Reinvented as Intel's dealmaker in California — working next to Andy Grove and Gordon Moore — Avram champions networking and multimedia long before they're fashionable, and learns to fight for ideas the company isn't yet ready for. A chapter about influence without authority, and how to make things happen from the inside.
During my period working on Gemini/BiiN (1984–1988), I had let Andy and Les know that I was not happy with my role at the company and even started to explore leaving Intel. Andy said two things to me that, as it turned out, were very important. First, he said that I should stay at Intel because the company accepted me for who I was. While it was a strange comment, Andy was right. Even though I was very different from the other Intel executives, I was always treated with respect.
That was, I suspect, because we all shared many of the same values: a focus on integrity, commitment, honesty, and results. Second, he said that if I ever needed something personally, I should just ask.
California, Here I Come
As it happened, a few years later, my wife, Arianne, was accepted into Stanford’s PhD art history program. We needed to relocate to California. Intel took care of all the expenses, even though the move was at my request. Arianne and I found a beautiful home in Palo Alto, which, by then, we could happily afford. My eldest son, Adin, went off to Brandeis University, so there were now only four of us. My office was located close to Gordon Moore and Andy Grove, fortunately, because this gave me a great deal of access.
Back in 1986, Les had taken over the Systems Group from Bill Lattin when Bill retired, and Les spent much of his time in Oregon where that group was headquartered. My new job was doing business development for that group, still reporting to Les. I was responsible for activities like acquisitions and minority investments in support of the various business units comprising the group. In some cases, this structure worked well; in other cases, not so well. The business units were very focused on the short term. They rarely came to me with deals, so I had to find opportunities that fit in with whatever they were doing. I, then, had to convince them to sponsor the deal. They tended to have little interest unless there was a short-term payoff. I was more interested in having a long-term impact, so this was disappointing. One group where I was successful in assisting was PCEO (the PC Enhancement Operation), an internal start-up that Les had originally sponsored.
My timing was perfect, it turns out. Intel was about to undergo a profound transformation, from being a manufacturer of semiconductor components to one of the leaders in the computer industry. The 1987 annual report title was “Intel the Architect of the Microprocessor Revolution.” In the process, Intel would become one of the most valuable and influential companies in the world, and I would finally be able to play a meaningful role in making that happen.
Strategic Long-Range Planning
In 1987, the year before I moved to California and just twenty years after Intel’s founding, Andy Grove was named CEO of Intel. He replaced Gordon Moore, who became the chairman of the board. Andy held the CEO position for eleven years and then became the board chairman until 2005. Grove, a Jew who had survived both the Holocaust and the Russian occupation of Hungary, and who’d come to the USA at the age of twenty-one not speaking English, would be hailed as one of the greatest CEOs of all time, if not the greatest. Much has been written about this remarkable man, most of which I will not repeat here.
In the ‘70s, Les Vadasz was tasked with creating Intel’s strategic planning process. One element of the process was called SLRP (strategic long-range plan). SLRP was initially bottom-up. As the company grew, SLRP became unwieldy, with more and more business units each presenting their own vision and plans. Intel had also evolved from being a semiconductor manufacturer of memory components into the world leader in microprocessors and, as Andy would describe it, “the foundation of the New Computer Industry.” In 1988, Andy changed SLRP to be top down. The result was that Andy now became Intel’s strategist. A consequence of this was that Intel’s focus became narrower and deeper.
The SLRP meeting typically lasted about three days and would take place in April. It had about fifty attendees, including Intel’s senior leadership and individuals of their choosing. Attendees were broken down into subgroups to work on various topics Andy presented. Subgroups would then review the results of their deliberations with the larger group, followed by a discussion with the full group.
It was meant to be a three-year look into the future, with the resulting plan that covered the same time frame. I used to joke that the SLRP discussions would end up dealing with the last year, the current year, and the next year. Even if it is a three-year view of the future, three years is not long enough to identify the potential discontinuity that could either provide enormous opportunity or possibly disrupt the current business. I am sure that those who were responsible for the development of Intel’s semiconductor technology had a much longer-term perspective, but this was not a subject for SLRP.
Andy began with a two-hour presentation reviewing past SLRP discussions, then outlining his views of the future and raising various issues and opportunities as he saw them. How I wish I had copies of those presentations! They reflected his superb analysis of the industry.
Andy, like the rest of us, had come to slowly understand that the personal computer launched by IBM in 1981 had effectively turned the computer industry on its side. This change presented both Microsoft and Intel with the opportunity to dominate the computer industry.
Something Andy clearly recognized.
