Chapter 7 - Data Communications: Market Order 1973-1979
7.1 Minicomputers, Distributed Data Processing and Microprocessors
Demand for modems and multiplexers surged from 1968 to 1970 due to the huge success of the terminal-based IBM System/360 and the commercialization of timesharing. And although timesharing suddenly collapsed in 1970-1971, the pick up in sales of mainframe computers from a low of 5,700 units in 1970 to a high of 14,000 units in 1973 meant sales of data communication products continued to grow at rates above 30% per year. The transition to terminal-based, on-line, real-time computing had happened and, combined with the increasing use of remote terminal access, data communications had become a rapidly growing business.
A next surge in data communication growth would arrive in the mid-1970s, caused not by mainframe computers, but minicomputers. The minicomputer revolution began between 1968-1972 with the formation of ninety-two new competitors. By 1975 sales totaled $1.5 billion. The first minicomputer markets of embedded and engineering applications created little demand for data communications. But by the middle of the 1970’s, minicomputers found a welcomed home in both large and mid-sized corporations performing financial and administrative functions. Driving this trend in large corporations was first the ever-expanding backlog of software development projects of MIS departments that frustrated financial and operational management, and second the need of remote operations for computing to invoice customers or keep track of inventory or generate timely reporting. In 1979, 81,300 minicomputers were sold compared to 7,300 mainframe computers. The demand for data communications products in the form of modems and multiplexers soared.
The genesis of the minicomputer is the same as that of the modem: the SAGE Project. In 1953, Kenneth Olsen, a recent graduate of MIT and one of 400 engineers hired to staff the SAGE Project, was reassigned to work as a liaison to IBM, the firm contracted to manufacture the SAGE computers. When it was time for a new assignment, Olsen went to work for an advanced engineering group at MIT Lincoln Labs led by Wes Clark. Clark had approval to build the first transistor computer, the TX-2, considered the first personal computer. In early 1957, Olsen left to test his entrepreneurial skills and, together with Harlan Anderson and thanks to venture capital provided by American Research & Development, they founded Digital Equipment Corporation (DEC) in August 1957. Their first product, released in 1959, borrowed significantly from the TX-2. In the fall of 1965, DEC introduced the first commercial minicomputer: the PDP-8. In 1967, DEC went public with a valuation of $77 million, 770 times its founding valuation. Every venture capital investor sought the next DEC, thereby contributing to the explosion in the number of minicomputer companies between 1968-1972.
As minicomputers invaded corporate America, a challenging new condition arose. Corporations found themselves with multiple mainframes and minicomputers. And it was a condition becoming increasingly pronounced. The need was obvious: to interconnect all the computers together so that data could be shared and that users would have timely access to all appropriate computers. It was the same need that had motivated IPTO of ARPA to launch the ARPANET. Only this was the world of corporate computing of the mid-1970s. To make sense of this weed-like growth phenomenon, a new vision of corporate computing emerged to replace the IBM mainframe “Host centric” model: distributed data processing (DDP). DDP represented a network of many interconnected computers often with no central host. Only for DDP to work, computers and computer devices had to be networked together, a problem with no comprehensive solution even as late as 1977.3
As early as 1973, data communication companies had introduced modems with diagnostics, a first step in creating networking products. Even these first primitive products necessitated more software and computational capabilities than any products that preceded them. Fortunately, LSI-semiconductor components and microprocessors made the evolution to network-based products possible. Microprocessors, at this stage, were simply components in data communication products. Then the early 1980’s, microprocessors would engine personal computers and fuel an explosion in networking, the second wave in computer communications after data communications.
The announcement of the first microprocessor came on November 5, 1971, from Intel Corporation, a venture-capital funded start-up founded in 1968. In announcing the world’s first “micro-programmable computer on a chip,” the 4004 microprocessor, Intel claimed it would usher in “a new era of integrated electronics.” It was advertising hyperbole as master understatement. The 4004 was as small as the first three letters of the word ENIAC, yet equaled in computational power that thirty ton first computer. Even so, Intel management debated whether or not to introduce it. How would the 4004 be used when the total installed base of computers numbered only 88,000? Eight years later, in 1979, an astonishing 75 million microprocessors would be sold4 – 329,000 as microcomputers5 – four times the number of minicomputers (81,300) and forty five times the number of mainframe computers (7,300).6 Microprocessors would revolutionize computer communications.
