• Selling our technology would recoup some of our substantial R&D
expenditures and open up a new income stream.
148 / LOUIS V. GERSTNER, JR.
• In a post-PC world, there would be high demand for components to power all the new digital devices that would be created for network access.
IBM Research
As I said earlier, it has been well known for half a century that IBM has one of the most prolific and important scientific research laboratories in the world. IBM has more Nobel laureates than most countries do, has won every major scientific prize in the world, and has consistently been the foundry from which much of the information technology industry has emerged.
However, the Research Division of the early 1990s was a troubled place. My colleagues there saw the company being broken up into pieces and wondered where a centrally funded research organization would fit in an IBM that was being disaggregated. When they heard I had decided to keep the company together, the collective sigh of relief that emanated from Yorktown Heights, New York (the headquarters of our Research organization) was almost audible.
One of the obvious but puzzling causes of IBM’s decline was an inability to bring its scientific discoveries into the marketplace effectively. The relational database, network hardware, network software, UNIX processors, and more—all were invented in IBM’s laboratories, but they were exploited far more successfully by companies like Oracle, Sun, Seagate, EMC, and Cisco.
During my first year at IBM I probed frequently and deeply into the question of why this transfer of technology invention into market
WHO SAYS ELEPHANTS CAN’T DANCE? / 149
place performance had failed so badly. Was it a lack of interest on the part of IBM researchers to deal with customers and commercial products? It did not take long to realize that the answer was no.
The major breakdown was on the product side, where IBM was consistently reluctant to take new discoveries and new technologies and commercialize them. Why? Because during the 1970s and 1980s that meant cannibalizing existing IBM products, especially the mainframe, or working with other industry suppliers to commercialize new technology.
For example, UNIX was the underpinning of most of the relational database applications in the 1980s. IBM developed relational databases, but ours were not made available to the fastest-growing segment of the market. They remained proprietary to IBM systems.
Getting Started
The easiest step we could take initially was to license technology to third parties. This process did not involve selling actual components or pieces of hardware and software, but it did allow other companies access to our patent portfolio or our process technology.
(“Process technology” is, as the name suggests, the technology required for the IT industry’s own manufacturing processes—the skills and know-how to build leading-edge semiconductor and storage components.) This effort—licensing, patent royalties, and the sale of intellectual property—was a huge success for us. Income from this source grew from approximately $500 million in 1994 to $1.5
billion in 2001. If our technology team had been a business unto itself, this level of income would have represented one of the largest and most profitable companies in the industry!
However, this was just the first step to opening the company store.
We moved on from simple licensing to actually selling technol
150 / LOUIS V. GERSTNER, JR.
ogy components to other companies. Initially, we sold fairly standard products that were broadly available in the marketplace but that, nevertheless, IBM chose to make for itself. Here we were competing with many other technology suppliers, such as Motorola, Toshiba, and Korean semiconductor manufacturers. The principal product we offered in the market was simple memory chips called DRAMs (pronounced D-RAMS).
Selling commodity-like technology components is a feast-or-famine business driven not so much by customer demand as by capacity decisions made by the suppliers. Our DRAM business made a gross profit of $300 million in 1995 (the feast), then proceeded to lose $600 million three years later (the famine).
We were not naïve about the nature of this business; its cyclicality was well documented. It turned out, however, that the downturn of 1998 proved to be the worst in the history of the industry.
Why were we in the DRAM business? Well, we really didn’t have any choice. We had to prove to the world that we were serious about selling technology components. Most of the potential customers for our technology were worried (quite appropriately) that they might begin to depend on us and that we would subsequently decide to turn inward again.
Consequently, riding the DRAM wave was the price of admission for us to enter the technology component marketplace. We had essentially exited the DRAM market by 1999, but by then DRAMs had given us an entry point. Now potential customers worried less about our reliability as a supplier or whether we had a serious commitment to this business.
We were ready to tackle the emerging opportunity in the components business: The change in computing that we’ve been talking about was driving a fundamental shift in the strategic high ground in the chip industry.
As I’ve discussed, the action was going to be driven by the proliferation of Internet access devices, exploding data and transaction
WHO SAYS ELEPHANTS CAN’T DANCE? / 151
volumes, and the continued build-out of communications infrastructure. All that was driving demand for chips—and, to our great delight, chips that would have fundamentally different characteristics from the lookalike processors that powered lookalike personal computers.
In this new model, value would shift to chips that powered the big, behind-the-scenes processors. At the other end of the spectrum there would be demand for specially designed chips that would go inside millions, if not billions, of access devices and digital appliances. And in between would be chips in the networking and communications gear.
This is the kind of sophisticated development activity that not only allows great technology companies to shine, but it also generates margins that support the underlying investment necessary to lead.
