This article is more than 1 year old
Sorry Google, it's boring old workloads that are pumping up AWS and Azure, not sexy AI
Spend big or go home
Comment Google Cloud Platforms's chief thinks the service will surpass AWS by 2022. Speaking at Forbes CIO Summit in Half Moon Bay, California, last month, Diane Greene claimed Google has "a huge advantage in our data centers, in our infrastructure, availability, security and how we automate things. We just haven't packaged it up perfectly yet."
As Google works on that packaging, Microsoft chief executive Satya Nadella assured that Microsoft will compete by complementing its IaaS and PaaS offerings with SaaS products like Office 365 and Dynamics – products for which AWS currently has no answer.
AWS, meanwhile, maintains a hefty lead on its rivals, even as increasing competition between the big three continues to bloody growth and/or profit all round. Normally, price competition comes as products commoditise. But in cloud, would-be competitors must spend big just to get in the game, and then run on thin margins to remain competitive.
It's damned if you win, damned if you don't. That "damnation" promises to pay big dividends, however, to the company that can figure out how to move enterprise workloads into the cloud.
A game only the rich can play
Last year Amazon, Microsoft and Google collectively spent $31.5bn loading up on data centres. While Oracle has claimed its paltry $1.7bn is enough to compete (because, of course, its data centres are super-duper), distinguished AWS engineer James Hamilton eviscerated that silly notion. Each of the big cloud providers has invested in highly tailored infrastructure to juice performance, but this is only part of the answer. He writes: "Extraordinary database performance won't change most of the factors that force successful cloud providers to offer a large multinational data center footprint to serve the world."
To be a big cloud competitor, in short, you have to be big. There simply is no way to budget one's path to a winning cloud position. Indeed, Google CEO Sundar Pichai told Wall Street analysts that Google's "largest growth in headcount and CapEx was in Cloud," giving a taste for the costs associated with keeping up with the cloudy Joneses.
Nor does this get much better once the largesse has been established. AWS implemented 51 price cuts by the end of 2016. Rather than wait for competition to catch up, AWS makes the cuts continually as it streamlines performance. While Google doesn't break out its Cloud revenues and profits, Microsoft owned up to the impact this price competition is having. As noted on the company's earnings call, such cuts took a toll on Microsoft Azure's profits in the last quarter, pushing the company to respond with higher-value SKUs that (it hopes) help it escape otherwise bruising price competition.
In search of differentiation
Still, as Nadella stressed in the earnings call, looking at Azure profits in isolation can be misleading, both because R&D or hardware spending on Office 365, for example, ends up benefiting the Azure team, and because customers who may choose Azure for a commodity workload end up making a strategic bet on Dynamics. The two feed each other, and looking at Azure's P&L without this context doesn't give a complete picture.
Not that Microsoft is alone in this. The heart of Google's cloud attack is AI and machine learning, which Google learns "on the job" with its own internal services built to support its gargantuan advertising business. Ditto AWS, which said in its earnings call:
We're also investing in new technologies such as artificial intelligence, machine learning. You're starting to see some of that show up in things like Amazon Go, our beta store that we have developed in Seattle, drones. We use those technologies a lot in our internal businesses and we're also developing services for AWS customers.
Cloud, in short, is a complement to other businesses at Amazon, Microsoft and Google, just as those businesses complement their cloud products. But each company brings its own particular spin, pointing to the future of public cloud competition: head-to-head competition without going head-to-head directly.
Betting the cloud on AI... or not
Well, that is, except for AI/machine learning, which was top of mind for each company as they reported earnings. Google wants to position itself as AI‑first. Microsoft says that companies look to it to "digitally transform with the state-of-the-art cloud services, AI, and new natural user interface technology". Ever cost conscious, Amazon's Jeff Bezos wrote to shareholders [PDF]: "We're excited to lower the costs and barriers to machine learning and AI."
Yet the problem is that most enterprises simply aren't ready for serious AI/machine learning projects, however much we like to pretend otherwise.
Fifty per cent of enterprises claim to be ogling machine learning in 2017, by one estimate. The problem with such numbers, insists Gartner analyst Nick Heudecker, is that they're completely wrong. As he grouses, the "same was said for big data, but only 15 per cent of orgs get to production. Will likely be much lower with ML."
Now factor in a minuscule 5 per cent of total enterprise workloads running in the cloud (public or private), according to McKinsey, and it's clear that the vast majority of the cloud market has yet to be won, and likely won't be won by cutting-edge ML projects that appeal to a lilliputian segment of the overall market. No, the real money in the cloud is going to belong to the vendor(s) that can help enterprises take baby steps.
Which is one big reason to believe that AWS's reign will come under increased pressure.
Microsoft, given its strength with traditional workloads, is actually extraordinarily well placed to capitalise. A Goldman Sachs survey of IT decision-makers found that Microsoft is CIOs' most strategic vendor today, and is expected to remain such for the next three years (AWS comes second).
Given that CIOs trust Microsoft and already give a large percentage of their workloads to them, it's reasonable to expect Microsoft to win a greater percentage of cloud workloads going forward, even despite the AWS head start.
Google, which is expected to be tech executives' fourth-most strategic vendor within three years (it currently doesn't make the top 10), has a whole lot of big data expertise going for it... and against it. Google may struggle simply because it is outpacing what enterprises feel capable of digesting. Everyone may want to run like Google, but vanishingly few are even remotely capable of doing so.
Which leaves us with AWS, the company doing most to define the enterprise of tomorrow. Clearly AWS will continue to lead. What isn't as clear is for how long, and by what margin. Its growth, while still impressive, has slowed over the last few quarters. Unlike Microsoft, AWS must build relationships with CIOs, something they're keen to do given its cloud leadership position.
Ultimately, it feels like the public cloud is coming down to a drag race between the cloud of the future (AWS) and the cloud that helps bridge the future with the past (Microsoft Azure). Everyone will talk about how their cloud is the best platform for this or that gee-whiz machine learning or AI workload, but the better strategy will be pitching their clouds as the best place for the boring enterprise workloads that actually power enterprises. ®