Google is positioning itself as the best place for AI workloads, but it's facing stiff competition from AWS and Microsoft.
Google CEO Sundar Pichai may acknowledge that the future is multi-cloud, but that doesn't mean he thinks all clouds are equal. In a revealing comment during the company's Q2 earnings call, Pichai dropped this bombshell: "[T]here is a tremendous cost to your business of being on the wrong [cloud] architecture."
In other words, Pichai believes going with anything but Google is something of a dead end.
Now, if only he could convince the majority of enterprises that have elected to go with cloud market leader Amazon Web Services (AWS), or leading contender Microsoft Azure.
Smart money on AI smarts
Which is of course what he spent the earnings call, as well as the company's Google Next conference, trying to do. Much of the focus of the earnings call (and conference) was just how much the company is spending to improve and expose its machine learning smarts. Lest this get lost, Pichai called it out: "The common thread you'll hear on today's call is the benefit of machine learning and AI, and how it's improving our products and generating great results for our users and partners."
Though Google is, at heart, an advertising business, Pichai didn't want to restrict machine learning and artificial intelligence (AI) to improving search ads. Instead, he talked about how the investments in machine learning extend to the cloud, touting Google Cloud Platform (GCP) as: "a natural extension of our long-time strength in computing, data centers, and machine learning."
SEE: Cloud migration decision tool (Tech Pro Research)
Those improvements haven't come cheap. Though not exclusively tied to GCP, Alphabet's capital expenditures (CapEx) jumped 95% year-over-year to $5.5 billion in the quarter. At the same time, Alphabet's biggest area for operating expenditure (OpEx) increases has been on its cloud business, with "the majority of our headcount growth...in technical roles and engineering and product management" within GCP, Google's Ruth Porat noted on the call.
It's clear Google/Alphabet is spending big to make a play for cloud workloads. It's also clear that the differentiated value it's peddling is "supercharged information," as Google Cloud chief Diane Greene put it in a blog.
They're doing it wrong
What's also becoming clear is that Google doesn't plan to simply pitch the virtues of its machine learning smarts, but also plans to swat down so-called cloud pretenders to the AI throne. Hence, Pichai's comment that "[T]here is a tremendous cost to your business of being on the wrong [cloud] architecture." Namely, any cloud but Google's.
Microsoft and AWS aren't likely to be persuaded, and both have far more customers than Google. Yet this still feels like a winning marketing strategy for Google. The Google brand exudes "rocket science." As consumers, we've grown up trusting it to spit out smart answers to our most inane queries. It's not hard to see those consumers-qua-enterprise buyers assuming that Google would be the ideal cloud for machine learning smarts.
And yet this might not be the real battleground. Half the battle is convincing the market that a particular cloud has the best machine learning and AI assets, but the other, and arguably harder, battle is to convince those same customers that a particular cloud can make those same technologies approachable for mere mortals. Google recognizes this, and has taken great strides with AutoML.
As 451 Research analyst Nick Patience has written: "Right now, a lot of knowledge is required to build custom models, and given there are only roughly one million data scientists in the world (compared to about 26 million developers of all kinds), many companies, including Google, see a major opportunity in enabling developers other than data scientists to be able to get their hands dirty with machine learning."
With the introduction of AutoML, for which Google has announced further advances this week, Google has pushed machine learning toward a more drag-and-drop experience, rather than hard-core coding in the command line.
SEE: Deep learning: An insider's guide (free PDF) (TechRepublic)
Google, however, doesn't have Microsoft's pedigree of handholding enterprises through difficult and complex systems. Microsoft's entire hybrid cloud strategy is centered on this fact: Change is hard, it takes time, and it requires a lot of vendor assistance. Amazon, for its part, has less experience with old world computing, but has proven itself a quick study in both figuring out hybrid as well as introducing relatively simple machine learning interfaces or services (like Polly) for mainstream enterprises.
In short, Google is right to position around its machine learning smarts, but it has strong, credible competition here. As such, we're likely to see real innovation, both in terms of machine learning features and accessibility, coming from the big three, benefiting enterprises across the board. In the case of machine learning, however, it's very possible that Google will be setting the pace.
- Deep learning: An insider's guide (free PDF) (TechRepublic)
- Moving fast without breaking data: Governance for managing risk in machine learning and beyond (ZDNet)
- Google Cloud Machine Learning Engine: The smart person's guide (TechRepublic)
- AI in business: Making machine learning work for customers (ZDNet)
- Breaking the tech giants' stranglehold on AI: Why machine learning needs to be decentralized (TechRepublic)