Each ARM-powered chip has to run some form of software from a simple monitor to an exciting new application. As we view the future, software is both important and complex. It is important since companies have to make software bets on future investments and it is complex because it covers so many different business models, markets and products. I believe the next 100 billion ARM-based chips will be driven by sensor networks. As in sensor networks will create data which will focus future software investments.
To help discuss how I got to this conclusion of sensors and data - I’ll use two examples. The first example is based on a quote which inspired me to think sensors and the second is based on a prediction (my crystal ball method) which drove me towards data and software as an important catalyst.
Let’s first explore the quote.
Growing up in Cambridge (UK) I always heard about Maurice Wilkes, unfortunately I never got the chance to meet him. He is remembered as the inventor of the first computer executing a stored program. Maurice Wilkes is a Turing Award Winner (the computing equivalent of the Nobel Prize). He was quoted as saying:
“Students sometimes ask my advice on how to get rich. The best advice I can give them is to dig up some old algorithm that once took forever, program it for a modern workstation, form a startup to market it and then get rich.”
From "A Half Century of Surprises", in Talking Back to the Machine: Computers and Human Aspiration, Ed. Peter J. Denning, Springer, 1999
Like all quotes you can argue various points - for me this quote effectively was true throughout the 90’s but like all industries the trends change. The 2000’s marked the rise of the ARM ecosystem and mobile computing. Now if we look back at the first 50 billion chips the concept of workstation was pretty much replaced by mobile device. All the latest innovative software targeted mobile devices e.g. the mobile apps stores exploded with popularity leaving the traditional PC software market to slow down. The mobile market has created and will continue to create new innovative software companies. This gives rise to my first revision of the quote, which replaces workstation with mobile device.
Revision 1: “dig up some old algorithm that once took forever, program it for a modern mobile device, form a startup to market it and then get rich.”
The growth in mobile is about providing access to platforms, if you look to the future “where are the next 100 billion chips coming from?” I think the answer is relatively simple, sensors - where the focus is on providing a standard solution across the network. Sensors are the end points of the network where all the information is created and begins. They are increasingly taking on more functionality. Throw in the business concept of the Internet-of-Things (sensor to server or possibly HPC) the world becomes a huge opportunity for the software ecosystem to create new exciting solutions that drive data from device computing to the cloud.
As an IoT example, let’s take a blood pressure sensor that is connected to the internet. Software technology has to be created to allow data to flow from the sensor through the network to eventually end up in a database on a server. This all has to be achieved securely and reliably. This brings me to my second and final revision of the quote (the next 100 billion chips) which takes us further down the path from workstation to mobile device to eventually sensor network.
Revision 2: “dig up some old algorithm that once took forever, program it for a modern sensor network, form a startup to market it and then get rich.”
I’ve talked about sensor networks driving the next 100 billion ARM-powered chips but let’s discuss the real underlying catalyst “data” and the software used to manipulate that data.
“The acceleration to consume more data, more intelligently with more computational-efficiency will continue to prevail throughout the industry for the foreseeable future”
Crystal ball time, when predicting the future it is better to start with one technology growth-constant that forces the other technologies to align and contribute to that constant. Making use of the QUOTE above, I can create an entirely new general prediction that data will drive the next 100 billion chips for the software ecosystem.
My growth-constant here is that data usage will continue to rapidly expand and that will align other technologies (computational-efficiency and intelligence). It doesn’t matter which market is viewed they will all follow the same common pattern. Data growth demand will outstrip hardware technology unless we can improve the software computational-efficiency. This means that there is an opportunity for software to be used to beat the projected Moore’s Law line by side-stepping the escalating costs.
I’m thinking the word cost in this context is all around total cost of ownership (TCO). Example, there is a drive to bring DNA sequencing down to $1, once we get nearer to that goal, more data will be created because more people will want their DNA sequenced – it’s a circle. This is an opportunity for the software ecosystem to take advantage of the situation by accelerating computational-efficiency beyond hardware improvements towards the end goal.
Finally intelligence, we can predict systems will get more intelligent - whether it is a new Natural User Interface (NUI) or the increase in capability at the datacenter. Data growth is driven by making access easier. The reason we see data growth continuing is because data is easier to obtain, modify and use.
The next 100 billion chips will build on the previous 50 billion, so mobile computing remains important but even today we can see the rise of sensor networks opening up huge new areas of opportunities for the software ecosystem with data being underlying catalyst.
The acceleration to consume more data, more intelligently with more computational-efficiency will continue to prevail throughout the industry for the foreseeable future. Taking data growth as a constant we can see the potential of new software revenue across the industry, especially around the alignment of computational-efficiencies and intelligence. This will effectively make data a new renewable energy for the software industry and increase our appetite to consume.
It’s a cycle, future software revenues will continue to grow since more sensors, create more data which requires more chips.
Are sensor networks going to be the driver? Is data the new renewable energy? What is the future? What will drive the next 100 billion ARM-powered chips for the software ecosystem? The future is exciting but speculating on the future is always difficult, so what do you think?
Andrew Sloss, Andrew is a Chartered Fellow of the British Computer Society and Senior Principal Engineer in the Strategic Software Alliances group ARM Corporate Marketing. He focuses on future software challenges and trends which covers software from firmware to HPC. Andrew is the ARM Bindings Sub Team leader for UEFI an international standard. He is also a Part-Time Lecturer at the University of Washington teaching Embedded and Real-Time Systems.