Over the next four years, IBM’s Technology Group went from nowhere to number one in custom-designed microelectronics. I’m happy to say that PowerPC has experienced its own renaissance here, quietly reemerging as a simpler, cheaper, more efficient processor used in a wide range of custom applications, including game consoles. Just consider that IBM’s contracts with Sony and Nintendo in 2001 hold the potential to produce more intelligent devices than the entire PC industry produced in 2000.
As a result—and here’s the important point—IBM, for the first time in its history, is now positioned to benefit from the growth of businesses outside the computer industry. This diversification does not take us away from our core skills; we have simply extended them to entirely new markets.
Our Technology Group is still young and developing. We can’t yet declare victory in this, the third of our growth strategies. It is a piece of unfinished business that I leave for my successor.
Nevertheless, while the economic benefits of our Technology Group strategy have been uneven, the underpinnings of the strategy are powerful and potentially huge for IBM. First, if one believes in the
152 / LOUIS V. GERSTNER, JR.
theory of building great institutions around core competencies and unique strengths, then exploiting IBM’s technological treasure trove is an extraordinary opportunity for the company.
Second, the evidence to date is fairly clear: The two companies that have enjoyed the highest market valuation in the IT industry over much of the last decade have been component manufacturers—Intel and Microsoft. Certainly one derives enormous benefit from its virtually monopolistic position. But there is no doubt that a strategy built around providing fundamental building blocks of the computing infrastructure has proven to be extremely successful in this industry.
17
Unstacking the Stack and
&nb
sp; Focusing the Portfolio
B efore you proceed with this chapter, let me make one disclaimer about the upcoming diagram. I am not going to launch into a primer on computing topologies. Instead, I want to use this oversimplified picture of the industry’s structure to illustrate the flip side of our work to restore IBM’s economic viability. This is about the bone-jarringly difficult task of forcing the organization to limit its ambition and focus on markets that made strategic and economic sense.
The diagram is commonly called “the stack,” and people in the computer industry love to talk about it. The stack shows most of the major pieces in a typical computing environment. At the base are components that are assembled into finished hardware products; operating systems, middleware, and software applications sit above the hardware; and it’s all topped off by a whole range of services.
Of course, it isn’t anywhere near as simple as this in real life.
By now you know that IBM’s strategy when it birthed the Sys
154 / LOUIS V. GERSTNER, JR.
tem/360 was to design and make every layer of this stack. But thirty years later the industry model had changed in two fundamental ways. First, very small companies were providing pieces inside the stack that IBM had invented and owned for so long, and customers were purchasing and integrating these pieces themselves.
Second, and just as threatening, two more stacks emerged. One was based on the open UNIX operating platform. The second was based on the closed Intel/Microsoft platform. In the mid 1980s, when IBM had more than a 30 percent share of the industry, the company could safely ignore these offshoots. But by the early 1990s, when IBM’s share position was below 20 percent and still falling fast, it was long past time for a different strategy.
We had to accept the fact that we simply could not be everything to everybody. Other companies would make a very good living inside the IBM stack. More important, just to stay competitive we were going to have to mount massive technical development efforts. We couldn’t afford not to participate in the other stacks, where billions of dollars of opportunity was being created.
I’ve already described the results of our decision to expand into the UNIX and Wintel markets—reinventing our own hardware platforms at the same time that we built new businesses in software, services, and component technologies. Just as important, we had to get serious about where we were going to stake our long-term claims inside the IBM stack.
The first and bloodiest decision was the determination that we would walk away from the OS/2 v. Windows slugfest and build our software business around middleware. Before the 1990s came to a close, we made another strategic withdrawal from a software market.
WHO SAYS ELEPHANTS CAN’T DANCE? / 155
The Stack
Leaving Application Software
For most of its modern history, IBM made and sold hundreds of business applications, for customers in industries like manufacturing, financial services, distribution, travel, insurance, and health care.
These were important applications for important customers, yet we were accomplishing little more than losing our shirts. Jerry York conducted an audit that showed over the previous twenty years IBM
had
156 / LOUIS V. GERSTNER, JR.
invested about $20 billion in application development and acquisition, with a negative rate of return of around 70 percent!
This was—and is—a very specialized segment of the industry: everything from small-business payroll packages to software for automotive design or even the sophisticated software used to simu-late biological and genetic activities. It has always been dominated by entrepreneurial companies that bring obsessive focus to their specialties—such as sales force automation or financial services. Interestingly, nobody has ever succeeded in building a broad portfolio.
When I questioned why we stayed in this business, I was told that application software was critical to the total solution (which was true enough) and that our problems were of execution, and therefore fixable. So we changed executives, tinkered with the strategy, and studied whether we should just buy a few of the best firms in the field. The first candidate was going to be SAP.
Three years, a lot of activity, and a few billion dollars later, we still weren’t solution leaders, and we weren’t getting anything close to a decent return on our huge investments.
