Background

The previous post discussed some of the strategies that can be followed by infrastructure as a service (IaaS) providers. At a high level, IaaS strategies are typically based on some mix of low price (offering lowest prices per unit of value), high differentiation (offering unique capabilities), and focus (selling to a more narrow group of customers). But what is the “right” strategy? Well, there is no single right or best strategy for all IaaS providers. Instead, the question must be answered individually by each IaaS provider. To arrive at a profitable answer, each provider must also first understand customer needs, segmentation of customers, their own unique capabilities, strengths and weaknesses of competitors, and many others.

Also suggested in the previous post was that Amazon Web Services is following a low price strategy with its EC2 offering, and doing a pretty good job. Doing a good job in part because their strategy, operational model and positioning appear aligned. Here is a fairly representative statement from Amazon: “You pay only for what you use, with no up-front expenses or long-term commitments, making AWS the most cost-effective way to deliver your application to your customers and clients.” The most cost-effective. Not the most differentiated or the most focused. Amazon is not claiming the most secure, the most compliant, the best hardware, or the best service. They are claiming lowest cost. This does not mean Amazon is the “best” IaaS provider. It means they are trying to offer the lowest cost.

New Developments

Since that post, Amazon has introduced spot pricing for EC2 instances. Briefly, “Spot Instances are a new way to purchase and consume Amazon EC2 Instances. They allow customers to bid on unused Amazon EC2 capacity and run those instances for as long as their bid exceeds the current Spot Price. The Spot Price changes periodically based on supply and demand, and customers whose bids meet or exceed it gain access to the available Spot Instances. Spot Instances are complementary to On-Demand Instances and Reserved Instances, providing another option for obtaining compute capacity.”

What does this mean?

In terms of strategy, this new move serves in part as a re-confirmation of Amazon’s low cost provider approach. It also provides great buzz and free marketing (such as this blog post) for EC2. The spot price alternative should also offer good value to the subset of customers that have workloads that can be run -- as well as terminated -- “whenever.”

The relationship of spot pricing to Amazon’s low cost provider strategy is interesting. Commodity pricing drives commodity providers to reduce costs as much as possible. Here is why: Commodities are not differentiated except by different grades or levels of quality. For instance, corn may be assigned a grade based on the percentage mix of kernel colors and the percentage of damaged kernels. In general, higher quality -- or grade -- commands a higher price in the commodity market. But this does NOT mean the provider with the most low grade corn (or even the most high grade corn) wins the battle for profits. Instead, each grade of corn establishes its own market price, driven by supply and demand. The provider of the corn does not determine the price, the market does. However, the provider influences its own profit levels by driving its costs to the lowest levels possible while still producing the target grade of corn. It is the difference between price and cost (i.e. price minus cost) that determines profits. Amazon probably would not initiate a commodity-like market for compute power without believing it had a strategic cost advantage.

There are a number of caveats related to whether or how well Amazon has established a “true” or efficient commodity market. Some have suggested that Amazon is not even offering market driven pricing. It is true that Amazon ultimately sets the spot price, but it is doing so based on supply and demand, following an auction process. In this model, some entity needs to serve as the market maker or the one who matches the offers for sale and purchase. It happens to be Amazon in this case. Some believe Amazon’s role means that market prices are not the result.

However, even with Amazon in this role, the result *is* market driven pricing. The buyers have freedom of choice and can decide whether or not to purchase spot EC2 instances at any given price. The real catch seems to be that Amazon is currently the only supplier. This means that Amazon does not have any direct competitors (at least for EC2 spot instances). So if the real point is that Amazon’s EC2 market is not reaching lowest possible prices due to lack of competition, then I would certainly agree. It is not a competitive market but a monopolistic market.

Remaining Questions

There are still a number of remaining questions. Will Amazon be successful in this approach? Will others try to duplicate it? Should others try to duplicate it? Will a truly competitive market for commodity compute services emerge? What technical, economic or social changes would need to occur to support this model? If you wish to discuss further, please contact me by email or Twitter.