Oracle made a lot of noise and published no fewer than 17 press releases this week, during its OpenWorld developer conference in San Francisco. But not all of the news is equally important. Judging by how much time Larry Ellison spent talking about it, the company’s new second-generation cloud infrastructure has the top priority.
Ultimately, the company is optimistic about its prospects in public cloud, despite that that market is 10 years old.
“In terms of actual deployments, we’re barely out of the first inning,” Robert Shimp, Oracle’s vice president of technology marketing, told VentureBeat in an interview recently at company headquarters in Redwood Shores, California.
But there are questions: Will many enterprises and startups test out and then stick with Oracle’s cloud infrastructure as a service (IaaS)? Will Oracle spend the money necessary to fund worldwide data center infrastructure expansion? Does Oracle have the team in place to frequently release and update services in order to stay competitive? Will Oracle just end up throwing in the towel with commodity public cloud, as HP did not very long ago? Will Oracle solve basic problems with the experience of setting up new computing resources to more closely resemble larger public clouds?
What’s certain is that people will be spending more, not less, on the public cloud in the next few years. In 2020, 30.2 percent of all IT spending will be on public cloud, up from 18.9 percent last year, IDC said in April. A 451 Research survey found that 6 percent of enterprise workloads are running on public clouds. If Oracle can hang in there, it will see revenue growth.
But there’s no guarantee that Oracle will increase its market share as it competes with Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP). And it probably doesn’t help that in 2009 Ellison called the cloud “water vapor.”
Still, executives are bullish. They think they can be more successful than Dell, HP, Rackspace, and VMware here.
Oracle cofounder and chief technology officer Larry Ellison demonstrates Oracle’s second-generation cloud infrastructure onstage at the Oracle OpenWorld conference in San Francisco on September 20, 2016.
Above: Oracle cofounder and chief technology officer Larry Ellison demonstrates Oracle’s second-generation cloud infrastructure onstage at the Oracle OpenWorld conference in San Francisco on September 20, 2016.
Image Credit: Oracle
“We’ve taken a very different approach,” Don Johnson, an Oracle vice president of engineering who previously spent seven years as an engineer at AWS told VentureBeat at OpenWorld. “We have fundamental investment and ownership of our software stack. “We haven’t cobbled together a collection of various pieces and tried to stand it up as public cloud.”
In a talk about the new cloud infrastructure at the conference earlier in the week Johnson told a packed room that there would be “quarterly, monthly, weekly, almost daily releases.” That’s bold, but it’s the way things work at AWS. It’s very different from the cadence of releases of Oracle database software
Johnson told me the ability to ship faster has to do with the fact that people have come to the Oracle cloud organization from web companies like Amazon, Dropbox, Facebook, Google, and Pinterest, and they’ve sought to put in place the types of tools they’ve used before to rapidly send out updates, monitor, manage, and automate. “The thing that gives me confidence is really the people,” Johnson said, adding that building the new Oracle cloud infrastructure has been “the most successful and the most fantastic engineering effort that I’ve been a part of.” He wouldn’t talk about the exact size of the company’s geographically dispersed cloud organization, but in 2014 Oracle said it was looking to hire more than 100 engineers to work at a cloud engineering center in downtown Seattle.
Johnson said that the Oracle cloud would provide a wider variety of operating systems for customers to choose from — when I tested it earlier this year, it had Oracle Linux and Oracle Solaris, but no Windows, no Ubuntu, no CentOS, no CoreOS. And he said virtual machines (VMs) would start up faster — I found that an Oracle VM took 3 minutes and 30 seconds, compared with just 30 seconds on AWS. It will be possible for people to manage Oracle cloud infrastructure with more orchestration tools — presumably such as Puppet, Chef, and Ansible — in the future too, he said.
And it matters that Oracle has hundreds of thousands of customers, plenty of which have been making the shift to Oracle’s cloud software anyway. “I don’t think we’re out there saying, ‘Hey, you want some compute? Who’s having that conversation, to be honest?” said Steve Daheb, senior vice president of Oracle cloud. Oracle is interested in moving companies’ existing applications into its cloud. “We’re going in and having those conversations,” he said.
At 2016 Oracle OpenWorld in San Francisco.
Above: At 2016 Oracle OpenWorld in San Francisco.
Image Credit: Jordan Novet/VentureBeat
Johnson pointed to Oracle’s vast financial resources as one of Oracle’s other strengths. And that’s warranted. In the quarter that ended on August 31, Oracle brought in $1.83 billion in net income on $8.59 billion in revenue. Income adds up.
“I think right now, we have about $70 billion of cash … to put out there,” Oracle co-chief executive Mark Hurd told reporters in a press conference following the OpenWorld keynote on Monday. It’s just a matter of how much money Oracle is willing to spend to further build out the necessary data center infrastructure, leaving aside other costs. Hurd shied away from committing to a dollar figure. But he did provide a general model about how infrastructure spending will work.