However, over the years, Andy got stuck, and Intel got stuck along with him. Though Intel was enormously successful during the 1990s, it did not use either this time or its wealth to adequately prepare for the future that would result from the Internet and, later, from mobile computing. Intel even had difficulties initially grasping the importance of notebook computers and the growing home computer market, in my opinion. The company became fixated on the performance of its microprocessor. Moore’s law would be the dominant force driving the company. Intel interpreted that as meaning ever-faster microprocessors. Increased performance required more power, which resulted in shorter battery life and made Intel less competitive in the portable market. Even though Intel no longer had to enable second-source suppliers of its products, companies like AMD, and later Cyrix, and many others, offered compatible microprocessors. Intel therefore still had competition, especially on the lower end of processor performance.
Intel’s superior semiconductor technology meant it could build higherperforming processors than other companies. Processor performance was a key aspect of Intel’s competitive strategy, but performance only matters to customers as long as there are desirable applications that can take advantage of it.
I find it ironic that Andy, who so clearly documented the importance of “strategic inflection points” for subsequent generations of business leaders, would end up missing so many. Andy used to say that two things drove Silicon Valley: paranoia and greed. Over time, Andy moved from paranoia to greed, from risk-taker to risk avoider.
Eventually, Andy recognized that the Internet was such an inflection point. I think that was one of the reasons he turned over the reins of the company to Craig Barrett. While he recognized it, he did not know what to do about it. Barrett had been very vocal about Intel’s inability to create new business. Living in Arizona, he likened the PC microprocessor business to a creosote bush that would prevent other plants from growing. But while Barrett understood the problem, he was not the right person to come up with the solution. Nor were any of his successors.
Vein Of Gold
When the IBM personal computer was launched in 1981, very few people foresaw that the control of the computer industry would move from system manufacturers to two key component companies, Intel and Microsoft, resulting in great transfers of power and wealth. IBM certainly did not understand what it meant to be dealing with the likes of Andy Grove and Bill Gates.
Back then, Intel believed that the profit in the computer industry would continue to be made by computer system companies like IBM and Digital Equipment. Microsoft thought that Unix could be the dominant operating system and was working on a version of it, Xenix.
But as the PC market grew, Grove and Gates decisively abandoned their existing plans and fully embraced the PC opportunity. Over time, both companies came to realize that their customers were not the PC manufacturers such as IBM, Compaq, Dell, and Gateway but rather the individuals and enterprises that purchased those companies’ products. Intel’s marketing moved from focusing on the engineers that would design personal computers to the individuals and enterprises that would use them. This resulted in marketing programs such as Intel Inside, which created a very strong ingredient brand. Microsoft also developed a robust branding effort.
Fueled by growing profits, Intel was able to invest in developing much of the critical technologies required to move the center of the computer industry to the PC. Intel and Microsoft were then able to extract most of the profit in the computer industry.
Exploiting the PC opportunity was brilliantly executed, and Andy was able to get the company totally focused. He had found a vein of gold and was committed to extracting every ounce. But he did not realize that after you extract all the gold, you find yourself in a very big hole. Dominating the microprocessor used in the personal computer became known as Job 1. All other businesses were called Job 2, but after a while, Job 2 came to mean doing things that would support Job 1.
While Andy wanted to see Intel develop new businesses, he offered his support for that objective inconsistently.
Intel’s performance under Andy’s tenure was astounding. This kind of success blinded Andy and the Intel board. With the exception of the period representing the collapse of the Internet bubble, sales and profitability continued to grow significantly for many years.
A N D Y G R O V E : C E O
Sales Profit Capitalization
1987 $ 1.9B $ 0.24B $ 4.3B
1998 $ 26.3B $ 6.10B $ 197.0B
Growth 13x 24x 45x
I N T E L P E R F O R M A N C E
Andy Grove’s amazing performance as CEO I have grown to believe that what makes the long-term difference is not only what you do when things go badly but also what you do when things go well. It is a great challenge to develop new businesses in the shadow of any business as successful as Intel’s microprocessor business. Intel, like so many other companies, was failing success.
Andy was an amazing manager, and he taught others in the company how to manage effectively. Much has been written about Andy’s management capabilities and his management style, from Stanford Graduate School of Business Professor Robert Burgelman’s Strategy Is Destiny (2002) to Measure What Matters (2018) by famed venture capitalist John Doerr, which details one of Andy’s most effective management tools, objectives, and key results. Without a doubt, Andy was one of the best managers of any major technology company anywhere in the world, perhaps even the very best.
Andy deserves recognition for much of Intel’s success, but along with that must come responsibility for its failures. While he was razor focused on exploiting the opportunity the PC presented, he did not lay down the foundation for continued success, including picking the right successor. Most of his fans and many of my former colleagues and friends at Intel will disagree with me. My example of a great CEO would be Steve Jobs, even though I think Andy was unquestionably a much better human being.
