Contributed by Rubin Worldwide
Written by Dr. Howard A. Rubin, Founder, Rubin Worldwide

Witnessing the Possible End of Moore’s Law and What to Do About It


On May 13, 2016 the MIT Technology Review published an article with an intriguing title, “Moore’s Law is Dead. Now What?” with the tag line, “Shrinking transistors have powered 50 years of advances in computing….” And in the article itself the introduction to the “Obituary” is clear “in a few years technology companies may have to work harder to bring us advanced new use cases for computers. The continual cramming of more silicon transistors onto chips, known as Moore’s Law, has been the feedstock of exuberant innovation in computing. Now it looks to be slowing to a halt.”

On February 15, 2018, Newsweek published a more disturbing piece, “The Future of Technology is Uncertain as Moore’s Law Comes to End.”

Both of these headlines, even if only partially true, portend a change in the dynamics of a core component of technology economics – “infrastructure economics.”

Moore’s Law essentially states that computing power doubles every two years while the cost of that computing power remains the same.  In short, computer power gets better, faster and cheaper.  This phenomenon has been discussed since 1965 when it was first theorized by Gordon Moore, one of Intel’s founders. As reflected in these articles, however, “the pace of advances in computing power is slowing, and it is taking longer than two years to achieve a doubling.”

Moore’s Law has had a profound impact on every organization’s infrastructure costs and almost none more so than in the banking and financial services sector in which infrastructure spending relative to revenue is typically 4% or higher according both the Rubin Worldwide research database and Gartner Inc’s most recent IT Key Metrics Database reports for Banking and Financial Services. .  To put this into perspective, consider the world’s biggest banks with revenue in the $70B to $100B range with infrastructure annual expense being generally over $3B a year.

To understand the economic impact of Moore’s Law and what it means for it to be dead, the context needs to be that of full infrastructure costs encompassing computing platforms, storage, end-user computing, network components, database technology, call center/service desks, and their management.  Such costs consist of personnel and non-personnel expenses, which run the gamut from salaries and benefits to hardware and software expense, hardware depreciation, software amortization, various overhead costs, and occupancy/real estate costs.

With Moore’s Law churning away in the background of this cost basis, its impact gets diluted because computer hardware and related depreciation represents less than 20% of infrastructure expense (with the balance being software, personnel costs, occupancy, etc.) but is still significant and is in the range of 2.5% to 5.0% of total infrastructure expense.  In some sense, it represents an annuity that firms can either take to the bottom line or reinvest.   However, this is almost a universal benefit to the technology community.  It is there for the taking, and even companies without any sort of infrastructure expense optimization program can simply “surf” it – essentially getting Moore’s Law benefits through their basic infrastructure refresh and acquisition cycles.

But based on the headlines, it appears it is coming to an end.  If this is true, it will become imperative to take action and put in place active programs for infrastructure cost optimization – surfing will no longer be an option.


Evidence of the End of Moore’s Law: Infrastructure Economics

The chart below shows unit costs of key infrastructure components as tracked from a database of more than 3,000 companies spanning 20 sectors during the period 2009-2017.

When viewed as a chained index using 2009 as the base year, a remarkable shift is evident between 2016 and 2017.  At the same time that the end of Moore’s Law was predicted, the data base of unit costs shows a major inflection point (2016-2017) at which unit cost went up!

The table below shows the trends from 2009 to 2016 and then the shift from 2016 to 2017.

There may likely be other cost drivers at work such as descaling of mainframe and UNIX environments, changes in labor rates, and even technology changes, but after 9 years of documented improvements of infrastructure economic efficiency reflected in unit cost something clearly has changed.

An alternative view – using a “market basket approach” – reveals the magnitude of the underlying economic shift.  Consider a mix of infrastructure services of the volumes shown below:

At the unit costs shown in the initial table, processing these volumes would have cost $1,415M in 2009 and dropped to $856M in 2016 and then rebounded to $910M in 2017.   Using an indexed perspective with 2009 as the base year at 100, the index value dropped to 61 in 2016 and then bounced up to 64 for 2017.

Perhaps what we have evidenced here is the first sighting of the actual end of Moore’s Law.  There are likely arguments other forces are at work and perhaps this is an aberration.

But no matter what the underlying dynamics are, something has changed.  And this change should be a cause for concern for those that have been surfing Moore’s Law and for those that have developed strategies to amplify its effects.


Implications and What to Do (Whether you surf or not)

There are a number of factors that influence infrastructure expense. Unit cost plays a key role but is only part of the total picture.  The other major driver is “demand/capacity” – the size of the infrastructure itself relative to the size of your business.  The basic equation is that too much “stuff” even at low unit costs likely equals too high an overall cost.  Therefore, whether surfing or not both demand management and unit cost management are critical competencies.

In the area of unit cost management, best in class performers are now focusing on

  • Labor productivity via automation, RPA (robotic process automation), and standardization
  • Labor economics via location strategy and sourcing models
  • Hardware and software costs via new vendor management strategies (such as Category Management and Category Captains)
  • Hardware and software accounting via reconsideration of depreciation and amortization models

In the area of demand management, the dominant strategy is to focus on driving out “valueless complexity.” Valueless complexity relates to multiple dimensions of IT “people, process, and technology” that can unfavorably impact the expense of the technology organization and result in under-performance of business value creation.  The dimensions encompass the application’s portfolio size/application complexity, architectural complexity, workforce complexity, infrastructure size and  complexity, sourcing/vendor management complexity, cyber/risk management complexity, organization structural complexity, and more.  Benchmarking is the key to get started in this area of infrastructure optimization.

In a typical organization, unit cost optimization efforts of various sorts have been underway for years while demand management has had a minimal amount of attention.  As a result, recent data indicates that perhaps the opportunity presented by demand management is 2-3x that of pure unit cost optimization.


And What Will Come Next

In this age of digitization and the race to the external cloud, infrastructure economics and its continual optimization will need to rely more heavily on externalities, the supply chain, and vendor management.  Concurrently, as the need for more elastic technology cost structures increases to cope with market volatility, cost transparency will be key and there will be an intense focus both on fixed versus variable costs along with cost response time (e.g., how fast the cost structure can change in reaction to changes in demand).

This all portends a future in which technology economics itself will become even more important as a competitive tool.  A future in which IT leaders will likely take on the role of an enterprise’s “Chief Technology Economist”.   The ability to effectively manage technology “monetary policy” will be a differentiator and with demand for infrastructure services and capabilities at the core of enabling IT strategy – digital or otherwise – mastery of infrastructure economics will be key.



Dr. Howard Rubin


Gartner Senior Advisor