Focus On: Silverlake Ten Years Later

Starting in 1986, D.H. Andrews Group (Cheshire, Conn.) became notorious for its predictions about IBM’s plans for "Silverlake," now known as the AS/400. The key message behind them was that something important was happening that would change the world of computing forever.

While the evolution of the AS/400 did not happened precisely as predicted, the D.H. Andrews Group (then known as ADM Consulting) predictions about the importance and impact of AS/400 were much more accurate than those made by any of the traditional industry analysis firms.

As the AS/400 enters its second decade, it is natural to wonder for how long it will remain an important part of the computing landscape. It is therefore a good time to take a fresh look at one of the most successful and unusual products that the computer industry has ever seen. It is also appropriate, and should be fun, to engage once again in some shameless gossip and speculation.

The AS/400 has been a money machine for IBM. Over its 10-year life, the AS/400 has directly accounted for over $35 billion in revenue and more than $8 billion in profit. Sales of related hardware, software and services would more than double those numbers. To put it in perspective, the AS/400 made more money for IBM than Disney made on its amusement parks in the same period.

Although owning an AS/400 cannot match a day at the Magic Kingdom for fun, it is almost as effective at creating happy customers. The powerful combination of high revenues, profits and customer loyalty would seem to make the AS/400 into one of the most valuable business franchises in the world. It is, therefore, curious that senior management does not always behave as if the AS/400 were all that important to IBM.

A new management team has been put in place an average of every 18 months during the AS/400’s life span. These frequent changes were not caused by dissatisfaction with results. In fact, each past AS/400 general manager has moved on to a bigger job within IBM.

In late 1997 the AS/400’s sixth management team arrived full of ambition and energy. The AS/400 Division’s budget for development and marketing was cut before they arrived leading to staff downsizing and the loss of a great deal of experience. The net result is a talented but inexperienced team with lots of new ideas. Fortunately, the new team inherited a product line that was the strongest it has ever been.

The AS/400 began life in 1988 as the world’s finest proprietary minicomputer. It soon became the dominant force in its niche and helped hasten the demise of the DEC VAX and other traditional minicomputers. A number of major shifts in the market, however, have made life uncomfortable for AS/400 developers during the 1990s.

Those challenges included the open systems and Unix movement, the rise of client/server, Microsoft’s entry into the server OS market with NT, and the sudden emergence of the Internet and Web browsing as a new computing style. The AS/400 reacted to each of these developments, but there has always been a lag between when the market developed a taste for something new and when the AS/400 found a way to adapt to it.

The underlying design of the AS/400 has always been its greatest strength. It gave the AS/400 the potential to adapt to radical change and emerge not just as adequate but in some cases as a superior alternative. Unfortunately, the full potential of the architecture could only be realized with a major redesign. The AS/400’s third General Manager, John M. Thompson (now IBM’s top software executive), saw what was needed and approved a massive project to re-invent the AS/400 in 1993.

The re-invention took five years and over $2 billion in development. Each year during that period saw improvements, but the early years were lean in terms of customer value. As a result, AS/400 sales stagnated and profits slowly declined. The attention paid to the AS/400 outside of its installed base also steadily diminished during this time.

It has been five years since Lou Gerstner became IBM’s CEO. During the Gerstner era, the AS/400 has been a major, but declining, source of profits. It has also been a product whose industry profile has steadily dropped. As a result, there has been limited motivation for Gerstner’s management team to invest the time needed to fully understand IBM’s most unusual product.

Gerstner’s grand plan for IBM has centered around growing the service business. This is understandable since he began his career as a consultant with McKinsey. There has also been a heavy focus on electronic commerce and the Internet. Server hardware and software contribute almost half of IBM’s profits, but are not seen as a market with high growth potential. Gerstner’s team thus tends to play down the importance of servers and takes the attitude that it does not matter which type of server a customer buys as long as it is one that IBM makes. This philosophy makes it politically difficult for AS/400 management to shout to the world that their product is better than other types of servers.

The relative importance of servers has declined during the 1990s for a number of reasons. The most important is that openness standards have made it possible to write applications that can be ported from one type of server to another. The cost of the four main types of servers (mainframes, Unix, Intel/PC, and AS/400) has also come closer together so that hardware cost is rarely the reason to choose one over the others.

A new and different AS/400

While Gerstner was shaking up IBM and wave after wave of change was hitting the information industry, the developers in Rochester kept plugging away. There wasn’t any one instant when everything was done, but a point was reached early in 1998 when it was accurate to say that the re-invention of the AS/400 was largely complete. All things considered, the results were impressive.

As luck would have it, the most important AS/400 improvements arrived just as IBM was re-organizing its entire server business. The result was that little was done at that time to point out what had been accomplished. The February 1998 AS/400 enhancements were announced before the new management team had been around long enough to hang the pictures in their offices much less comprehend how a highly unusual product had undergone such radical changes.

