Thomas Kurian (Google Cloud)
Google Cloud has always delivered cloud computing a bit differently. Filled with propellerhead engineers, the technology giant sometimes seemed to be building for über geeks like itself.
Early customers included Snap and Spotify, great companies but hardly filled with Oracle DBAs holding on for just one more year to collect their pensions. Back in 2017, I wondered if maybe, just maybe, Google Cloud should turn this cutting-edge tech cred to its advantage, helping enterprises “run like Google.”
Four years later, however, Google Cloud seems to have found a comfortable middle ground between “scary cool” and “hella boring” services. Describing announcements from the vendor’s Google Next event, Tom Krazit and Joe Williams write, “Wanted the next killer cloud service? Sorry, not this week. In terms of exciting stuff that Google Cloud was actually planning to ship, this was a light event.”
This is not to say Google had nothing to announce -- quite the opposite. The real story of Google Next is just how much it seems to finally understand the importance of being a boring enterprise software company. The real story on cloud these days is who can integrate diverse cloud services best. The cloud is about to become much more interesting, precisely by becoming much more boring.
Making cool boring
First, it’s important to recognise just how much Google Cloud has done to help the industry to, yes, “run like Google.” Consider Kubernetes. As suggested in a recent Kubernetes documentary, “Google had to make a bold move in the cloud space to be the long-term winner.”
By market share estimates, Google Cloud is not the winner in cloud today, but “today” isn’t very interesting in a market that still registers just six per cent of overall IT spending. By open sourcing and then fostering a vibrant community around Kubernetes, Google Cloud gave itself a seat at the table that it didn’t otherwise deserve. It wasn’t the first mover (that was Amazon Web Services) and it lacked trust with CIOs (that was Microsoft Azure).
By enabling enterprises to become dramatically more productive with containers, Google Cloud helps connect developers to its vision of how all enterprises should build. Yes, this means a bigger market for cloud rivals such as AWS and Azure to consume, but it also creates a vastly bigger slice of the market pie for itself.
The same is true in artificial intelligence (AI) and machine learning, areas where analysts and others regularly give Google Cloud the lead. Data science is finally becoming a thing, as I noted recently, precisely because data infrastructure has finally become accessible by mere mortals. Google Cloud has played a big role in crafting this data infrastructure.
It has also played the long game, open sourcing TensorFlow and other tools to help make data scientists more productive.
This is thoughtful strategy, as Brookings Institution Fellow Alex Engler argues: “By making their tools the most common in industry and academia, Google [with TensorFlow] and Facebook [with PyTorch] benefit from the public research conducted with those tools, and, further, they manifest a pipeline of data scientists and machine learning engineers trained in their systems.
"In a sector with fierce competition for AI talent, TensorFlow and PyTorch also help Google and Facebook bolster their reputation as the leading companies to work on cutting-edge AI problems.”
Through Kubernetes, TensorFlow, and various data infrastructure initiatives, Google Cloud has managed to make its rocket science approachable to mainstream developers. As such, it might have made sense for the company to announce a “gee whiz” service or two at Google Next 2021. It didn’t, and that shows the other part of Google Cloud’s strategy.
Making boring cool
Most of what was announced at Google Next were customers. Lots of them, and not the Snap/Spotify type (though there were some of those). Deutsche Post DHL, General Mills, Siemens Energy, Walmart, and Wendy’s, to name a few.
Focusing on Walmart during the event was almost certainly a slap at AWS/Amazon, but the main story was that Google Cloud appeals to snoringly dull companies with safe, predictable services. In the enterprise world, boring is very, very cool. As Google Cloud’s Richard Seroter told me, “It’s no longer about shipping a slew of niche or duplicative services. It’s about completeness and broader concerns” like security.
Yes, Google Cloud also announced the Google Distributed Cloud (see Scott Carey's analysis), which, as Forrester analyst Jeffrey Hammond points out, could prove to be anything but dull: “The distributed cloud preview and the Postgres support for Spanner will both turn out to be pretty important. The next wave of cloud-native apps will be serverless and planet-scale.”
But such things were already on offer from AWS (Outposts) and Microsoft (Azure Stack). Google wasn’t breaking new ground.
Much of what Google (and the other cloud providers) need to do now is make all those cool cloud services play well together, and not merely in their own walled gardens. It’s already apparent that most enterprises won’t buy everything from just one cloud provider, though they may well choose a predominant vendor.
Asking customers to only buy one cloud’s services is asking them to believe that one cloud will be the source of all innovation. That’s silly. As Sun Microsystems Cofounder Bill Joy once opined, “No matter who you are, most of the smartest people work for someone else.”
The ultimate cloud winner will be the one that enables the richest ecosystem, with innovation coming from its own services, open source communities, and partners (who will almost certainly compete with those first-party services). By “enable” I don’t just mean “it’s available if you dig deep enough in our marketplace.” No, I mean deep integration with the services that customers want most.
The next wave of cloud innovation will come from integration of the old rather than relentless shipping of the new. Game on.
