The public cloud changes everything about IT, and IBM (NYSE:IBM) is destined to be one of the victims. I tried to make the case to Warren Buffett in a six-page letter sent in April 2013. My effort to persuade obviously failed – he still owns the stock – but the argument still holds. Here is the letter, a narrative that includes Jim Sinegal of Costco (NASDAQ:COST), Amazon Web Services (NASDAQ:AMZN), Netflix (NASDAQ:NFLX), Bob Feller and Michael Jordan.

April 10, 2013

Attn: Warren Buffett
RE: The Vulnerability of IBM’s Operating Model

Dear Warren,

As you’ve counseled for many years, investors should wait for the fat pitch – a pitch straight down the middle of the plate. As a fellow shareowner of Berkshire and allocator of capital, may I offer: IBM looks like a Bob Feller fastball coming right at our head.

First, a discussion on why IBM’s operating model is suddenly quite vulnerable, then the scary part about IBM, the rough equivalent to that Iowa farmboy’s heater.

IBM’s operating model

The reason IBM’s moat is problematic is explained by Herbert Spencer’s “survival of the fittest” metaphor. As Spencer pointed out, the key to survival is not necessarily derived from being biggest, fastest, strongest. It’s about fit. Those who best fit the environment have the best chance to survive. Those poorly designed for the environment perish.

An example you’re probably familiar with is the white moth/black moth comparison from 19th century London. The rapid buildup of soot due to industrialization led to the white moth population getting wiped out. Against a black backdrop, it was easy for birds to find white-moth lunch.

Note the causal connection. It was a change in environment that sealed the white moth’s fate, not anything the white moths did or didn’t do.

The same is true for IBM. The company is not facing difficulty because of anything it did. It’s facing difficulty because of forces exogenous to the operating model; namely, the environment in which IBM competes has suddenly and fundamentally changed.

IBM is an elite, iconic name, the “Michael Jordan of the IT world.” As M.J. learned when he tried minor-league baseball during a year away from basketball (because of gambling allegations), greatness doesn’t always port when the game gets changed. Being the best at twenty-foot jump shots doesn’t mean you can hit a curveball…or, in M.J.’s case, a fastball, slider, and change-up.

IBM’s competitive environment

Computing power, it turns out, is remarkably fungible, an ideal fit for distribution from a central source – what the media calls the “cloud.” What’s happening to the distribution of computing power is similar to what happened to the distribution of electric power in the mid-19th century.

(Nicholas Carr has written about this, though it’s not clear if he originated the analogue.)

When electricity production first became viable, companies built customized electric power plants, assembling them in their respective backyards. Then, when technology made it possible (through the development of alternating current), the distribution paradigm changed from discrete power generation to centralized because it’s more efficient.

For electricity users at the time, switching was a no-brainer. They traded large capital expenditures for low variable expense. Entrenched interests tried to defend the status quo, but there was no competitive advantage to having a power plant in the backyard, so economics drove the paradigm change.

How does this relate to our position in IBM?

IBM is in the business of selling, installing and servicing computing power plants in the backyard of blue-chip businesses and government agencies. Just like with distribution of electricity, and seemingly overnight, the backyard computing power plant is no longer the most efficient way to distribute computing power.

For a tiny fraction of the cost of the backyard computing power plant, users can tap into computing power by going online and connecting to a cloud service host.

In terms of user experience, it’s like comparing travel on a horse and buggy to a Ferrari. You’re either tapping into the limited computing power assembled in your backyard, or into the vast power of a centralized, shared infrastructure. Computing projects that used to take weeks are now completed in an hour or two on the public cloud, and at a tiny fraction of the cost.

The fundamental problem facing IBM, Oracle (NYSE:ORCL), Hewlett-Packard (NYSE:HPQ) and other traditional players in IT: The public cloud is a low-margin quasi-utility.

As Adrian Cockcroft, CTO of Netflix said, “there is zero revenue for traditional IT” in the public cloud. That sounds like an overstatement, but it’s not too far from the truth. At least we can say it’s tough to find any high-margin revenue in the public cloud model. Users pay low variable expense (prices have been dropping from cheap to super-cheap) with no capex, so how can IBM make money in the public cloud?

