Investors love cloud computing. There is no universal definition of "cloud computing," of course, but a little mystery only fosters more enthusiasm. I would like to offer a sober view based on several years of experience of building applications for the cloud.
Cloud computing dates all the way back to 1950s. It was not just a theoretical concept - it was widely used since the days of the first mainframes. The Internet gave it a major boost. Anyone with a web-based email account benefits from the cloud technologies every day. Thousands of companies are building their own clouds to run their applications, like Office 365 by Microsoft (MSFT) or Creative Cloud by Adobe (ADBE). Clouds offer lower costs, faster development of new features, better integration with outside services, simplified deployment of new versions, and easier customer support.
Cloud computing delivered as a service to other companies and individual software developers is a relatively new phenomenon. Dozens of companies offer these services in the U.S. alone. They all offer slightly different solutions, but ultimately they all promise to host and run applications on shared computing infrastructure. I divide these companies into three large groups.
Hardware and Software Vendors
Large suppliers of information technology - IBM (IBM), Hewlett-Packard (HPQ), Oracle (ORCL), Microsoft and CA Technologies (CA) - were slow to warm up to the cloud technologies. Cloud computing is a very new concept to many of their customers. There must be a serious reason for a large corporation to move their critical applications to the cloud, and the most compelling attraction is the lower cost. This puts the IT technology providers in an awkward situation: in order to sell their cloud services they have to undercut their own sales of hardware, software and related services to the same customers. Not surprisingly, they were not pushing the cloud until it became clear that they may lose their customers altogether if they don't act.
Gmail played a very important role in the evolution of cloud computing. Originally most of the IT providers were very skeptical that corporate customers would trust their email service to an outsider. Little by little, however, customers started moving to the Google's platform. Success stories from smaller companies and organizations encouraged the larger ones to take a closer look. Soon a trickle of new customers turned into a massive migration of email services from in-house to the cloud . In 2012 Google announced that 5 million businesses, 66 out of the top 100 U.S. universities, and government agencies in 45 states adopted Gmail. This was a major wake up call to the rest of the IT industry.
It takes time, however, between the moment a company like Hewlett-Packard or Oracle realizes what needs to be done, and the time it can actually do it. Their massive sales forces were traditionally motivated to move large volumes of hardware and software. Investors still punish these companies when they come up short of their revenue targets.
It does not help that the new cloud technologies are mostly cannibalizing the existing sales. IT vendors cannot abandon their existing products, so they have to incur the costs of building out the cloud infrastructure on top of their existing costs, while the revenues stagnate or decline. This is, obviously, not a very healthy trend.
At some point the giants will be able to enjoy the lower cost structure of cloud infrastructure, but it may take two decades before most of the legacy software is finally decommissioned. Fortunately for the vendors, they can often stop spending money on the development of legacy products, and manage the support services to maximize the cash flow.
Note also that these companies have robust ecosystems of partners, distributors, resellers, solution providers, consultants, etc. These ecosystems will be hit hard as the migration to the cloud accelerates, and there will be some backlash to companies themselves through lower sales of development tools and related support services.
Finally, the cloud changes the competitive landscape. As large organizations get more comfortable moving their internal processes to the cloud, they have a wider selection of providers to choose from. A company using IBM's servers and database software for payroll decides to outsource is as likely to choose ADP (ADP) or Paychex (PAYX) as IBM. Claims processing? Look at Xerox (XRX).
It is also getting more and more difficult for traditional vendors of information technology to compete against the new breed of open cloud platforms that we discuss below. For example, Google App Engine, Amazon Web Services and Microsoft's Azure offer highly efficient and scalable non-relational datastores. These datastores offer significantly lower costs, require little or no administration, and offer built-in unlimited replication and scalability. The advantages are so big that I would never advise a startup company, or a company working on a new project, to even consider a traditional RDBMS. Fortunately for Oracle, IBM and Microsoft, most software developers grew up in the world dominated by relational databases that these companies supply. This is not a long-term deterrent though. The move away from traditional database software to non-relational datastores will become more and more pronounced each year.
The bottom-line for investors is that the move to the cloud is not going to be easy for these companies:
- They are likely to incur more costs, not less, until the transition is substantially complete, which may take a decade or more.
- They will face lower revenues as cloud cannibalizes their existing sales.
- They face increased competition and, as a result, a possibility of lower margins.
- They are losing their pricing power as open cloud platforms contribute to price transparency, shifting bargaining power to their customers.