With computers there is always another wave of improvements just around the corner. In the case of the AS/400, the next wave came in September 1998. The AS/400 Division therefore has taken this opportunity to draw attention to the new value proposition their product line now offers. Tom Jarosh, the sixth AS/400 general manager, has convinced Gerstner to spend much more than they have in the past on AS/400 promotion with the goal of improving market awareness and image.

The story Jarosh has to tell is a good one. The AS/400 has always had a number of core strengths: high reliability, great security, lots of business applications, ease of operation and highly integrated software that limits the need for technical support. In addition to retaining these strengths, the AS/400 has overcome its most important weaknesses including:

  • Performance. The AS/400 has offered 64-bit RISC processors for four years. It took a second generation of processors and microcode for them to achieve the high performance that 64-bit processors are capable of.
  • Top-end capacity. For years the average increase in the capacity of the largest AS/400 was 70 percent. In 1997 the top end increased by 470 percent. Further large increases are planned so that the rate of growth is no longer an issue.
  • Openness. The long and painful process of incorporating openness standards is finally largely complete. The last major pieces were the Java language and a thread capability compatible with Unix and industry standards. Together they mean that an increasing number of applications written to industry standards will be able to run on the AS/400 with little need for customization.
  • Hardware cost. A modest price premium remains for some of the large AS/400 models but the medium sized Model 170, the new custom servers and the other e-Series server models all now can compete with Unix and Intel alternatives on price alone.
  • Groupware. The addition of Lotus Domino transformed the AS/400 from a weak and costly system for serving office functions including e-mail and Web serving, into a cost, capability and performance leader.

The AS/400’s capability is therefore now better than its reputation. There is a great story to tell, but it is not an easy one to explain. Jarosh, however, has a secret weapon in the form of Malcolm Haynes. Many AS/400 followers have seen the three highly entertaining multimedia CD-ROMs, including "FATSO vs. AS/400," explaining various facets of the AS/400 that Malcolm has created. Jarosh has given Malcolm, who calls himself the AS/400 Minister of Propaganda and Enlightenment, a chance to let his fertile imagination run free. It will be fun to see what results.

What Will Happen Next?

The real question is what will happen next to the AS/400. It is easy to paint a dim picture of the future since there has always been an ample supply of AS/400 critics. The case offered by those currently throwing rocks goes something like this:

  • NT is an unstoppable force that Microsoft will use to take over the traditional AS/400 market.
  • The AS/400 image is beyond repair. Buyers will continue to see it as a proprietary minicomputer that was nice in its day but is overpriced, slow and whose applications are out of date.
  • IBM management no longer cares much about servers and is too absorbed with services and software to do much about the AS/400.
  • The base of loyal AS/400 advocates is shrinking. RPG programmers are spending their evenings learning Visual Basic or other languages.

There is some validity to each of these points. Overall, however, there is an even stronger set of reasons to be optimistic about the AS/400’s future:

  • Both NT and the AS/400 can succeed, especially now that the AS/400 can run NT on co-processors under the covers. Early euphoria over NT is rapidly fading as its limitations become apparent and delays in improvements continue.
  • The AS/400 image could turn around when buyers grasp the value it now offers versus Intel or Unix servers.
  • As results improve, IBM management will pay more attention.
  • Java is the language of the future. The AS/400 could easily become the best server for Java applications. The native 64 bit processors, the affinity between the structure of Java and the AS/400’s architecture and the innovative way Java has been built into it are the reasons this line of thinking may prove valid.

1998 therefore will likely be a turning point year for the AS/400. If the Jarosh media campaign captures the imagination of buyers and sales continue to increase, then a positive growth cycle could begin. The improving image will lead to more sales which will free resources for additional advertising and even greater awareness.

D.H. Andrews Group current predictions for what will happen to the AS/400 therefore include:

  • 1998 will see year to year revenue growth of over 15 percent, the highest in many years.
  • Market image and buyer awareness will gradually improve.
  • There will be a large surge in orders starting soon. Owners of the large number of outdated CISC systems will determine that it is more cost effective to buy new RISC models than to upgrade. The Year 2000 problem will force these customers to do something soon.
  • Lotus Domino users, who now tend to only buy Intel or Unix servers, will discover the AS/400 in large numbers. If the AS/400 can establish a major presence in this market unit sales could soar.
  • Java will become the leading language for application development, but not until after the Year 2000 comes and goes. This could create a great deal of demand for AS/400 servers.

Jarosh’s team has a wonderful opportunity before them. They must fix the image and awareness problem, get IBM management behind then, increase the number of people selling and supporting AS/400s for customers, and move fast before Microsoft eliminates the near term NT weaknesses. IBM must also succeed in its extensive campaign to help Java mature and gain acceptance. Most importantly, IBM needs to find a way to grow the base of AS/400 talent.

Netting out all the above, the worst case scenario for the AS/400 appears to be that it will grow slowly but remain largely focused on the needs of current customers. In the best case, the AS/400 will realize its full potential and rival Intel and Unix servers in total share of the server market. Most likely, the next five years will see growth in AS/400 sales at a much higher rate than in recent years and many new converts.