IBM operating model doesn’t fit, and there’s no easy fix

The public cloud is not an environment in which IBM can generate much profit because, as (NYSE:CRM) CEO Marc Benioff said, “IBM needs to sell a box, and the cloud is not a box.” IBM can’t tell its customers to tap into a computing utility (the public cloud) on a pay-as-you-go basis – as opposed to constructing a power plant in their backyard – because it won’t support its high-margin operating model.

IBM’s operating model revolves around selling a bundled product of hardware, software and service. The company’s markup on cost in the aggregate is roughly 100%, and if you back out services (which are run at breakeven), the markup on hardware and software is north of 200%. Of course, high markups are fabulous, but only if they are defensible.

Hardware is increasingly getting commoditized – Google (NASDAQ:GOOG), Amazon (AMZN), Netflix (a former IBM customer) and others assemble their own servers at about 1/3rd the price of equipment from Dell (DELL), Hewlett Packard, etc. That much is well known. What’s not well known is that software is getting a lot more competitive, with margin pressure likely to follow – again, causation can be traced to the public cloud.

Application developers are like kids in a candy store. Any software developer with a smart idea can rent the Ferrari to build, test, integrate and deploy applications. In the old environment, it used to require an investment of many millions of dollars in computing infrastructure before you could be a legitimate player in enterprise software. Proprietary code is another barrier to entry that’s going to the wayside. The wide availability of open source code is allowing small, nimble software startups to tackle just about any programming problem. It’s why disruptive innovation in software is happening from the ground up, not from the top down.

IBM is telling its customers…

Here’s what IBM is telling customers: Yes, you need to get moving toward the cloud. Only thing is, for security and reliability and control, you should build a “private cloud.” What’s a private cloud? You might have guessed this much: It’s a computing power plant in your backyard, cloud bells and whistles included.

As an alternative, IBM customers need to build a “hybrid cloud.” Once again, it involves building a private cloud in your backyard, though this time it’s directly connected to the public cloud. Whatever it’s called, the bottom line is this: IBM needs its customers to build something big and expensive and local.

It’s a tough sale. When they construct a private cloud, customers effectively give up everything that is wonderful and beautiful about the public cloud. Trading capex for low variable expense doesn’t happen when you have a private cloud. And there’s waste involved in the private cloud: Instead of paying only for what you use, private clouds are built for overcapacity (they have to be able to handle peak usage). And then there is that other beautiful thing about the public cloud: It’s the functional equivalent of driving a Ferrari on the cheap. Spiegel (a German media company) recently converted the file format of 20,000 movies using the computing power of the public cloud; it took 2 days instead of 2 months, and the cost was less than its electric bill would’ve been using a private cloud.

The way I see it, IBM has one hand to play and it’s a bad hand. It has to defend the status quo (capex invested in the backyard). IBM’s economics depend on it, but the economics of customers are pushing the industry in an altogether different direction, toward commoditization of hardware and software, combined with low-margin, pay-as-you-go computing power.

But wait a minute, isn’t there another door open to IBM? Can’t IBM get behind the public cloud and make some profit, even if it is low-margin?

Here, the story takes an interesting turn

The dominant player in the public cloud is AWS (Amazon Web Services, a division of They’ve built out a centralized computing power plant – many call it cloud infrastructure, or infrastructure as a service (IaaS). It’s a computing utility upon which everything else is built, e.g., storage, applications, database, etc.

How did AWS capture 75% of the cloud infrastructure market, with $2.5 billion in expected revenue in 2013 and top line growth of 65%? This is where the story takes an interesting turn. Established IT players, including IBM, are seeing prospects for their operating models dwindle because of someone wholly unaffiliated with AWS and the industry. In fact, he’s retired.

Enter, stage right: Jim Sinegal.

While the media portrays Jeff Bezos as a keen admirer of Sam Walton, it was actually the guy on the other side of Lake Washington that really captured his imagination. Bezos sees the Costco operating model as having the ultimate moat. Who in their right mind would open a warehouse club anywhere near a Costco location? With 2% net margins and boatloads of capital needed to get scale, the answer is easy: Nobody.

Bezos believes that Jim Sinegal’s invisible commitment to never mark up anything more than 15% (even if they could easily pass on the mark up, Costco resists the temptation), builds customer loyalty, even when customers aren’t consciously aware of it. By way of comparison, Wal-Mart’s average markup is 34%, and Target’s is 46%.

