For all that’s been said about data capital, its impact has been extremely isolated. A few unicorns have pranced onto the scene, churning data into profits, but most have had their horn shorn off.
A few established companies have seen their digital initiatives bring true transformation, but most are disappointed with the results. And the Organization for Economic Cooperation and Development (OECD) says that overall productivity growth has bumped along at less than one percent for the past nine years.
Yet data is undeniably important, and companies are doing their utmost to take advantage of it. So what’s the hold-up? The answer lies in the nature of technology – cloud computing in particular – and the structural impediments of business technology that prevent businesses from taking full advantage of their data resources.
The basic premise of cloud is that you rent someone else’s computers so you can build and run your own software. You can build anything you want -- as long as you’re prepared to manage the technology yourself.
But real economy companies, like retailers, manufacturers, and healthcare providers, need to forget about both the hardware and the software so they can focus on the data – on transforming it into real capital. This is why businesses need autonomous cloud.
In an autonomous cloud, you build what you need, and try to automate the rest. It’s not just the infrastructure that has to manage itself; the operating systems, databases, diverse storage methods, even applications and analytics have to tune, secure, and repair themselves without human intervention. This requires machine learning embedded in every service that constantly observes its own activities, improving its future performance even as it accounts for changing workloads, data structures, and security threats.
To understand why the autonomous delivery and protection of data is so crucial, you have to look at the economic identity of data and the way that plays out in large firms.
Data is an economic factor of production in digital goods and services. If you design a new fraud detection algorithm but lack the data to train it, you can’t build that service. It’s like if you design a new office tower but lack the financial capital to pay for it. You can’t build it.
This means that data is a kind of capital—an asset, but a strange one. So strange that even though The Economist declared data the world’s most valuable asset a few years ago, accounting rules still won’t let you put it on the balance sheet.
Data lacks market signals
But the oddest thing about data as an asset is that the vast majority of it never goes to market. Instead, most data gets used by the same firm that creates it. This is like oil companies becoming multi-billion dollar firms by burning all the crude they dig up instead of selling it.
This means that almost every large firm pursuing digital transformation is also considering how to expand an internal data economy. The supply side comes from applications, sensors, and devices creating data assets not just in greater amounts, but in ever-greater diversity. The demand side consists of a growing number of increasingly diverse analytical models and algorithms applying that data in a wider array of business processes and decisions.
However, since the supply and demand of data are inside the same company, there are no market prices to indicate which datasets are more productive than others for a given use, or which might be of higher quality, or maybe more recent, and so on.
Without market signals to set prices for datasets, or market competition to bring those prices down, most companies are suffering from high transaction costs between the supply and demand sides. The costs to get any dataset from its point of creation to its many points of use are likely higher than they should be. It’s hard to know because most companies don’t measure their data creation and use this way.
Bringing down the cost of data
In that case, how do you bring down the time, cost, and effort to get data from its point of origin to its many points of use? By automating every possible linkage in that web of interconnections, from harvesting the metadata of diverse datasets into a centralized exchange, to tagging, enriching, and transforming datasets to suit target analytics and AI models, to model selection and hyperparameter tuning, all the way to the automatic creation of interactive visualizations for executives and managers.
When data capital flows become frictionless – when the costs of creating new data from existing data are radically lowered – then companies increase the return on this asset. This may come in the form of increasing revenues as a result of data-driven actions and decisions, or of decreasing the costs to create and use these data assets, or both.
To begin this journey, you have to start with the fundamentals. This is why Oracle first brought to market the Autonomous Database, and then Autonomous Linux, the operating system to run it on. And these autonomous services rely on an ML-infused cloud infrastructure that can deliver fast, guaranteed performance surrounded by layered security with multiple forms of automated threat detection and sequestration.
But the point of all this technology is to fade into the background. By reducing the time, cost, and effort of getting data from its point of origin to its many points of use, more companies will see a real payoff from digital transformation, leaving us all better off.