What Andy lacked was vision. For me, a great CEO has to have a combination of vision, leadership, and management ability. Yes, Andy would come to recognize the importance of the Internet and the changes that would result in the nature of the computer industry, but it was too late. He failed to develop an organization that was able to embrace the future; instead it clung to the past. Nowhere would that become more evident than when Paul Otellini, then CEO, would turn down Apple’s request that Intel manufacture the processor for the iPhone. Of course, Paul could not have foreseen the impact of that decision at the time. But he should have realized that things would not stay the same and that there would be disruptions. Apple was already very successful with the iPod.
While this book is more critical of Andy than many other books written about Intel, I should be clear that while he and I had a difficult and complicated relationship, I had the utmost respect for him professionally and personally. I think Andy, if he were still alive, would have agreed with much of my criticism. A year before he died, suffering terribly from Parkinson’s disease, he apologized to me for not listening more to what I said. This meant a great deal to me. But I have to take responsibility for not getting him to hear me. In writing this book, I came to realize part of the problem. Andy was comfortable with the concrete. He wanted facts. But in thinking about the future, one has to use intuition and imagination. Andy wanted to know what I knew, not what I believed. But how could I know the future? I also believed you could influence the future. While you could not control it, you could bend it. One of our biggest disagreements came about because of our divergent views on the role of telecom companies in creating residential broadband.
— the future is too important to be left to chance
Expanding Intel’S Business
Initially, in my new role, I had difficulties being effective. While I shared Intel’s values, I struggled culturally; how I looked, dressed, and acted were just so different from the other Intel people that I was often not taken seriously. This reminded me of my difficulties as a young child when I returned to school after my year at the convalescent home at the age of eight.
At one point, I asked the human resources department if they could find someone who could help me adjust my behavior so I could be more effective. Intel was working with a psychologist who specialized in understanding corporate culture. After a few sessions with me, he said that I was too enthusiastic about my proposals and did not appear to be balanced by not presenting the downside. I was, in other words, too positive. He also thought I was relying too much on my intuition, not enough on objective data. This was also true, but I was a strong believer in instinct. Instinct is when you know something but don’t know how you know it. Not knowing how you know something does not make it not true. But in the highly analytical engineering culture of Intel, it certainly made it suspicious.
After working with the psychologist, I learned to be much more concrete and balanced in my presentations. I gave as much time to the downside as the upside, and I kept my enthusiasm at bay. His advice was great. Over time, as I had more successes, my views were given more credibility inside Intel, and I could revert back a bit.
While doing the research for this book, I interviewed many who worked either directly or closely with Andy. All held him in high regard. Often, they would say that while Andy could be very opinionated, he could be persuaded to change his mind when confronted with facts and data. This made me realize that one of my issues with him was that I could not present facts and data about how I thought things would unfold in the future. I was using my intuition.
It turns out that I did not have much to add to Intel’s existing business, because my background was in computers and not in semiconductors. I am slightly embarrassed to admit that I never once set foot in a semiconductor factory during my time at Intel. Naturally, I looked for opportunities that fit my skill set; in retrospect, I realize this limited my impact. In Intel’s results-oriented culture, writing reports and making presentations would have little impact. I needed to take action and make things happen. I would need to become an “activist strategist.” But how?
It is very important for companies to expand their business.
Successful companies usually begin with one product. Over time, they add additional products, which they sell to existing customers, and they acquire new customers/markets for existing products, including geographical expansion.
S E L A S F O E G A T N E C R E P
S C U R V E
100% • • • 80% • 60% • 40% • 20% • • • 0% • • • Slow Growth Fast Growth Slow No Growth Products normally go through what is called an S curve. In the beginning, a product may not have enough features/capability to attract many customers. As capabilities are added or, sometimes, as prices are lowered, there are more and more customers. After a while, all the potential customers have bought the product and the market turns into a replacement business. It becomes harder and harder for a company to add capabilities that are so compelling that customers will want to upgrade. Sales level off and can even decline. Generally, the time for a company to add new business is when the current business is experiencing rapid growth. That is difficult because the business that is growing rapidly consumes most of a company’s resources and attracts the best talent.
V E R T I C A L vs. H O R I Z O N T A L I N T E G R A T I O N End User Compaq Cisco [PCs] [networks] Microsoft Intel Seagate Broadcom [software] [microprocessors] [disks] [semiconductors] Graph illustrating vertical integration Companies have a number of options for expansion. They can do a vertical integration; networks would develop to be formidable businesses. Intel’s strength with the semiconductor and its software skills could play an important role in developing a leadership position in the nascent computer-networking business, I thought. Microsoft was struggling with networking technologies and was not entrenched.