At first, he said, the company spent money to put in place the data center gear that’s necessary to run software as a service (SaaS) for its customers. Then came spending for the equipment necessary to provide platform as a service (PaaS) to customers. “You’ll see more investments into infrastructure, but remember, it’s riding behind the wake of the existing SaaS and PaaS infrastructure that’s out there,” Hurd said. He suggested looking at Oracle’s expenditures for SaaS and PaaS. The company is taking “that same approach now into infrastructure,” he said.
If nothing else, the company’s most recognized leader has changed his rhetoric when it comes to the cloud. On Tuesday Ellison spent much of his keynote talking about how his company’s cloud services outperform Amazon’s and undercut them on price. “We’re more than 100 times faster than Redshift, using half as many CPUs,” he said. Oracle duly issued a press release and published details on how it ran its benchmarks.
In a blog post encapsulating this round of Oracle cloud announcements, Gartner analyst Lydia Leong wrote that it should be seen as “a positive move for Oracle, but one with many open questions about its future.” That’s fair.
I would just add that it seems Oracle has no immediate plans to back down in cloud infrastructure. If anything, it’s just getting comfortable.
“There comes a point when every startup founder does some soul-searching and asks the question: What is it that I actually want out of this?,” explained Ben Medlock, SwiftKey cofounder and chief technology officer (CTO), to a packed audience in London.
Medlock was speaking as part of the Re.Work Deep Learning Summit in the U.K. capital, but before matters turned to machine-learning, the computer scientist and entrepreneur was quizzed about the recent acquisition of his startup by the mighty Microsoft.
For the uninitiated, SwiftKey is the popular mobile keyboard app that enables users to type more quickly on touchscreens — it learns your writing style over time, based on historical data, and serves up “next word” suggestions accordingly. The company also boosts its predictive typing smarts by scanning texts from myriad third-party sources, thus learning common word-order sequences.
Though SwiftKey had built a reputation in the consumer realm for its Android app, behind the scenes it carries out significant research into artificial intelligence, machine learning, and natural-language processing (NLP). It also offers a software development kit (SDK) for third parties that allows them to integrate its language-learning technology into their own services. And in 2014, it finally launched for iOS after Apple opened up to third-party keyboards.
SwiftKey had raised north of $20 million dollars since its inception in 2008, and, as with any firm with investors to please, there comes a time when thoughts steer toward one of two possible exits: going public or being acquired.
“We started out just wanting to solve this problem of helping people to type,” explained Medlock. “But we were very clear that we never wanted to run a sort of… maintenance business. We were only interested in carrying on if we could see that we were growing and exploring new ideas.”
For a company such as SwiftKey, an acquisition always seemed to make more sense — it just never felt like a billion-dollar business that would take the public markets by storm. But the company wasn’t thinking about an exit when it first started talking to Microsoft.
“When we started talking to Microsoft a couple of years ago, we weren’t actively looking for an exit at the time — but it kind of prompted one of those soul-searching sessions,” continued Medlock. “And I think you kind of look at the core growth engine of your own company and what’s generating the cash and how likely it is that will continue to grow and how likely it is that will give you the opportunity to explore new areas of research.”
Ultimately, the decision to sell boiled down to one key question: How did the opportunity to grow as an independent company stack up against the near-infinite resources and reach provided by a much bigger company?
“I think if we had felt that being an independent company would give us the opportunity to keep growing and develop in a rapid way, we would’ve stayed independent,” said Medlock. “But we felt that — weighing things up — we would have a better chance of exploring some of these new ideas as part of a bigger company. We founded in 2008, and we felt that we’d done pretty much everything we could do as a tech startup in London.”
A sign of the times
These days, it seems any startup that gains a semblance of traction is snapped up within a couple of years. That it took eight years for SwiftKey to be acquired is perhaps a sign of not only where it was founded, but when.
“If we’d been on the West Coast of the U.S., we probably would’ve been acquired earlier,” said Medlock. “When we started in 2008, it felt like there were very few tech startups in London at all, and incredibly few that were really deep technology-focused.”
Though London has emerged as a hotbed for tech startups, it may have taken longer to gain the attention of suitable buyers, because the U.K. lacks the status of Silicon Valley. But the journey from small-time startup to one with millions of dollars in VC cash and hundreds of employees is one that Medlock doesn’t regret for a second.
“The life experience that you gain through the trials and tribulations of building a startup business — I think there is no way I would’ve traded that for an earlier acquisition and less heartache and stress and sleepless nights,” he said. “When you look back on it, it’s not really the value you’re proud of, it’s all the things that you went through and the people you worked with. And for me, that journey is well worth waiting for.”