Counting By The Billions

The semiconductor industry has been on cruise control since the advent of the personal computer. By 2002, a total of 1 billion PCs had been shipped, according to Gartner, and by 2008 that number had doubled. But that was nothing compared with the smartphone. In 2014 alone, Gartner reported sales of 1.2 billion smartphones.

Both of those markets will remain healthy for years to come. Despite predictions of the PC’s demise, the less-expensive tablet’s emergence, or of the smartphone’s imminent slowdown, PCs and smartphones still account for a huge amount of semiconductor content—processors, memories, storage, PCBs, hardware accelerators, discretes, embedded software, and much more. People still buy new phones every couple of years, and they need new computers every four to five years.

What’s worrying semiconductor companies these days isn’t that there will be fewer chips developed each year. It’s that there will be slower rates of growth in markets that consume the same chips by the billions of units. Those markets, fueled by a sure path to shrinking features and reducing costs every two years, have generated the largest concentration of profits in a given time span in the history of business.

They also have fostered a disaggregation, which has created wealth across a wide swath of technology companies. Specialization is always the most efficient and cost-effective way to get products to market on time. The re-aggregation of the industry, both in the current wave of semiconductor and IP company consolidation, and the move by companies such as Apple to take more chip-making in-house, are clear signs that this business cycle is ending.

What comes next is hardly bad news for the semiconductor industry. The Internet of Everything, including connected cars, and a slew of medical devices, will result in billions more chips being sold. Those numbers could rise into the hundreds of billions of units. The difference is they won’t be the same chips, so the economies of scale will have to come from somewhere else.

This represents a whole new challenge because the next phase isn’t about shrinking features every couple years or adding more functionality onto a chip. It’s about customizing chips for specific markets quickly, packaging them in new ways, and building all of this for a reasonable cost with low power and an overriding security architecture.

These aren’t just application processors, although those will certainly be part of the mix. They include everything from sensors connected to logic and memory for measuring the flow of liquids through a pipe to advanced and very flexible processors used in cloud-based servers, and everything else in between. And none of them will be made in the kinds of volumes that have defined the PC and smartphone markets, even though in aggregate the volume of IoE chips will outnumber them by a factor of 10 to 100, depending upon whose numbers you believe.

What’s important in these markets isn’t necessarily the speed of the chip. It’s the speed with which the chips are designed, verified and manufactured in small batches. That requires new applications of tools, new packaging approaches, more emphasis on platforms, flexible ways to connect everything together, new architectures, new equipment and new ways to use existing equipment.

If all works as planned (and the plans are still being formulated), the great industry consolidation of 2014-15 will look more like a retrenchment than a contraction. Growth will come from many new areas and markets, and probably new companies, which will use design tools, equipment and even IP far differently than in the past. For the companies that grasp these changes, and which can position themselves to take advantage of them, the opportunities are huge. For those waiting for the next billion-chip-per-year markets, it might be time to stock up on water and extra batteries.