Something has been bothering me off and on. It started with a speech Charles Stross made years ago trying to imagine what would happen once the electronics revolution ended.
My take is this, if your work has anything to do with computers, telecom or IT you could be unemployed in 15 years. This seems like a rather long time, after all you could be unemployed much sooner than that for all sorts of reasons, but do listen.
To give a little background, in 1965 Intel co-founder Gordon Moore observed that the number of transistors they could fit into a microchip doubled about every two months. Much to everyone’s surprise, this held true ever since and is enshrined as Moore’s Law.
Time after time as the industry confronted seemingly insurmountable obstacles, people predicted the end of Moore’s Law, but they were proved wrong as the industry adopted new techniques and technologies, shrinking transistors further.
Smaller transistors are faster transistors and you can fit more of them in a microchip. Unfortunately, the faster you run a microchip, the more heat it produces. Between 1995 and 2005, Intel increased its processor speeds from 200MHz (200 million clock cycles per second) to 4GHz (4 billion clock cycles per second). A piece of arithmetic typically takes only a couple of cycles to do, so you can see how powerful these machines became.
By this time, you could warm your coffee with the heat your computer made. So chip makers switched gears, in 2006, they halved the clock speed but put in 2 computer cores inside 1 chip. This has been the trend since then with even mobile phone chips today having as many as 8 cores.
This year, marked the last ever ITRS (International Technology Roadmap for Semiconductors) report. In 1993, major US tech companies like Intel and IBM joined to coordinate their planning effort for the future growth of their industry. This meant all equipment and material suppliers, all major customers, designers and industry players could anticipate and prepare to meet the requirements for the next decade. This period is coming to an end as each company goes off to pursue its independent research. As technology change becomes incremental, there is less need for coordination.
In its last report, ITRS predicted that transistor sizes could stop shrinking by 2021, almost a decade sooner than their 2013 prediction.
This is coupled with another trend. In 2001, 19 companies made computer chips. Today there are 4, Intel, Samsung, Global Foundries and TSMC. In other words they seem to be following the classical consolidation pattern when an industry matures and advances become fewer and marginal. Moore’s law is showing its age and may die soon.
The times, they are a-changin
This does not mean that computers, phones, watches etc. will stop getting better. There are still a loads of tricks engineers will continue to try which could fuel improved performance for say another decade. Just in time for my generation to reach its mid-career crisis.
That is why I am interested in this now.
Let’s talk about the tech ecosystem. The telecom giant Ericsson is cutting about a fifth of its workforce. Most of these people worked in older, less profitable technologies. Technologies which are facing increased competition, lower profits and need fewer engineers because, well most of the work on them has already been done.
Ericsson and a lot of other telecom players are counting on 5G technology which is expected to take off around the world sometime around 2020. This promises to be many times faster and more efficient than the current 4G which we are just getting used to.
This illustrates a key point. All large tech profits (and jobs) come from exploiting the latest innovations. Practically all these innovations are powered by improvements in computing technology. All app development, social media or video streaming at their core rely on increasing bandwidth and computing which would have been impossible 10 years ago. It is only because phones became cheap, powerful and could reliably access the internet that somebody though of starting a food delivery app.
Without ever advancing computing power, it becomes increasingly harder to find new untapped niches where a small company can survive. In other words, if you want a lot of start-ups, you have to keep generating new frontiers.
Worst hit is going to be software. Unlike hardware, cars or shoes software doesn’t need to be changed every few years. When the hardware stops evolving, so will the software that exploits this hardware. Only whereas the hardware may need to be replaced once in a while, the software can be re-used as often as needed with no need of a programmer.