Google made a series of announcements today at its Google Cloud Platform GCPNext event in San Francisco, and, notably, recently signed up Spotify to run on its platform. But I found the event somewhat disappointing; I expected more new product announcements and a clearer focus on mainstream enterprise adoption. More broadly, I think Google’s nascent platform—despite a recent Amazon Web Services (AWS) defection by Dropbox and a move by Apple to shift some work to Google—shows that AWS continues to be dominant in cloud computing, and Google will have to work even harder to catch up.
First, some background. Google recently bought Diane Greene’s company Bebop and put her in charge of Google’s cloud businesses. With her background founding and running VMware, she should be able to steer some enterprise business towards Google. But she also needs to create an enterprise-sales and support culture at Google to manage and sustain new enterprise relationships.
Google, at its core, has an advertising and media-based culture, and most of the enterprise businesses it has managed to get as cloud customers have, so far, been in these markets. The other market that is growing for Google is retail: Many retailers want to avoid providing business to AWS, Amazon’s mammoth cloud business, as they compete directly with Amazon’s main retail arm. Home Depot, for example, has announced it is going to use Google’s cloud.
When Google restructured and formed its Alphabet holding company, executives left the cloud business inside mainstream Google. They didn’t mention it at all in the announcement unveiling Alphabet. This was terrible positioning for GCP–can you imagine Amazon announcing a re-org and not mentioning AWS?
At the time, I proposed that Alphabet should have created an “E for Enterprise” business and put the Google Cloud and Google for Work businesses inside that unit. Then, the company could work on creating a distinct enterprise sales and support culture. When the company hired Greene, I said it again.
This week company executives talked about how Greene has hired more salespeople and is making Googlers actually go and meet with enterprise customers, which is a good step in the right direction. Google also announced new regions (locations that customers can specify when they launch applications) in Oregon and Japan, and plans to eventually have ten more regions worldwide, an expansion move I have previously identified as critical to enterprise public-cloud adoption. So the company is finally working to match AWS and Microsoft Azure’s global rollout plans.
At today’s GCP Next event, Google brought out Google Inc. CEO Sundar Pichai to introduce Greene and then Eric Schmidt, now the executive chairman of Alphabet. These execs mostly talked about Google innovation and architecture—not new customers or new products. SVP of engineering Urs Hoetzle then talked about the state of GCP, and VP of product Brian Stevens introduced Coca Cola, Disney and Spotify as on-stage guests. The focus, therefore, was on products that the company had already announced—not new announcements—which was disappointing. For new server-less computing and machine learning applications Google have a compelling story, but they don’t appeal to the kind of mainstream enterprise applications that are currently migrating to AWS.
This is all very different from the way AWS runs its public events. AWS chief Andy Jassy traditionally runs the opening keynote at AWS Re:Invent from start to end; AWS trots out customers early and frequently; and the company announces new products along with the customers who have been beta testing them. I wasn’t that impressed with the way Google ran the show. It wasn’t as slick and rehearsed as Google I/O or AWS Re:Invent, and the company over-ran its keynote timeslot through the first set of breakout talks. I didn’t think the company properly leveraged Greene to focus on the moves the company is making to capture new enterprise business. Their customer examples were in line with Google’s existing sweet spot of media and advertising.
Urs Hoetzle talked about what comes next for Google: For one thing, he said “NoOps” application development will get easier with automation from App Engine and Kubernetes. He thinks we need automated security at scale based on immune systems for security deeply embedded in the apps and networks. He made the point that server-less architecture is deeply embedded in GCP. Use of the term “NoOps” is likely to turn off enterprise customers, who have large operations teams. However, his focus on server-less architectures was good, although Hoetzle didn’t mention Google’s new AWS Lambda competitor, Google Cloud Functions.
The Stackdriver monitoring console was the main “new” product Google demonstrated as a beta, though there was no mention of pricing. Google bought Stackdriver last year, so this isn’t a surprise, and the functionality is really a subset of what several other tools already provide. Stackdriver doesn’t provide a compelling new capability for GCP.
I think GCP is making good progress, but it has a long way to go–and if anything, the platform is falling further behind AWS and Azure rather than catching up. They are able to leverage the innovations and scale of the Google mothership for new applications in analytics and machine learning, but AWS is innovating faster in more areas and has huge scale itself, so that’s not enough.
Google is still saying it’s early days for cloud, but I disagree. There is a huge amount of enterprise business up for grabs right now, and they are missing out on most of it.