Moore’s law states that the number of transistors in a CPU doubles every two years, outpacing the rate of most any other physical process. This has had an enormous impact on computing, with CPUs now containing billions of transistors and running at speeds in the Gigahertz range. In short, it is because transistor sizes are continuing to shrink (or stay constant) and CPU speeds are increasing that you can run modern software on your current-gen computer. In business decision-making, managers typically examine the two fundamental factors of: the cost of selling a product and the cost of delivering a product to customers. With traditional software delivery, there is one additional factor that impacts both of these costs.
This is a subject I have discussed before, with a longer treatment here. One misperception that holds firm across all industries is that Moore’s law will continue to hold true indefinitely. Indeed, it has been 13 years since Intel’s 36-nm fabrication process began. But the laws of physics do not confer immortality on any process, and certainly not on a world where transistors are manufactured in rooms the size of your living room.
Five Secrets About Enterprise Software That Has Never Been Revealed :
1. Moore’s law is dead –
This is one of the most underappreciated truths in computing, but the scientific and business communities have yet to come to terms with this. While we may still be able to rely on CPU speeds doubling for the next few years, it will be far longer than that before transistors are manufactured at a density high enough to keep that pace up. CPU manufacturers are currently playing catchup with Moore’s law, releasing new manufacturing processes every 2 years (instead of every 18 months). As soon as they can get ahead of Moore’s law again, look for their cadence to slow down substantially.
2. Moore’s law is not the only problem –
The real problem is not Moore’s law, it’s the fact that this exponential increase in transistor density has not been matched by an increase in bandwidth to the rest of the system. It is because of this that software has begun to stall out while hardware continues to progress. This doesn’t mean that individual operations aren’t getting faster every year – they are. But they are getting faster at a rate far slower than transistors (and therefore CPU speeds), and bandwidth improvements cannot match what Moore’s law allows for. In short, this means that all CPUs (and therefore software) will essentially hit a “speed wall.”
3. Software is the new industrial revolution –
The way Moore’s law is changing industry forces a shift in how we think about software. To understand why, consider an analogy with the industrial revolution. In the 1800s, steam engines were poised to replace muscle power in industry, but it wasn’t until the 1870s that steam-powered textile mills began to take over from water-powered factories.
It was by this time that electricity became a viable alternative, and so electric motors took over for steam engines replacing water for generating power for industry (ironically enough using some of the very same components). If you were an industrialist in the 1870s, you were either in the steam engine business or the steam-powered textile industry.
Software is now the industrial revolution, and for the same reason. CPUs are a necessity for running most software today; the rest of it will be able to run on hardware that doesn’t use CPUs. The traditional business model that dates back to Windows 3.1 (run once, sell many times) is no longer viable – instead we will look at how to make software more like electricity than oil.
4. The software equivalent of the Internet will change the world –
Software is where the “Web” came from. Most people don’t realize that the first Web browser, Mosaic, was designed by a National Science Foundation student in 1993 and released in 1994 (not until 1996 did Netscape release what would become the dominant browser). It was ten years later (1998) when Netscape released Prodigy, which took advantage of Mosaic’s capabilities. Prodigy continued to be updated and extended over time, introducing a string of impressive innovations such as pop-up ads between pages and user profiles.
During this same time, the World Wide Web Consortium (W3C) was formed to standardize HTML and eventually led to the release of HTML 4.0 in 1997. The difference between Mosaic and Prodigy was that unlike Mosaic, Prodigy had an immediately viable business model. It didn’t take long for Internet Explorer to follow, eventually resulting in the current browser dominance by IE (and very nearly Netscape).
At a high level, the story above is what is going to happen with enterprise software. Software innovation doesn’t happen in a vacuum; it usually builds on top of existing technology (just look at how many new software companies build their products on top of AWS). So the fact that there is no dominant software platform to base such a technology on is not an inhibiting factor.
5. Hardware must be controlled –
Nothing controls anything the way software does, and this applies to hardware as well as to software. When Moore’s law slows down, this will present enormous problems for device manufacturers because there will no longer be a clear delineation between their “hardware” and someone else’s “software.”