Furthermore, the development of computer networks would also help accelerate the adoption of personal computers to the benefit of Intel’s main microprocessor business. At first, it seemed an ideal area to explore. I realized that while Intel had many of the technical skills to develop networking products, it lacked the marketing and business expertise required. It was doubtful that Intel had the time to develop those skills internally before other companies achieved strong positions. I wanted to acquire a company that could jump-start the process, that could form the nucleus of a networking business.
Les was very supportive of this idea. He even arranged for me to make a presentation to the Intel board of directors on networking. I used the automotive industry as an analogy. I explained how the oil companies, like Standard Oil, pushed the development of highway systems by building small demonstration “highways” and providing free maps. In my description, the highway system represented computer networks, while the cars were the computers. I explained that once the highways were built and legislation limited driving speed, the automotive industry found itself no longer competing on speed but by using racing stripes, car seat fabric, and bumpers. To illustrate this point to the board, I even brought a chip upon which I had painted a racing strip.
If we did not want the same thing to happen to us, I told the board, we had to make sure that computer networks were built and that they were fast. Several members of the board at that time understood more about the computer business than did the management of Intel. This included Max Palevsky, who had founded Scientific Data Systems.
While none of the board had experience with networking, which was still in an early phase of development, they were supportive of my exploring this area.
The networking industry was very fragmented then, with some companies offering network adapters, others offering hubs and routers, and still others offering software and servers. I believed that Intel could become the leader in networking by rolling up a number of these smaller companies. To accomplish this, though, we would need talent to lead this effort. I concluded that the only way to do so would be through acquisition.
As a novice with respect to mergers and acquisitions, I sought the help of investment banks in New York City, like Goldman Sachs and Morgan Stanley. When visiting their offices, I was treated very well as a representative of Intel, an important public company. I discovered that the bankers employed some of the best chefs in New York to work in their executive dining rooms. This seemed to be a big part of their sales approach. I had almost no prior experience working with investment bankers, besides the ill-fated Franklin IPO. The bankers were very enthusiastic about helping me, as they had a number of interests with respect to Intel. They traded Intel stock for their customers, and they would also earn a great deal in commissions on any acquisition Intel might make. They also wanted to increase their standing with Intel, whose treasury now contained significant assets.
The bankers had analysts that followed various industries and made recommendations to the firm’s clients on which stocks to buy or sell. I met with many of them. They helped educate me, not only with respect to potential acquisition candidates in the networking business but also about the technical aspects of how acquisitions worked. There was so much for me to learn. I felt like I was back at school (then again, I never really went to school). In addition to being coached, I read university textbooks on the topic of mergers and acquisitions (M&A). I had to become an expert in high finance in a very short period of time.
I arranged for several of the investment bankers to visit Intel. In presentations to Andy, Les, and Arvind Sodhani, Intel’s treasurer, they outlined strategies for Intel to enter the computer-networking business. These presentations were helpful. There was definitely high-level support for acquiring a networking business, in theory.
We could not invite the bankers to Intel’s executive dining room, because we didn’t have one. In Intel’s cafeteria, everyone from Andy Grove on down stood in line and waited. And everyone had to bus their own dishes. We also had no assigned parking spots or, really, any other perks. Sometimes Intel could be egalitarian to the extreme, probably a result of the founders’ humble roots. I will never forget Andy telling me that he had a vacation home in the ski area of Lake Tahoe, which he owned together with another Intel executive to save costs. He said he did not really like going up there because he had to do so much maintenance work himself. Andy must have been worth more than $100 million at the time, but it did not occur to him that he could hire people to do that work.
A Bit Of Background On Networking
Before personal computers and local area networks, there was time-sharing. Individuals might use a terminal connected to a mainframe or minicomputer. Terminals normally only had an alphanumeric, monochromatic display. They connected to computers by serial port. The display speed was very slow by today’s standards; you could see each character written one at a time on the screen. When personal computers were first introduced into businesses, they used a serial port to connect a PC to the time-sharing computers and ran terminalemulation software. As more and more PCs appeared within the organization, people sought to find a faster way to connect them to what we would now call servers, developing the concept of a client/server network. Email was a very important factor in driving this.
Initially, computer networking was developed to allow mainframe and minicomputers to communicate with each other. In 1975, for example, to connect PDP-11s together, Digital Equipment developed DECnet, a very powerful set of software protocols that could have become the standard for computer networking had Digital not kept it proprietary. Networked systems included computers that were located near each other, computers that were located within a campus, and, eventually, computers that were located remotely from one another.
The Internet evolved from a way to connect computers that were both local and remote. The key to the Internet became the router, a device that can pass information from one computer to another via many paths. Cisco would come to dominate this aspect of networking. Each computer then had to have a way to attach itself to a router.
