Startup world that is nothing short of amazing—especially in this economy. (Some of the decks on Slideshare's site are not from last year’s startup events; however even those other companies appear to be alive and well.) Amazon can’t take all the credit for this track record; however it does seem to be a solid data point that validates the value proposition."

That is amazing, if it holds true.