As tech clichés go, “big data” falls somewhere between “cloud” and “social” on the list of terms you’re probably tired of hearing — ideas so broad they mean hardly anything at all. Nevertheless, since we first looked at the growth of predictive analytics last year, spending in that area has only increased.
The redundantly named but allegedly reliable investment research firm Markets and Markets predicted this year that corporate spending on logistics analytics — big data for supply chains — should grow from slightly more than $7 billion this year to more than $29 billion in 2019. That’s an overall corporate figure, but electronics appears likely to play a central role to the boom.
Just last week, a Royal Bank of Scotland analyst told Financial Times (subscription required) that interest in predictive analysis spiked after the infamous electronics supply chain break that followed the 2011 earthquake and tsunami in Japan. The disaster disrupted the world’s supply of a broad range of electronic components. One particular cause of panic was the gargantuan chip orders pending at the time for the auto industry. Autos are seen as a vanguard product in integrating electronics, and the disaster in Japan served as a warning for other industries. As home appliances, houses, and even bathroom fixtures become increasingly wired (the vaunted Internet of Things), supply chain risk in electronics becomes supply chain risk for everything.
Could big-data analytics really prevent logistical breakdowns from a devastating, random event like the disaster in Japan — or even a lesser threat, like a sudden shift in currency exchange rates or fuel prices? In theory, yes. Where old-school modeling might take into account the likely sources of occasional interruptions (Japan’s seismic risk isn’t new, for example), the only real remedy for a supply chain manager is to calculate the possible costs and warehouse accordingly. That is a hit-or-miss process, and it’s expensive if your ballpark figures are wrong.
In just the past 18 months, a number of analytics startups have shifted the emphasis toward using big data to source new supplies in real-time as interruptions are happening. (One of the shops we highlighted last year, Precogs, has gone on to prominence.) Looking at production and consumption in more than simple supply/demand terms, and correctly predicting where production is most likely to create overruns and shortfalls — even before the factories themselves know — significantly improves the odds of successfully responding to surprises.
By how much? Even the analytics firms can’t predict that yet. But the spending increases on real-time data suggest a gelling consensus that the costs, which are considerable, are worth it. And with electronics supply chains likely to become more intertwined with non-electronics industries, those costs are spreading wider. As the scale grows, the cost of collecting and analyzing all that data will explode. But so will the cost of not having it. Soon, it will stop being a choice at all.