How’s this for gumption? AWS has taken Sinegal’s markup strategy and applied it to the public cloud. Not only has AWS kept markups for its public cloud hosting service to a low, flat rate, but as it achieves economies of scale it lowers prices – 23 times in the last five years – and each time AWS lowered prices in the absence of competitive pressure.

Here is a recent snippet from Werner Vogel’s (Amazon CTO) blog; he’s discussing a database product on AWS (the emphasis is mine). Imagine how the competition felt waking up to read what the largest player in the category was up to last night:

In the year since DynamoDB launched, we have seen widespread adoption by customers building everything from e-commerce platforms, real-time advertising exchanges, mobile applications, Super Bowl advertising campaigns, Facebook applications, and online games. This rapid adoption has allowed us to benefit from the scale economies inherent in our architecture. We have also reduced our underlying costs through significant technical innovations from our engineering team. I’m thrilled that we are able to pass along these cost savings to our customers in the form of significantly lower prices – as much as 85% lower than before.

The details of our price drop are as follows:

Throughput costs: We are dropping our provisioned throughput costs for both read requests and write requests by 35%. We are also introducing a Reserved Capacity model that offers customers discounted pricing if they reserve read and write capacity for one or three years. For customers reserving capacity for three years, the price of throughput will drop from today’s prices by 85%. For customers reserving capacity for one year, the price of throughput will drop from today’s prices by 70%…

How cool is that, anyway? Lowering prices in the absence of competitive pressure?

Of course, AWS is not lowering prices because it’s charitable. It’s cementing relationships with customers. As a bulwark to moat, the implications are fun to think about (thank you, Jim Sinegal). The largest player in a low-margin utility market generates economies of scale and automatically gives it back to customers in the form of lower prices. So how is the #2 player supposed to compete? (Beats me.)

Where does all of this leave IBM?

The company is in a tough spot. The game has changed and it’s not a change that favors Big Blue. IBM has already lost the nascent public cloud infrastructure market, as AWS has got scale, is the low-price leader, and is using Sinegal pricing tactics. Game. Set. Match.

IBM’s best chance (it’s only chance?) is to convince its blue-chip customer base that, despite the promise and allure of the public cloud, customers still need computing power plants in their backyard. The strategy has a chance in the short-term, given widespread confusion about the cloud (IBM can argue security, reliability and control issues).

Long-term, it’s hard to imagine how IBM can thrive in the new environment. The company has to convince customers to forgo the public cloud, with its low variable expense and superior computing power. Security, reliability and control are increasingly seen as curable issues – even non-issues – for the public cloud. Economics are not on IBM’s side and, at the end of the day, the struggle for existence among operating models is all about economics.

Here comes Bob’s fastball

This is what makes IBM’s competitive situation as scary as a Bob Feller fastball aimed at our head: The move to the public cloud is already entrenched among smaller companies, and upmarket migration appears to be accelerating. As venture capitalists will tell you, entrepreneurs and startups are all building their IT infrastructure in the cloud. (There may be an exception, but I couldn’t find one, even in companies with over $100 million in funding.) This suggests IBM’s long-term prospects are particularly troublesome, as it will have to wean tomorrow’s enterprise customers off of a low-expense utility model and onto the high-capex client-server model.

People in the industry tell me at least half of Fortune 1000 companies are already experimenting with the public cloud. Further, most government agencies are looking hard at the public cloud as a way to save money. Companies and government agencies are generally testing new workloads at this point (as opposed to migrating existing operations), but a couple renegades are moving much faster.

Samsung (OTC:SSNLF) recently moved one of its computing hubs to the public cloud, saving $34 million in the process, and Netflix is in the final stages of moving everything (its entire IT infrastructure!) to the public cloud. That’s a big deal for the likes of traditional IT, since Netflix is one of the biggest consumers of computing power.

I’ll conclude our story where it began, with Bob Feller on the mound. If it wasn’t for a change in the competitive environment, IBM might have been a soft toss down the middle of the plate. But the environment has changed, and now there appears to be a little white sphere hurtling toward our head at a high rate of speed. My suggestion, respectfully submitted: duck.

Best Regards,

Arne Alsin