A number of competing technologies developed and deployed in the early 1980s, called local area networks (LAN), could connect computers and devices located within a relatively small geographical area. One of these technologies, Ethernet, was first developed by Bob Metcalfe at the Xerox Palo Alto Research Center. Later supported by Digital Equipment and Intel, Ethernet came to dominate LAN technology and became the way computers would attach themselves to routers. Servers were developed around the same time to support email and shared printers.
Connectivity across large geographical areas, called wide area networks (WAN), developed much earlier in response to the need to connect mainframe and minicomputers to each other over large distances.
Much of the technology now used for Internet communication came out of this. Using this same technology, clusters of local area networks could be connected together. If you have high-speed Internet in your home, for instance, you’ll notice that your router has one port labeled “WAN” and a bunch of ports labeled “LAN.” This is the same general architecture as was first developed for enterprises. In 1999, Wi-Fi was introduced, with Apple offering the first commercial product. This would prove consequential in the proliferation of computing but did not play a role in my thinking at the time.
— can’t afford anything we want and
don’t want anything we can afford Networks consist of computers that utilize networking technologies like Ethernet to connect to other computers both locally and remotely. While Ethernet provided a very effective way for computers located close to each other to communicate, other technologies would be needed to let computers connect both to other local computers but also computers that were located remotely. The router is what makes all this possible. While there were a number of router companies, I was convinced that Cisco, which had been formed in 1984, would be the winner in this new market.
So in 1988, I wanted Intel to buy Cisco, which was still small then.
We might have bought them for less than $100 million, but Andy thought that was too expensive. Cisco went public in 1990 at a valuation of $224 million. Sometime in 1994 or 1995, Intel took a look once more at Cisco. Once again, Andy passed at the opportunity, saying that Cisco at $4 billion was too expensive. I then bet Andy that one day Cisco would be worth more than Intel. He laughed at the thought. On May 2, 1998, Cisco passed Intel in market capitalization.
Looking back, I may not have done a good job of explaining why Cisco would be so successful. Most likely, I did not appreciate how little understanding Andy had about networking. That’s on me. But it was very hard to “teach” Andy anything, at least in my experience. Les was the opposite. He really wanted to understand, and he had a big appetite for knowledge.
Having been turned down by Andy with respect to Cisco, I then considered 3Com as a possibility. Ethernet provided a way for multiple computers to all connect to each other over the same network and a small geographical area, and a device called the router allowed both local and remote connections. The founder of 3Com was Bob Metcalfe, who had developed Ethernet while at Xerox PARC. I knew Bob from my time at Digital when I gave his new company, 3Com, a very large order for Ethernet cards. Though the leader in Ethernet, they were having issues in their market, which was getting to be very competitive.
Andy thought they were a “piece of crap.” When he told me that, I angrily responded, “I got it; you don’t want anything we can afford, and we can’t afford anything you want.” That was pretty much true.
Looking back, I don’t think Cisco, 3Com, or any other potential acquisition would have entirely worked out. At that time, though, I was not yet aware of the problems of integrating such acquisitions into a company like Intel. I believe Cisco would never have become the company they became under Intel ownership. While Cisco had a strong technical capability, they were really customer focused. 3Com might have been a more successful acquisition, but as part of Intel, it would have had difficulties competing with network leaders like Cisco.
The next company I considered was Sytek, the original broadband company formed in 1980. In 1982, General Instruments (GI), the leading supplier of products to the cable television industry, began a series of key investments in Sytek that led them to own 57 percent of Sytek. Though Sytek had ample opportunities to raise venture capital, they believed a strategic investment from GI made more sense. GI was interested in Sytek’s development of MetroNet, one of the company’s main efforts and one with which GI could help directly. Sytek and GI’s collaboration on MetroNet could have been the foundational technology for interactive television and became one of the forerunners of broadband Internet.
Sytek combined off-the-shelf components that were readily available in the television industry, coupled to a key Ethernet component that Intel had developed. There was a demonstration of this technology in Sacramento, California, in 1982. This was the first metropolitan area network, I believe. Later, in 1990, a Boston-based company, LANcity, would do something similar, which they actually brought to market.
Though General Instruments was interested in the residential market, the cost of TV set tops capable of connecting to this network was estimated to exceed $1,500, which made it unviable as a consumer product. Consequently, GI began to lose interest in Sytek. In 1992, Intel and GI would partner, utilizing many of the same concepts to develop the cable modem technology that evolved into what is now utilized by the cable industry for residential broadband.
Fortunately for Sytek, its executives ran into twelve members of the IBM Personal Computer Group at a networking seminar in 1983.
