We’re used to associating Google with the best. The cutting edge. However, this dominance has yet to assert itself in the race to dominate cloud computing. A race currently being led by Amazon Web Services.
In constructing the most valuable company ever, Google has amassed an unparalleled cloud stack, appealing to end users and developers alike. Yet, at least when it comes to owning enterprise advancements in the cloud, Amazon and Microsoft – not Google – are simply better positioned to dominate enterprise-focused cloud services. To keep up with Amazon and Microsoft, Google has prioritized machine learning as its point of differentiation, most recently by publishing a set of machine learning APIs to address natural language, speech, vision and translation. Ben Thompson, over at stratechery.com expertly addresses the market positioning and points of differentiation of the cloud giants on his blog. However, as far as enterprise adoption of the cloud is concerned, Google’s pursuance of machine learning to make headway against other industry giants is more nuanced than simple tech investments and open-sourced machine learning APIs.
Google’s biggest opportunity does indeed lie with machine learning, but its greatest weakness also relates to sourcing the information that makes machine learning useful to enterprises. For Google, productizing machine learning for business applications (as opposed to natural language and vision) depends entirely on the type of data it can capture. Google dominates in terms of harvesting data available for its own products, and can use this information to solve problems related to speech, search, vision, etc. It will be years before more enterprise organizations develop applications capable of matching Google’s prowess for machine-learning cloud advancements, or, more importantly, attract the kind of technical talent with the skills to make use of it
However, Google struggles to gather and make use of key data points outside of its own network. This data is what’s considered “first-party” to an enterprise and is highly guarded – generated by the myriad operational systems currently sitting in various silos across expensive on-premise data centres. And based on a recent study by McKinsey, “the overall share of enterprise workloads in the public cloud is still in the low single digits.” Unfortunately for Google, this is where the data required to solve many of an enterprise’s problems actually worth solving exist. On the plus side, however, based on the aforementioned McKinsey study and my own experiences in the marketplace, attitudes are certainly shifting.
Machine Learning Opportunities Lie With Businesses
Of particular interest in today’s environment are problems that impact business profit and loss (P&L) directly, such as sales forecasting, promotion optimization, customer lifecycle management, etc. Google’s highly integrated consumer-centric business model (that has led to such amazing products like search, tag management, google analytics, etc.) offers little experience navigating these business needs. Google needs first-party transactional data to solve P&L problems, not image, speech or text data.
Obtaining entirely new sources of data may seem daunting, but it’s a task well worth its rewards for Google. In fact, Google’s rumoured interest in a dunnhumby acquisition may have been motivated, partly, by the opportunities lying in enterprise first-party data.
While the company struggles to collect first-party transactional data, retailers and other enterprises have invested millions of dollars in interfaces and systems that capture messy first-party data – from point of sale machines, loyalty programs, merchandising systems, and a host of other operational systems. And most importantly, they have done little in effectively consolidating all this data for impactful applications. This is an area where Google’s Cloud Platform can be leveraged.
The Amazon Paradox for Enterprise Retail
As a technology company that dabbles in retail and other services, Amazon is predisposed to excel at profit-focused problem solving using machine learning. Amazon has the experience and technologies (for example, Amazon Go) to gather and make use of first-party transactional data, among other data, to create impactful analytic solutions. However, Amazon is also the thorn in many a retailer’s side. Save for a few leading-edge retailers turning to Amazon’s vast distribution capabilities to expand their own operations in geographies where brick-and-mortar solutions are a poor growth option, handing data over to Amazon would amount to suicide.
However, Google (and Microsoft for that matter) can gain ground on Amazon Web Services. And they must do so quickly or risk losing a key aspect of cloud-computing market share. There’s no denying that Google intends to become more of a B2B product company. To do so, however, Google first has to convince enterprises retailers, telecommunication companies, banks, and other large enterprises to entrust Google Cloud with their first-party data.
Second, and in order to remain competitive, the enterprise cloud giants will need to partner with third-party providers that specialize in making the most of these data-sets. Rubikloud’s enterprise data platform, for example, is engineered with all the scalability, elasticity and retail specificity to both pipe massive quantities of operational data to the cloud and deploy machine learning driven solutions. Rubikloud’s own suite of machine learning solutions, furthermore, address P&L concerns directly. And, most importantly to the cloud providers, Rubikloud is a cloud-first platform. Enabling access to vertically-integrated, machine learning ready platforms like these are exactly the kinds of differentiation that will make a cloud provider a destination for large enterprises.