AWS and GCP have operating margins of roughly 40% (41% and 39% respectively). Almost half of every dollar spent on cloud computing is profit, which only benefits customers if the companies choose to reinvest that profit.
More worryingly, this margin demonstrates that this industry is already extremely consolidated, and we don't have any reason to believe it won't become more acute over time. Monopolies are always bad for customers, eventually.
But we needn't be all doom and gloom. I seem to remember someone once saying "your margin is my opportunity", and 40% margins sure seem like a massive opportunity to me!
Cooperatives recapture profits for customers
A cloud computing cooperative is an extremely elegant solution to this problem:
A member cooperative is an enterprise that is owned and governed by its customers. The model has a long history, and many companies you recognize are actually cooperatives (Vanguard, Land O'Lakes, Sunkist, Bob's Red Mill).
They tend to be created in industries where large fixed investments are needed to create basic infrastructure for the market to function at all. If such industries have many small companies, then it isn't reasonable or profitable for any of them to make those investments individually, and the market is vulnerable to monopolistic companies slowly controlling the supply chain.
Cloud computing fits this description precisely. Even in the dotcom bubble days it was extremely expensive to buy a server and put it in your dorm room, and now internet users have grown accustomed to the much higher performance of complicated multi-region deployments in front of CDNs and load balancers. To say that building cloud computing infrastructure is expensive is a laughable understatement.
Cooperatives remove the adversarial divide between producer and consumer, making them the same party. The fact that the cooperative exists for no other reason than to benefit its users, and they control it directly, means the decisions it makes are always directionally aligned with their best interests (as long as good governance can control and prevent principle agent effects).
A cloud computing cooperative could easily allow customers to reclaim almost the entire 40% margin that's currently going to the public clouds. For many customers a cloud computing cooperative would yield even more than 40% savings, since the bulk of their bills is markup on "serverless" products rather than on the actual underlying compute. And it would completely remove the threat of monopoly and other strategic risks.
If this is such a good idea why hasn't it happened yet?
The short answer is simply lack of incentive
The normal incentives of markets don't promise much reward for an individual person to found and build a cooperative, and getting a bunch of companies that are probably competitors to spontaneously work together without any leading voice is the definition of a very difficult coordination problem.
This is why many of the most prominent cooperatives were founded by absolute lunatics (John Bogle and Dee Hock come to mind). People who were driven by idealism (and maybe a bit of bitterness) rather than the promise of a huge payday.
I, Blaine Hansen, happen to be a lunatic. I want to start such a cooperative, at the earliest point I can actually achieve. The things that drive me are:
- I want a cloud computing cooperative to exist, even for my own personal use.
- I don't want the world to get more enshittified and monopolistic than it already is.
- I want to push forward the science of governance, and I need to actually step into the arena to do so.
I don't have the perfect resume to start a public cloud company, but I think I have a good enough intersection of skills, experience, and interests to start this cooperative (if I get a lot of help from the right people):
- I've been a software engineer my entire career, so I can build a team to write the cooperative's user-facing platform.
- I've been on the painful customer side of expensive cloud computing (BigQuery gets expensive fast).
- I'm capable enough at writing and presenting and speaking with people that I can pitch an idea that sells itself.
- I've been an avid student of governance and institutional design for years, and so can make cogent design decisions about how a bunch of cloud tenants should share costs and make decisions. (My sketches for pricing and governance are below in the details section.)
Isn't the technical side of this insanely hard and expensive?
Yes! At least in general.
But an important development has recently made it much more achievable: Oxide Computer. Their racks are designed from the bottom up to be as simple and open as the public cloud. It's conceivable that a cooperative with a fairly small expert staff could put a few racks in colocation sites and use that as a workable foundation to prove the model and expand membership and capability.
That means a relatively small amount of initial investment from a relatively small group of founding members could be enough. We don't have to solve an industry-wide coordination problem to make this happen.
This is also in the best interest of Oxide Computer since:
- It creates a way to sell their product to the long tail of the market that can't justify buying their own racks. Customers of sufficient size or those with specific needs will continue to "graduate" from public clouds, and this cooperative won't be an exception.
- It gives many more public cloud customers a way to exit those walled gardens and run on fully open stacks, making an eventual transition to full rack ownership even easier.
- It reduces the market power of the public clouds, meaning they have less ability to bully and trap their existing customers, and lure in new ones.
We're building to this slowly
Building an organization like this is very difficult and expensive, and it will require quite a few companies to band together and provide startup pledges. So for now we're only talking to potential members and putting together financial models.
If you're interested in being one of these founding members, or even an eventual member, get in touch with me either by email [TODO email address] or grab some time to talk [TODO booking page].
Let's work together to make the world better, take control of our destinies, and save a ton of money.