By that time, the IBM PC was doing extremely well and was becoming a significant contributor to IBM’s profitability. The PC group needed LAN technology to serve their market. IBM was in the process of developing its own proprietary network technology called Token Ring, which was meant to compete with Ethernet. The PC group got tired of waiting for Token to be developed. They decided to use Sytek’s technology because it could be used to connect local networks but also to connect enterprises including buildings miles apart. IBM made a small investment in Sytek. The PC group began marking Sytek’s product, which was called LocalNet 2000. IBM and Sytek jointly developed a critical piece of networking software, NetBIOS. Later, Microsoft would incorporate this into their operating system.
Eventually, internal IBM politics killed the Sytek approach and required their PC group to wait for Token Ring. At the same time, the rest of the PC industry moved to Ethernet, which eventually won out and still is the standard for local area networking.
Sytek was in financial trouble once IBM had terminated their purchase agreement and moved to Token Ring. General Instruments, which no longer had any strategic interest, wanted to sell its 57 percent stake. As a result, the Sytek board decided to sell the whole company and engaged an investment banker to conduct an auction. Intel was contacted by their investment banker in 1989, and we began to evaluate making an offer. Les and I liked Sytek’s CEO, George Klaus, and thought he had the potential to lead an Intel networking business. The company had a number of strong technologists who could have helped Intel a great deal.
After many internal meetings, we finally decided to make an offer.
I then learned an important lesson about managing the acquisition process. Companies up for sale often hire an investment banker to represent them, and the banker may recommend an auction. The banker prepares an initial presentation on the company and reaches out to a number of potential buyers. Next, they narrow it down to a few possible buyers that appear to be very serious and a good fit. Those companies get an opportunity to review a lot more material about the company for sale and meet with all the key employees. The potential acquirers then place bids. The company to be acquired chooses among those bids based on a number of factors, with financial concerns typically prioritized the highest. This process tends to work well for the seller, as buyers who know they have competition will tend to offer more in order to be successful in their bid.
If the potential buyer has a lot of people involved in the deliberations around an acquisition, there is a good chance that someone will say that the price being considered is too high. Usually, then, others will jump on board to say that the price was too high. Later, if you can get the business for the lower price, they will claim they were right. If you lose the bid, they will say the acquirer overpaid. This happened exactly in the case of Sytek. We lost the bid to Hughes Networking, a division of the Hughes Aircraft Company, which later created DirecTV.
Sytek became Hughes LAN Systems. Our bid was $20 million less than the Hughes bid of $87 million. After the acquisition, their focus shifted to using satellites for data communications. Eventually, this part of Hughes, Hughes Communications, was sold to EchoStar.
I found this loss very upsetting. It was my real first attempt at an acquisition. I realized that Intel would probably never successfully bid in an auction because we would always underbid. The process works differently when you reach out to a company that has not been thinking of selling and enter into negotiations.
With the advantage of hindsight, I now think that had we successfully acquired Sytek, the evolution of both the networking business in general and Intel could have proceeded very differently. With Sytek’s talent, we would have developed the semiconductors needed for broadband. Coupled with Intel’s existing Ethernet capabilities, Intel could have become the leading semiconductor supplier in the broadband market, the position later occupied by Broadcom. Perhaps we would have used the knowledge we gained in the acquisition to develop network server technology, as well. Sytek would have been a better cultural fit with Intel than the later acquisitions we made.
The Antibodies Attack
Having failed to acquire Sytek, we ended up buying a small networking company, Jupiter Technologies, in July 1989. Jupiter, a software company, dealt with protocol translation. Before TCP/IP became dominant, different computer systems had varied and incompatible methods of communicating. Jupiter wrote software that translated among these different systems.
What I wanted in Jupiter was the talent to build Intel’s networking business. Jupiter’s CEO, Jim Flach, moved out from the Boston area to work at Intel in Santa Clara. He was made vice president of the Systems Group, reporting to Les. Then the Intel antibodies came out and attacked the “foreign body.” Finance, legal, and human resources all would swarm over the “body” of any acquisition, forcing it to adopt Intel’s way of doing things. Intel people would treat individuals within the company being acquired as second-class citizens. Though I could provide some cover to senior executives from the acquired company, I could not alone manage the magnitude of the reaction from the Intel “immune system.” I was so naïve. Integrating a new company is a very difficult task that requires dedicated, experienced management that we did not have in place. Jim Flach stayed at Intel for three years before leaving to become a partner at Accel, one of the leading venture capital firms.
In May 1991, we also acquired a division of LAN Systems, a company founded and led by Tyrone Pike, a serial entrepreneur. Tyrone joined Intel but did not stay long. However, Ed Ekstrom, the person who led the division we acquired, stayed for many years and made many contributions. PCEO was already reselling LAN Systems’ network diagnostic products.
