The new ‘Law’ in the industry to go along with Moore’s law has been dubbed Bezo’s Law which measures the cost of cloud computing over time. This law says that over the history of cloud that the cost of a unit of computing power has reduced by 50% every three years.
Bezo’s Law is based upon the Total Cost of Infrastructure Ownership (TCIO) which is a calculation of the cost of owning and operating a data center. There are a lot of cost components to TCIO including hardware, software, power, labor and overheads. Most companies that operate data centers don’t publish their costs in enough detail for an outsider to calculate such costs accurately, so there is no easy way to measure the current cost metric precisely
But it is obvious that the overall costs of operating data centers continues to drop. For example:
- There has been a lot of emphasis on lowering the power consumption of data center equipment and power is a significant component of data center operating costs.
- The cost or routers and switches has continued to drop, and companies like Amazon and Google have developed their own hardware which has supposedly cut their costs more than companies buying commercial equipment
- Software costs are getting cheaper through the advancement of software defined network techniques. And this same software should lower labor costs over time.
Bezo’s law is lately in the news because we are now reaching the point where it is probably cheaper to buy cloud services from one of the large providers like Amazon than it is to build your own data center. Amazon has essentially achieved an economy of scale that a smaller provider can’t match by building from scratch. This is not a unique economic phenomenon. Economic theory predicts that industries that benefit from economy of scale will eventually be dominated by the most efficient firms.
There are a lot of major implications for the telecom industry if it’s now cheaper to buy cloud services than it is to operate your own data center. I would expect to see all of the following within a few years:
- One can expect the smaller and less efficient data center providers to fade away over the next few years as their margins get squeezed and finally killed by more efficient operators. One would expect to only find a small number of data center providers within a decade.
- It no longer makes sense for corporations and governments to have their own data centers. Eventually cost savings will become compelling enough that you can expect a faster and faster migration of corporate IT functions to the cloud. This also implies dire consequences for employment for IT people who have specialized in providing services that can be replaced in the cloud.
- Makers of routers and switches are at risk because as the number of companies left in the data center business decrease, the only vendors that survive will be those that sell to those handful of companies. This also implies a squeeze on the margins of IT equipment since the large cloud companies will have the bargaining power to insist on low prices.
- ISPs will have a hard time justifying operating their own routers and servers and one can expect ISP functions like DNS, email and storage to move to the big cloud providers.
- One would expect a proliferation of specialized cloud software companies. The big cloud providers are likely to offer generic software. While the hardware will all be owned by a few large companies like Amazon and Google, the specialized needs of different industries is going to be met by specialized software that works with the cloud.
- This will accelerate the shift to software defined networking and one can expect to see things like cable TV headends, cellular base stations, voice switches and other hardware platforms all migrate to the cloud as well. Makers of those kinds of hardware also face a bleak future.
- This will put even more pressure to have very fast Internet connections. We will soon not have just a digital divide, but a cloud divide where there are those who can benefit by the cheap cloud and those that can’t.
This shift is gigantic for the telecom industry and we have reached a significant tipping point. In the future when we show a timeline of the history of the telecom industry (like I did yesterday), we will see and entry saying ‘2014 – When the Cloud Won’. One can debate if we have already reached the tipping point or if it’s next year. But what can’t be debated is that cheap cloud resources are going to change our industry in major ways. A lot of vendors we are used to working with will disappear and companies like Amazon and Google will become entrenched in telecom (and in myriad other industries).