However, one thing we were doing exceptionally well was irritating the heck out of the leading application providers—companies like SAP, PeopleSoft, and JD Edwards. These companies were in a great position to generate a lot of business for us if they were inclined to have their applications running on our hardware and supported by our services. Why? Because customers often bought the application first, then looked to that software provider to tell them which hardware to run it on. As long as these companies saw us as a rival, we were driving them into the arms of competitors like Sun or HP.
One example: The segment of IBM that produced applications for distribution and manufacturing customers set a stretch goal to increase sales by $50 million (from a base of about $100 million). It ran ads and promotions and sales contests, and it hit its target. In the process, it alienated every software company in that segment of the market. Those companies, in turn, stopped recommending our hard
WHO SAYS ELEPHANTS CAN’T DANCE? / 157
ware and contributed directly to a $1 billion decline in sales of one of our most popular products.
By 1999 we were finally ready to admit to ourselves that we could never be as single-minded as application providers that were in business to do just one thing—and do it better than anyone else. We exited application development but saved the very few pieces of software that IBM had successfully developed and marketed in the past. Thousands of software engineers were reassigned to other work, laboratories were closed, and investments were written off or sold.
Important as it was to stop deluding ourselves about our profi-ciency in this part of the stack, just as important was the message that we were prepared to work with the leading application software developers. What we said to them was: “We are going to leave this market to you; we are going to be your partner rather than your competitor; we will work with you to make sure your applications run superbly on our hardware, and we will support your applications with our services group.”
And rather than just having lunch with them and saying “Let’s be partners,” we structured detailed commitments, revenue and share targets, and measurements by which both parties agreed to abide.
The first company we approached was Siebel Systems, which had a leadership customer relationship management software package.
Its CEO, Tom Siebel, was understandably enthusiastic about the prospect of having IBM’s worldwide sales force and services organization marketing and supporting his product. But based on what he’d observed of IBM’s agility (or lack thereof), Siebel told us he doubted we could structure a deal on his timetable. He bet the IBM
team a bottle of fine wine that the whole process would break down due to what he called “cultural impedance mismatch” between Siebel and IBM.
Five days later Tom was picking out a fine Chardonnay. The contract was signed and we announced the relationship and the new alliance program. Over the next two years we signed 180 similar partnerships.
158 / LOUIS V. GERSTNER, JR.
In hindsight this looks like a no-brainer, given that it dramatically improved the economics of our business and was entirely consistent with our overarching strategy of being the premier integrator. Software companies that in the early 1990s viewed IBM as a major competitor are now very important partners. The amount of incremental revenue we realized is in the billions, and we achieved significant market share gains in 2000, then again in 2001.
The IBM Network
Some may think that the task of moving data from centralized computers to distributed computers, or from one manufacturing site to another, or from one country to another, would be the natural domain of telecommunications c
ompanies that had been providing voice transmission for nearly a century. However, until very recently telephone companies had minimal skills in data transmission, and voice services were based on a totally different technology. Moreover, the industry was nationalistic, monopolistic, and highly regulated.
Global telecommunications companies did not emerge until the mid-1990s.
So in the spirit of “If they need it, we will build it,” IBM in the 1970s and 1980s created multiple data networks to allow its customers to transfer data around the globe. We filled an important void.
By the early 1990s, however, the telecommunications companies were shifting their focus dramatically. Driven in part by deregulation, as well as by the revenue potential of digital services, all of the world’s major telecommunications companies were seeking ways to create a global presence, as well as digital capability. In the parlance of both the IT and telecommunications industries, they were talking about moving up the value chain. United States companies that had provided telephone service to customers only in a certain geographic sector of the country were suddenly investing in Latin
WHO SAYS ELEPHANTS CAN’T DANCE? / 159
American telephone companies. European telephone companies were joining consortia and building wireless networks in remote parts of the world.
In a period of about twenty-four months, the CEOs of nearly every major telecommunications company in the world traveled to Armonk to talk with me about how their companies and IBM could team up to create digital services. The proposals presented to us ran the full spectrum—from modest joint activities to full-blown mergers.
However, affiliating IBM in one way or another with a telephone company made no sense to me. I saw little to be gained from a partnership with a regulated company in a different industry. Besides that, we had enough problems in IBM’s base business. I wasn’t inclined to take on additional challenges.
What did occur to me was that we had an asset that most of these companies would be seeking to build over the next five years. And if the world was moving in the direction we anticipated—toward a glut of many networks (the Internet wasn’t even an important consideration at the time)—then the value of our network would never be higher. So we chose to auction it off to the highest bidder. We thought we’d be doing well to get $3.5 billion. But the frenzy eventually produced a bid of $5 billion from AT&T; that was an extraordinary price for a business that produced a relatively tiny percentage of IBM’s profits.
Who Says Elephants Can't Dance?: Leading a Great Enterprise through Dramatic Change Page 15