Consultants and Emerging Vendors
For example, ServiceNow, which is a pure play on the cloud technologies, grew its revenues by 73% yoy in its most recent quarter, but it failed to produce any profits. Cloud technologies are complex and expensive, and selling them to its corporate customers requires a large and expensive sales force. At this point it is far from certain that ServiceNow can survive as an independent player. Despite this impressive growth in revenues, the company is still very small compared to its competitors. It has a market cap of $7.24 billion, or 17 times this year's revenues. The stock trades at next year's P/E multiple of 350. It will be a tough price tag to swallow even for the largest of the potential suitors. I believe the valuation has to come down by 50-75 percent before NOW becomes an attractive stock. At this point this stock is a very good short candidate.
Several companies took a different approach to cloud computing - what is now known as Platform-as-a-service, or PaaS. Companies that offer the most advanced platforms include Google (GOOG), Microsoft, Salesforce.com (CRM), and Amazon.com (AMZN). Rackspace Hosting (RAX) is trying to join this group.
Platforms have their unique advantages:
- Customers can open accounts online. They can deploy and manage their applications themselves. While live technical support is available, it is rarely used unless an issue needs to be resolved.
- Customers provision computing and data storage capacity as needed, often in automated fashion. All advanced platforms now offer auto scaling features: customers define rules for adding or releasing computing power based on workloads, and the platform adjusts the available resources accordingly.
- Customers only pay for resources consumed.
- There are no minimal revenue or term commitments, making it much easier for the new customers to try the service or migrate less critical applications first.
- Customers are less concerned about platform dependencies as moving applications and data from one open platform to another is much easier than migrating away from a proprietary cloud.
These features made PaaS the ideal option for startups and individual software developers. Many of these startups grew into very large companies (e.g., Netflix). These days many established companies are also actively migrating their in-house applications to these platforms.
All open platforms experience phenomenal growth in the number of applications, volumes of data and computing loads. It does not mean, however, that revenues are growing at the same speed. All PaaS players are competing on price - when one player drops the prices, the competitors match or exceed these cuts almost immediately. Since building robust cloud platforms is very expensive and requires a lot of capital, I believe that all of them are barely profitable at this point, if at all.
Here are a few more important considerations:
- Platform-dependency is a major concern for most of the potential customers. Platforms become more attractive by using popular standard technologies instead of proprietary solutions. The same approach, however, makes them more vulnerable. Moving an application from one standards-based platform to another is often as simple as copying disk and database snapshots.
- Platforms try to attract new customers by offering more runtimes and services. For example, Google App Engine recently added a runtime for a popular PHP programming language, which they snubbed for years. Microsoft did the unthinkable - it added Java to its Azure platform. These changes lead to even more competition between the platforms.
- It is relatively easy to mix and match different cloud technologies or services. For example, an application running on Google App Engine may use Amazon's S3 storage instead of Google Cloud Storage. This makes it nearly impossible to price a service at a premium to the competition.
- There is plenty of competition from the startups like CloudBees.
Analysts tend to assign very high valuations to cloud providers despite a very competitive nature of this business. For example, shares in Rackspace, which grows its revenues at 16% yoy, trade at 60 times next year's earnings. Ben Schachter from Macquarie Capital gives the following estimate for Amazon Web Services:
Using our estimate of $3.8bn for 2013 AWS revenues, and applying a ~5x multiple based on the comps noted above, we arrive at a valuation of ~$19bn for the business on an EV/Sales basis (equating to ~$41/share of AMZN stock). Importantly, we believe this to be a conservative valuation multiple, as AWS revenues are growing much faster than any of the comps incorporated above. At an 8x valuation multiple, we estimate the AWS business could be worth $30bn as a stand-alone company, or ~$66/share.
I believe that this rose-colored optimism is unjustified.
While most analysts correctly anticipate the rapid migration of computing workloads and data to the cloud, they tend to greatly overestimate the earnings potential of cloud technologies. Cloud is a very competitive field where establishing a dominant position with a corresponding pricing power is impossible. I believe that most of the current revenue and earnings expectations are overly optimistic.
Therefore I recommend:
- a short position in NOW and RAX,
- a bearish bias when evaluating traditional IT vendors (IBM, HPQ, ORCL, MSFT), and
- a very cautious approach to companies where cloud offerings represent a relatively small part of their business (AMZN, GOOG, MSFT, CRM). Exuberant optimism regarding AWS is a contributing factor to my bearish view on AMZN.
Additional disclosure: I may initiate a short position in NOW over the next 72 hours.