— the challenge of an acquisition is keeping the talent
Though the LAN Systems acquisition was successful, I was learning how difficult it was to acquire companies within Intel. I had thought that by buying some smaller companies, we could get both critical technical skills and entrepreneurial management, which together would help Intel to create its own networking business. This did not work out as planned. The management of acquired companies generally did not want to be part of Intel. Entrepreneurs by nature greatly preferred to run their own businesses. Ultimately, we did not successfully acquire the senior talent we needed to develop our own networking business.
I still believe that networking was a great opportunity for Intel but did not know how to get it to happen.
Looking back, I have to conclude that to build a new business within Intel, there had to be a champion—a person willing to put their career on the line to lead the effort. Perhaps, if I had told Les and Andy I wanted to start a networking business for Intel, they would have agreed. But my experience at Digital and then later at Franklin had convinced me that an operating role was not for me.
Zero-Billion-Dollar Businesses
When Frank Gill took over the Systems Group from Vadasz, it already had a small networking business providing Ethernet cards and print servers. Slowly, Frank came to the same conclusion as I had, along with many others: Intel had an opportunity to create a large networking business. Frank wanted to grow the business internally; he was not interested in acquisitions or strategic investment. The Systems Group was renamed the Small Business and Network Group. Eventually, this business did more than a billion dollars in annual sales, but Grove would still not invest in it. He required that the business be selffunding. Andy even taunted Gill, comparing the microprocessor business with Gill’s network business: “Frank, you sell a billion dollars in one year. I earn a billion dollars in profit every quarter. Why should we invest in your business?” By the time Andy asked this question, he might have been right. The networking business was maturing, and companies like Cisco were very entrenched.
Intel’s revenue and especially its profit were so high that Andy saw a billion dollars as a rounding error. I would sometimes refer to the new business efforts as the zero-billion-dollar businesses. On the other hand, Intel Capital would invest in start-ups that barely had revenue and were sustaining substantial losses and watch them grow into companies with billion-dollar revenues, but none of them did this in their first year.
Early in 1998, Frank Gill had proposed acquiring FORE Systems, which was a communications switch company. Craig Barrett, who had just become CEO, very much supported this acquisition and brought it for approval to his first Intel board meeting as CEO. Andy Grove was now chairman of the board. The cost of the acquisition exceeded $2 billion. In a very unusual move, Gordon Moore, who was serving as vice chairman, spoke against the acquisition. Then the other directors agreed. Frankly, I doubt that many of them even understood the issues. Andy, who had favored the acquisition, just sat the discussion out. Soon after this rejection, Gill was gone. He left Intel soon after Barrett was made CEO, a job that Gill most likely wanted for himself.
Mark Christensen, who had reported to Frank, took over, but the networking business slowly died out. While Frank never thought I was a “real man” because I did not run a business, several of my networking investments earned more profit for Intel than Frank made the whole time he ran the Systems Group.
By the end of the ‘90s, Intel came to recognize that there was a major opportunity to provide the critical silicon or, as Intel would call it, “the building blocks” of the Internet. The company started an acquisition spree starting with Level One Communications, which was purchased for $2.2 billion, followed by the purchase of a Danish company, Giga, for $1.25 billion, which specialized in the chips needed for high-speed optical communications. Intel started building a portfolio of specialized semiconductor companies. It acquired tens of companies. From discussions I had with a number of individuals that were involved, it appears that they overpaid, perhaps did not do an adequate amount of due diligence, and had difficulty with integrating the new companies within Intel. Intel had hoped that its manufacturing strength would allow the company to offer the communications/networking components with a price advantage to other companies. Intel would be selling these components to companies like Cisco, Juniper Networks, and Lucent, who, in turn, sold to network operators like AT&T. Once again, Intel was several steps removed from the end customer. In 2002, network operators experienced a severe decline in their sales, and they cut back on their purchases from companies like Cisco, who stopped buying product from Intel. In 2007, Intel sold off its optical communications business to Encore for $85 million.
Doubling Down
After I left, Intel acquired many companies during this period—more than fifty. It looks to me like they did not really have a strategy and were not very good at due diligence and terrible at integration. Intel probably wasted many billions of dollars. Paul Otellini, who was then CEO, wrote in the 2007 annual report that Intel was divesting itself of this business to focus on markets where Intel CPU architecture gave the company a competitive advantage. Intel had once again failed at establishing a new business. It was a very expensive experiment.
Intel would have quite a successful program to accelerate the adoption of Wi-Fi driving standards, integrating Wi-Fi with processors like Centrino. Intel Capital carved out $150 million for early-stage investments in companies focusing on Wi-Fi opportunities. Sriram Viswanathan, whom I had trained, led that successful effort. However, while this certainly resulted in an expansion of PC sales and therefore Intel microprocessor sales, Intel did not participate directly in the market with Wi-Fi products.
