Google Cloud Platform (GCP) may be rocking, but if so, Alphabet is keeping it a closely guarded secret. As ZDNet’s Larry Dignan has posited, maybe “GCP’s revenue doesn’t stack up to its rivals yet.” No one would be surprised by this, given the lead of several years that AWS, in particular, has had. What would be a nice surprise is to have Google substantiate that its machine learning strength in cloud is paying off with mainstream customers.
SEE: The impact of machine learning on IT and your career (free PDF) (TechRepublic)
What we know
Sadly, to echo Dignan, we know very little about how Google’s cloud business is doing. In fact, it’s as little as we knew a year ago when then Google cloud CEO Diane Greene said everyone was “grossly underestimating” Google’s cloud revenue but then gave the convoluted “$1 billion each quarter” figure that obfuscated more than it clarified. After all, Greene was mixing the more mature (and larger) G Suite “cloud” number with GCP to add up to $1 billion. If anything, it told us GCP was doing worse than expected, simply because she refused to give a real number.
Fast forward a year, and now no one but AWS and Alibaba are willing to share their real cloud revenue data. The one nugget of potential data we got from Google’s earnings call was CEO Sundar Pichai’s comment that “Google Cloud is a fast-growing multibillion-dollar business that supports major Global 5,000 companies in every important vertical.”
Before we rush to judgment that “Google Cloud” is only GCP, Pichai goes on to lump in G Suite…again.
So we know that Google’s overall cloud business (including G Suite) is doing better. We also know that Alphabet “more than doubled both the number of Google Cloud platform deals over $1 million as well as the number of multiyear contracts signed,” said Pichai. Importantly, these deals are happening with mainstream enterprises like Bed, Bath & Beyond, HSBC, and more. Finally, Google CFO Ruth Porat also claimed to be seeing “a really nice uptick in the number of deals that are greater than $100 million.”
That’s a lot of money and a sign that enterprises are taking GCP very seriously.
We also know that, unlike Oracle, Alphabet is investing heavily in data centers to support future growth. In 2018 the company spent $25 billion, up 102%. While Google CFO Ruth Porat said on the earnings call that this growth would moderate in 2019, and while not all of it is data center-related, it’s still a sign of a company that is investing big to win big.
What we wish we knew
So that’s not nothing, but it’s not nearly as transparent as we’d like, particularly because Google’s Cloud revenue is a nice proxy for how well machine learning (ML)/artificial intelligence (AI) are doing in the enterprise. On the earnings call, Pichai called out ML as “driving differentiation” for its cloud business, and cited the UK’s Telegraph as a media company that bought GCP “because of [Google’s] data and machine learning expertise.”
Indeed, it’s because of this differentiation and the customers that come with it that Alphabet spent so heavily on the technical infrastructure needed to enable ML at massive scale. It’s also why Alphabet has been so smart about open sourcing key ML technology like TensorFlow to feed its cloud business.
We don’t really know how well it’s working. Given that Google has done so much to lower the bar to using higher-order ML by mainstream enterprises, it would be great to see if it’s working. We know from The Telegraph’s case study that companies like The Telegraph are using GCP to “analyze digital behaviors to understand audiences,” processing up to 4TB of data in under a minute, taking the former daily processing of advertising campaign performance data from 2.5 hours to 8 minutes.
This is amazing. It’s also not likely unique, but we simply don’t have the details to know how many more case studies like this GCP might be generating. We don’t know the scale of machine learning adoption through GCP, but that data would offer great insight into how ML is doing across the industry, given that Google arguably makes it easier to get value from ML efforts than most other alternatives. As much as we talk about machine learning, the reality is that adoption remains anemic.
Getting more insight into GCP, however, would help us get a better read on gathering momentum in ML adoption.