Internet Access Speed Growth is Linear, but in a Moore's Law Way

You might not know it from the stream of quarterly updates on “average” Internet connection speeds around the world, but a long history of speed advances confirms that consumer Internet access grows about as fast as Moore’s Law would suggest.

So even if it seems very little is happening, quite a lot is happening, all the time. You couldn’t tell that from quarterly or even annual changes in typical access speeds.

In the third quarter of 2014, for example, global average mobile Internet connection speeds dropped 2.8 percent to 4.5 Mbps, and the global average peak connection speed fell 2.3 percent to 24.8 Mbps in the third quarter of 2014, according to Akamai.  

On an annual basis, average mobile Internet connection speeds globally were up 25 percent from the third quarter of 2013, though. That implies a doubling of speed about every four years.

Most people would likely agree that usage grows faster than that.  Based on traffic data collected by Ericsson, the volume of mobile data traffic grew by approximately 10 percent between the second and third quarters of 2014, implying annual growth of more than 40 percent. But that’s usage, not average speed.

The global average fixed Internet connection speed saw a slight decline in the third quarter of 2014, dropping 2.8 percent to 4.5 Mbps. Global average peak connection speeds declined slightly in the third quarter, dropping 2.3 percent to 24.8 Mbps.

Those sorts of figures are hard to square with the notion that typical speed doubles about every 18 months to two years.

Logic seemingly would suggest that is unlikely. Communications networks--especially those of the fixed variety--are expensive construction projects. Such networks also are subject to local, state and national regulations, interest rates, economic conditions, changes in tax laws and changes in demand curves, all of which should slow rates of change, compared to rates of change for semiconductor products that follow Moore’s Law.

Shockingly, then, some studies have shown that even on twisted-pair copper telephone networks, speed doubled about every 1.9 years.

Other studies show similar results: some say an Edholm's Law shows that Internet access bandwidth does increase as Moore’s Law would predict.

Of course, experts have argued for decades about whether Moore’s Law would end. That debate still hasn’t been settled. But some argue that communications bandwidth would continue to improve on a Moore’s Law pattern, even if classic Moore’s Law slowed or flattened.

That’s a foundational assumption. If access bandwidth really does grow at Moore’s Law rates, then gigabit access networks are inevitable, no matter how crazy that seems.

But that is going to obvious first in the developed regions that have been at it the longest, in North America, some portions of Asia (Japan, Korea, Taiwan, Singapore) and parts of Europe.

Other regions with tougher economics might still be on the curve, but will start at slower speeds, as did Internet access in the more-developed regions.

The global broadband adoption rate (at least 4 Mbps) edged up slightly in the third quarter, gaining one percent and growing to 60 percent.

The global adoption rate of access at speeds of at least 10 Mbps was up 22 percent in the third quarter, following 65 percent increases seen in both the first and second quarters of 2014.

South Korea had the highest average connection speed at 25.3 Mbps but Hong Kong

again had the highest average peak connection speed at 84.6 Mbps.

Demand is going to grow as well, given both streaming popularity and new video formats including 4K video. With 4k adaptive bitrate streams generally requiring between 10 Mbps to 20 Mbps of bandwidth, markets where 4K streaming is widespread will face new investment requirements.

Though it seems improbable, and even when quarterly or annual statistics do not fully show the progress, Internet access speeds do grow about as fast as Moore’s Law would suggest. It’s astounding, really.