Intel also started to assemble a number of companies to provide the technologies needed on the client side of residential broadband.
The leader in that market was Broadcom, a company in which Intel had once owned a substantial share of stock. Intel considered buying an Israeli cable modem component company, Libit, which we had invested in during my time at Intel Capital. Texas Instruments acquired that company instead. Then, years later, Intel acquired all of Texas Instruments’ cable modem business. It would take another eight years, but Intel would become the leader in broadband components, twenty-five years after we started working on the first cable modem.
Then in 2020, Intel would exit the market.
Looking back, I realize that Intel in the early ‘90s should have acquired Broadcom and Qualcomm when those companies were still small. Those acquisitions, along with the acquisition of a leading graphic chips company like Nvidia, would have created a much stronger platform for Intel’s future. In 2021, Qualcomm is worth almost as much as Intel. Nvidia is worth twice as much as Intel. Broadcom was acquired in 2015, and while it would have been a successful acquisition, I think, it would not have been as meaningful as the other two.
My Last Acquisition Attempt
While I was exploring how to create a significant networking business, I was also trying to help Les grow the Systems Group’s PC business. In 1990, I found a successful UK company, Apricot Computer, in Birmingham, which had built an IBM PS/2-compatible computer.
I believe this was the only PS/2-compatible computer ever built. My interest in Apricot was primarily focused on the strength of their technical team. I thought that the acquisition of Apricot would have given us a solid platform for European expansion.
My attempt to acquire Apricot turned into a real disaster. Les had criticized me for not working closely with key internal groups. He was more or less correct, so with Apricot, I worked really hard to get all the key people in the Systems Group to review and approve the deal, particularly the technical people. Then, at a meeting to approve making an offer to buy Apricot, some of the technical people surprised me and started attacking the deal. They were calling in from Oregon while we were in Santa Clara. I couldn’t understand why they were doing this; the technical people had all previously told me they supported the deal. Then Andy started in on me, which, believe me, was not fun. To my surprise and great disappointment, Les, rather than defending me, piled it on as well. The deal was rejected. I was very angry with Les for not supporting me. This was one of only two times we had a major conflict. This was a big one; I stopped speaking to him at all for some time.
I did not give up on the deal. Those who know me well can testify to my tenacity. I began discussions with Giorgio Ronchi, the CEO of Memorex Telex, a company selling IBM-compatible disks and terminals. We discussed setting up a joint venture to sell IBMcompatible PCs from Apricot through the Memorex Telex sales force.
With Memorex Telex interested, I brought the potential acquisition of Apricot back to Intel. As I expected, having a serious customer for Apricot’s products changed Intel’s thinking about the acquisition. Still, it took me many months to get the deal approved. Then we had to complete due diligence.
Dick Boucher, to whom I reported while Les was on sabbatical; Arvind Sodhani, Intel’s treasurer; and I were the senior team responsible for the due diligence. This was to be my first large-scale acquisition.
Every department in Intel—finance, legal, human resources, and security, among others—had to get involved to verify everything Apricot had sent over and look to bring any problems to the surface. Some twenty people came with me to the UK to conduct the due diligence on Apricot. I have no idea how it came about that so many Intel people were involved. In retrospect, I should have made sure I understood in advance the totality of the effort. For instance, since we would be buying assets like buildings, the people that managed Intel’s facilities wanted to be involved.
Soon after we started our work at Apricot’s headquarters, I realized that I had lost complete control over the process. After three days, the chairman of the board of Apricot Computer, Roger Foster, asked me to come to his office. He was calling off the deal, he said; he would never subject his people to working for a company like Intel. Pointing to his window, Roger showed me what was, for him, the final straw. There we saw an Intel crew digging a big hole in his parking lot, looking for any hidden industrial pollutants: just standard Intel procedure, a step that would have made sense if we were buying a factory that manufactured semiconductors.
Boucher and I collected all the Intel people and returned to the USA. Apricot was the last time I tried to complete this kind of acquisition at Intel. Larger acquisitions of operating companies are very difficult in high tech, although some companies, such as Cisco, have developed skill at doing this. John Chambers, Cisco’s CEO, discusses this in depth in his book, Connecting the Dots. Acquisitions of smaller companies to get talent and technology can make sense, as Apple, Google, and Microsoft have demonstrated. While I still believed that Intel needed to create new businesses, I no longer saw how I could contribute to making this happen.
← Prev: Chapter 8 — Failing Upward
→ Next: Chapter 10 — Adventure Capitalist