On Jan. 2, technology news outlet The Information published a report suggesting that hyperscale cloud vendors Amazon (AMZN) and Salesforce (CRM) are aggressively moving down the path of migrating operations off Oracle’s (ORCL) database technology and onto open-source and/or internally developed solutions. The Information’s report cites sources with knowledge of the projects underway at Amazon and Salesforce to overhaul their technology stacks, though neither company had directly commented on the story as of this writing.
While the news of Salesforce’s efforts to move off Oracle are not necessarily new (the rumored Project Sayonara has been a relatively open secret for some time), the news of Amazon potentially shifting its internal operations would lend credence to our belief that Oracle’s database business faces a murkier future than consensus believes.
While we are maintaining our wide economic moat rating for Oracle at this time, we remind investors that our negative moat trend rating, which also remains intact, revolves around our view that customer switching costs are in fact decreasing for Oracle database customers. The shares have retreated roughly 2% on The Information’s report and have fallen roughly 13% from their September highs; they now trade near our unchanged $46 fair value estimate.
Oracle has routinely denied rumors of this ilk (including The Information’s report) for some time, and while we do believe database migrations are much more challenging than some may let on, we think the company has been too nonchalant about the improvements made in the open-source database community and by rivals over the past several years. Our moat trend rating does not suggest customers will migrate away from Oracle across the board, but we do think enterprises will increasingly look to downsize their reliance on Oracle’s database technology, particularly for workloads migrating to the cloud and for net new workloads.
We believe certain workloads will always be best served by high-performing databases such as Oracle’s, though we note that the company is no longer alone at the top of the database performance food chain. Microsoft’s (MSFT) SQL Server has made substantial strides in recent years, aided in large part by that company’s need to scale up its database technology for Azure. In fact, Microsoft landed ahead of Oracle in Gartner’s November 2017 Magic Quadrant for Operational Database Management Systems on both ability to execute and completeness of vision. Meanwhile, Amazon has been active in developing its own databases (such as Redshift), and Gartner estimates it amassed 5% share in the $34 billion database management market in 2016, up from essentially 0% share just a few years ago. Oracle’s share retreated from the mid-40s to just over 40% during that time.
Further, Oracle’s lack of positioning in the markets for public cloud infrastructure as a service and platform as a service creates challenges for customers that want to reap the benefits of the environments offered by Amazon, Google (GOOG)/(GOOGL), Microsoft, and others. Coupled with the rise of increasingly popular open-source solutions such as PostgreSQL for mission-critical applications, these factors contribute to our belief that Oracle’s customers are facing decreasing switching costs.
Despite Oracle’s marked progress in software as a service and other portions of the cloud value chain, the company’s top and bottom lines are still driven in large part by databases. Gartner estimates Oracle’s total revenue from databases approached $14 billion in 2016, more than 35% of annual revenue in fiscal 2016 and fiscal 2017. If we assume a 45% GAAP operating margin for the database business, this business would have been worth nearly 50% of the company’s operating profit in fiscal 2016 and fiscal 2017. Oracle needs to quickly (and profitably) scale up its cloud businesses in the event that customers do begin to seriously migrate away from its database offerings. To Oracle’s credit, the SaaS business continues to see a strong upward lift on its gross margins, though the broader IaaS/PaaS business continues to operate at heavily dilutive levels of profitability compared with the broader business. IaaS/PaaS GAAP gross margins have fallen from a peak of 58% in the first quarter of fiscal 2017 to just 39% in the first quarter of fiscal 2018, while revenue grew only 27% over that period despite redoubled efforts in the company’s IaaS business.
While we would be surprised if either Amazon or Salesforce outright confirmed The Information’s report, this smoke could mean a fire. In the case of Salesforce, reports of the company’s potential migration away from Oracle date back to 2013, when Salesforce and Oracle announced a nine-year partnership centered on Salesforce continuing to consume Oracle database licenses, Oracle-branded Linux OS licenses and support, and Java middleware, while Oracle worked to integrate its enterprise resource planning solutions with Salesforce’s best-in-breed customer relationship management platform. There were rumblings before and after the deal’s signing that Salesforce had been growing weary of its relationship with and reliance on Oracle and the latter company’s technology. Project Sayonara gained traction in a May 2016 Forbes report suggesting that Salesforce planned to be completely migrated from Oracle technology by 2023, which would coincide with the end of the nine-year stated partnership. We view this outcome as more likely (though still not an absolute certainty by any stretch) following The Information’s report, which cited a source close to the situation as stating that Salesforce continues to develop alternative solutions for its Sales Cloud and Marketing Cloud offerings, which represent 37% and 10% of our fiscal 2018 subscription revenue forecast, respectively, excluding Commerce Cloud.
Turning to Amazon, The Information’s report suggested that a migration away from Oracle has its roots in the early 2000s, referencing Amazon’s high-profile outage in 2004 as a turning point in Amazon’s perception of its reliance on Oracle as a technology provider. The report suggests Amazon’s retail unit has switched to internal solutions running open-source NoSQL databases that handle both customer and order information. Amazon has not confirmed this report, and we would caution investors that a database migration of this size would be a massive undertaking, as many databases have multiple dependencies across a boatload of applications in large enterprises, creating a significant degree of execution risk when undertaking a migration. However, large companies with substantial IT resources such as Amazon may in fact be in a position to provide a blueprint for leaving these environments, particularly when examining the financial implications of consuming open-source database technology. An April 2015 Gartner report, “The State of Open-Source RDBMSs,” postulated that an enterprise running a single Oracle 12c database license across two cores would spend $473,100 over three years; a comparable deployment of Postgres Plus would cost just $41,400 over the same period. Even if we assume discounting on Oracle’s part, this still represents a massive potential cost savings, should enterprises more aggressively pursue database migration projects.