The origin of the microprocessor was not some profound vision or carefully crafted product development strategy. In the summer of 1969, Busicom, a Japanese calculator company, approached Intel needing integrated circuits for its next-generation calculators. Intel’s Marcian E. “Ted” Hoff, Jr. considered their design of twelve chips too complicated and beyond Intel’s manufacturing capabilities.7 Still wanting to find a solution, he reflected on the design of the DEC PDP-8 minicomputer he was using. The PDP-8 had a very limited set of instructions, but because it had lots of memory, complicated control and logic operations were possible. Fortuitously, Intel manufacturing was perfecting a new process to create reliable chips of 2000 transistors, chips complex enough to implement such a design.8 In the late fall, Busicom executives approved Hoff’s design. Only by the time samples of the four chip set were ready, competition had driven end user prices for calculators to unexpectedly low levels, forcing Busicom to renegotiate for lower pricing. In the process, Intel regained rights to sell the 4004 chips for non-calculator applications; a right, however, that meant little if there was no demand.9 Ed Gelbach, the senior vice president for marketing at Intel, argued new applications should be the focus, not sizing the market potential as ten percent of the minicomputer market. Even though stretched for resources, management took the gamble. By February 1972, an encouraging $85,000 of MCS-4 chip sets had been sold.10
The story of the next generation 8-bit microprocessor was much the same. In late 1969, Computer Terminals Corporation (CTC), later to be renamed Datapoint Corporation, approached Intel and Texas Instruments (TI) to design and manufacture a LSI chip for a new intelligent CRT terminal.11 But before either company could produce chips, CTC abandoned the product. Intel continued their development program, however, and in April 1972 introduced the first 8-bit microprocessor: the 8008. Next, management explored whether recently advanced memory fabrication technologies might be used manufacture the 8008.12 It quickly became obvious that to do so required significant layout changes of the 8008. Management decided to re-design the chip and in 1974 introduced the 8080, the first second-generation 8-bit microprocessor.
By July 1974 nineteen microprocessors were either available or announced.13 The other early entrants were large existing corporations: Texas Instruments, Rockwell, Fairchild, National Semiconductor, Signetics, Toshiba, AMI and Teledyne Systems. These firms grew rapidly in 1973-1974 to keep up with demand and then had to adjust to a mini-recession in 1974-1975. Even so, new product introductions continued. By mid-1975, there were forty microprocessors and, by 1976, the number would grow to fifty-four. In 1974, revenues of microprocessors totaled $37.7 million. It was the year the installed base of microprocessors passed both minicomputers and mainframe computers.14 In 1975, revenues were projected to reach $450 million by 1980, an annual compounded growth rate of 50%.15
The first real competitor to Intel’s 8-bit processors did not emerge until 1976 and when it did it came from a firm formed by former Intel employees. In July 1974, Ralph Ungermann, manager of microprocessor development systems, and Federico Faggin, a key member of the 4004 team, left Intel and talked of doing something together. Before they could act, Exxon Enterprises persuaded them to start a semiconductor components company to support Exxon’s intended office systems businesses. Ungermann and Faggin founded Zilog Corporation. Soon after, they recruited Masatoshi Shima, the lead designer of the 8080 joined them. The goal of the microprocessor architecture was obvious: a chip both compatible with, and better than, the 8080. In 1976, Zilog announced the Z-80. It was an instant success and threatened the lead established by the 8080.
By 1976, a surge of next generation 16-bit microprocessors would come to market.16 Intel, late to market with its 8086, still held the significant advantage of a large installed base of 8080s that could be upgraded easily to the code-compatible 8086. Rumors circulated that Zilog would repeat their 8-bit Z-80 success with a 16-bit microprocessor. Only when they introduced their Z-8000 in early 1979, they made a fatal mistake: the Z-8000 was incompatible with the Z-80. This error nearly killed Zilog and created significant problems for many of the firms using the Z-80 and needing to upgrade to a 16-bit microprocessor.
Intel would win the battle of 16-bit designs. In 1979, Intel introduced the 8088, an 8-bit processor with a 16-bit architecture, and, by doing so, opened up a market that would seem to have no end: personal computers. Personal computers would transform the already emerging Networking wave of computer communications and alter the fates of the data communication firms that had entered Networking with products rooted in their data communication technologies.
“Tomorrow’s people to determine DDP’s price tags,” Data Communications, Sept. 1977, p. 33
Robert N. Noyce and Marcian E. Hoff, Jr., “A History of Microprocessor Development at Intel.” IEEE Micro, Feb 1981, p. 8
A microprocessor sold as a computer, not embedded in some other product, is a microcomputer
CBEMA, Industry Marketing Statistics
Robert N. Noyce and Marcian E. Hoff, Jr., “A History of Microprocessor Development at Intel.” IEEE Micro, Feb 1981, p. 9
Ibid, p. 13
Stephen P. Morse, Bruce W. Ravenel, Stanley Mazor and William B . Pohlman. “Intel Microprocessors – 8008 to 8086,” Computer Magazine Oct 1980, p. 43
From PMOS to NMOS
Robert N. Noyce and Marcian E. Hoff, Jr., “A History of Microprocessor Development at Intel.” IEEE Micro, Feb 1981, p. 14
Ibid., p. 17
Gene Bylinshy, “Here comes the second computer revolution,” Fortune Nov 1975, p. 138
National Semiconductor was actually first, announcing it’s Pace chip in 1974