Earlier this month, at the Cloud 2020 summit, in Las Vegas, a group of service providers, vendors, buyers and pundits met to discuss the future of cloud infrastructure. One of the tracks was on the economics of infrastructure services, and included John Cowan from 6Fusion, whose cloud metering technology is a key piece of the puzzle if cloud is to be treated as a commodity, James Mitchell, founder and CEO of Strategic Blue, and former Morgan Stanley commodities trader, and Joe Weinman, author of the fascinating book Cloudonomics.
Out of this track emerged some interesting debate around the viability of the whole notion of a cloud futures market. Jonathan Murray, EVP & Chief Technology Officer at Warner Music Group, followed up with an article presenting his perspective, from the skeptical end of the spectrum. This article presents a view from the optimistic side.
I’m indebted to John Woodley, investor and board advisor at Strategic Blue, for his insights and comments. John is a former Managing Director at Morgan Stanley, where he was co-Head of non-oil commodities for EMEA. He was the first person to be hired by a Wall Street bank from a power utility, becoming a founder member of the Morgan Stanley electricity trading desk. As one of the world’s foremost authorities on commodity market development and the pricing of non-standard pseudo-commodity products, his perspective is fascinating.
Jonathan explicitly self-identifies as one who is unconvinced about the prospect of a futures market for cloud. His article is fairly long, and his argument well made. The core point he puts forward is essentially: “There is no volatility or friction so we don’t need intermediation.”
Is there friction in the cloud market?
Jonathan doesn’t explicitly define friction, but I think he means that in today’s world, primary buyers can easily find primary sellers, and as such there’s no need for an intermediary. This is a fair point. In the old world, this was impossible. In a market without intermediaries, in order to get the best price, you need each of the buyers to make as many calls as are there are sellers. Let the number of buyers in the market be ‘b’ and the number of sellers be ‘s’. In a tiny market, with 10 buyers and 10 sellers, you need b*s calls in order to achieve the best price, or 100 calls in total.
However, in a world in which buyers are also resellers, the number of calls needed in order to secure the best price becomes much larger - somewhere around the factorial of the number of buyers plus the number of sellers - (b+s)!. In our tiny market, (10+10)! is 2432902008176640000. If one entity in the middle takes calls and keeps note you only need b+s calls and not (b+s) factorial. This is a conventional argument for the existence of a market - it offers huge savings. On these grounds, it’s hard to imagine Jonathan thinks there isn’t a market, a forum for price comparison, but certainly in today’s world the web is that intermediary near as dammit is to swearing. Jonathan correctly describes markets as places where supply is matched to demand, and indeed, we use markets because they have proven better than manual or -so far- computational planners in correctly matching supply and demand efficiently.
There is, of course, very significant friction in the cloud market, per a different definition. Look at the way cloud providers sell their services. They want long-term commitment, in their own currency, by credit card. By contrast, the cloud user wants minimal commitment, the best flexibility to move should a better deal emerge, or a better technology offering appear. They’d rather not have to keep a credit card on file, and in many cases, for example in India, currency fluctuations make a huge impact, and they’d greatly prefer to pay in their own local currency. In this situation, an intermediary who buys from the providers on terms which suit the provider, and sells to the buyers on terms which suit the buyers is very attractive.
Let’s examine whether there’s a need for intermediation against the assertion that there is no volatility. I think that assertion is false. It strikes me as similar to the assertion that there was no petrol price volatility in the US in 1979 or more recently in the Northeastern US after Hurricane Sandy, because the posted price at the pump did not change. The reality of course is that the underlying value changed so radically that people would queue for hours to get the scarce commodity and even threaten each other with violence to get precedence.
To be sure we have not seen that in cloud yet but we have seen scarcities. John Woodley cited an exemplary case with which he was particularly familiar. The Morgan Stanley IT department was approached to lend space at a datacenter because a major movie producer could not get enough capability on the open market to render a movie before the all-important Christmas release date. Had the IT department called a Fixed Income trader before agreeing to the deal I am fairly sure the producer would have had to pay more in cash or kind than a few T Shirts for the deal.
So, I have to ask: Is there no state of the world in which a sudden demand for many thousands or millions of instances might occur? Are there that many sitting idle? No, a far more productive and interesting line of thought is how the demand surge might occur and how to be in the right place at the right time to benefit. For an interesting example, consider the events surrounding Silver Thursday, in 1980, in which the Hunt Brothers had hoarded a third of the world’s supply of silver, inflating the price by a whopping 700%. At the same time, anecdotally, a certain bank held for some strange reason a large part of the silver smelting capability, and thus were able to charge greatly increased prices for people wanting to melt down the family silver, to sell as ingots. In John Woodley’s view, it is tremendously likely that we will see scarcity and volatility in the cloud markets, and a wealth of opportunity and demand to intermediate.
What about contractual preference?
I touched on this when discussing the idea of friction in the cloud market. A second but no less important driver for intermediation was not addressed in Jonathon’s article. In certain markets, the contractual preferences of buyers and sellers diverge dramatically. In such cases the insertion of a financial intermediary is very helpful and both parties are prepared to pay the rather minimal cost such intermediation causes. An example that is blindingly obvious can be found in electricity. Power plants are only used when the demand is sufficiently high to cover the operating expenses of the power plant. Power plants are built on borrowed money. A power plant owner needs to finance the debts used to build the plant, and so wants a fixed payment per month from their customers. By contrast, a retailer of electricity simply wants to be charged on a per use basis, and wants nothing to do with fuel cost pass-through, or a fixed monthly fee. Think about the way datacenters are funded. It’s exactly the same model - why do you think cloud providers offer such a huge discount for a commitment to usage? It’s the same! In the electricity markets, intermediation provides a welcome service, in which the intermediary is the buyer to every seller and the seller to every buyer, ironing out the contractual wrinkles, and benefiting the market as a whole. There’s already evidence of demand for this kind of service in the cloud market today.
The fundamental point Jonathan makes in his argument is that because there is no volatility or friction, there’s no need for intermediation. However, there is already evidence of price fluctuation and scarcity, and in terms of the delta between the way buyers want to buy and sellers want to sell, there is clearly friction. The cloud market behaves like and looks like a blend of the electricity and coal market, both of which are highly intermediated and heavily traded. Yes, it’s early days, but it seems very likely indeed that we’ll see the same kind of behavior in the cloud.