The first desktop computer dates to earlier than you probably think. And officially at least, it was an accident. Great inventions often are. But it was surprisingly similar to desktop computers that followed it.
Design work on the first desktop computer commenced in 1969. Yes, you read that right. It predated the Apple II by several years, and the IBM Personal Computer and IBM compatibles by a good decade. And it wasn’t built in Silicon Valley either. But this ahead-of-its-time oddball is the direct ancestor of your modern desktop or laptop computer, right down to the Intel processor design.
Computer Terminal Corporation
Computer Terminal Corporation was founded in July 1968 to build CRT-based replacements for Teletype machines to connect to time sharing systems, the equivalent of what we would today call a minicomputer or a mainframe computer. Its first terminal, the Datapoint 3300, was immediately successful, and CTC soon was selling them to the likes of Digital Equipment Corporation and Hewlett Packard to sell under their own names.
By November 1969, CTC, fresh off an IPO, started work on a successor. Their idea this time was to build an intelligent terminal that could connect to a wider variety of computers. Victor Poor and Harry Pyle, two CTC engineers, worked out a design for a processor on a living room floor over Thanksgiving weekend. The company founders, Phil Ray and Gus Roche, asked if the design could be implemented on a single computer chip. Poor and Pyle believed it could, and estimated the cost at $100,000.
Getting the design into silicon
Ray and Roche arranged to have dinner with the presidents of Intel and Texas Instruments, CTC’s two main chip suppliers. There they presented the two companies with Poor and Pyle’s design for an 8-bit processor and worked out a deal. Intel president Bob Noyce thought it was possible, and interesting, but was concerned it could affect their other lines of business.
It turned out Noyce was right. Just not in the way he expected.
Neither TI nor Intel finished the chip quickly enough for CTC. TI’s implementation of the chip was ready in March 1971. By most accounts CTC rejected it due to reliability issues. Intel’s implementation took even longer, and Intel didn’t have anything to show CTC’s until the end of 1971.
Meanwhile, CTC’s main motivation for using a single-chip processor was heat. They found they could solve their heat issues by using a switching power supply. So CTC decided to implement the processor using discrete TTL logic along with a switching power supply and released the Datapoint 2200 in May 1970, before either chip was ready. Then CTC stuck with its discrete TTL design for performance reasons. Intel wasn’t able to beat CTC’s performance until about 1982.
So CTC invented the microprocessor and then ended up not using it. So the first 8-bit microprocessor ended up not powering the first desktop computer after all.
The Intel 8008
Later that year, Intel found another potential customer for the 1201, calculator maker Seiko. Intel worked out a deal with CTC. Intel let CTC out of its contractual obligation of $50,000 in return for being able to keep the chip and sell it to others. Roche objected, but the rest of the decision makers at CTC didn’t see a need to retain rights to the chip and opted to save $50,000. Letting the Intel CPU get away was perhaps a worse business blunder than Decca Records rejecting The Beatles.
When the chip was finally ready, Intel renamed it to the 8008 and released it in April 1972, priced at $120. Intel revised the design and it became the 8080, which then led to the 8085 and the hugely successful 8086 family, which included the Intel 8088 CPU used in the original IBM PC.
Intel implied that the 8008 was derived from the 4-bit 4004 CPU, but they were separate projects commissioned by different companies. TI filed a patent that covered the 8008, and tying the two chips provided Intel a workaround.
Reception in the marketplace
Intel’s chips captured the attention of Gary Kildall, who wrote his CP/M operating system for the chip. Kildall’s invention led to a generation of desktop and portable computers that were very similar architecturally to the Datapoint 2200. An upstart named Microsoft, founded by college dropouts Paul Allen and Bill Gates, developed its first product for an early desktop kit computer powered by an Intel chip as well, the Altair 8800.
But all of this was happening in the mid 1970s. Companies like Pillsbury were already buying Datapoint 2200s and using them as desktop computers by then. By some accounts this was the intent all along. By other accounts, CTC saw how Pillsbury was using the machine and decided to promote it as more of a general purpose computer.
Texas Instruments TMC1795
Texas Instruments’ version of the chip, the TMC1795, wasn’t successful. Originally called the TMX1795 while it was in development, CTC rejected it for various reasons. Voltage fluctuations made the chip unreliable. The TI chip also had broken interrupt handling due to a mistake in some Intel documentation, and it’s unclear how an Intel mistake ended up in TI’s chip. CTC had a design with old-fashioned conventional TTL logic ready to go 10 months before TI had production silicon. And finally, while TI’s chip outperformed IntelOne reason CTC rejected Intel’s chip was performance. Their TTL implementation ran faster than Intel’s integrated design. It’s possible, even likely, that TI’s performance lagged behind as well.
But TI didn’t quietly fade away. TI tried unsuccessfully to sell it to other companies, including Ford. But they positioned it as a microcontroller more than as a general-purpose CPU. The TMC1795 and its revised version, the TMC1795A, were unsuccessful and Texas Instruments shifted focus back to its highly profitable calculator chips.
Had TI thought to try to market the TMC1795 as a general purpose CPU, history may have worked out differently. Especially since TI had about a 10-month lead on Intel. Arguably the TMC1795 was the world’s second microprocessor, having just missed beating the Intel 4004 to completion. But since TI never found a customer, it never entered large-scale production. The second successful microprocessor was the Intel 8008.
The idea that the TMX1795 never worked is a misconception, however. In 1996, TI demonstrated a functioning system based on the TMX1795. But at that point, the argument was academic. The chip was a technical success but a marketing failure. It didn’t power the first desktop computer or the second.
In the 1990s, TI partnered with Cyrix to try to compete with Intel’s x86 CPUs, but with limited success.
The Datapoint 2200
The Datapoint 2200’s industrial design and form factor was a product of its era. It was the size of an IBM Selectric typewriter, and the screen, which could display 12 rows of 80 characters, was the same shape and proportion as an IBM punch card. It looked like it belonged, even though it didn’t look like the device CTC sought to replace. The screen was about half the height of later screens used by PC clones that ran MS-DOS. But for a disruptive technology to catch on, it needs to look right for its time period.
Neither Intel nor TI was able to deliver the microprocessor on time, so CTC built the Datapoint 2200 using discrete TTL logic. Instead of a microprocessor, the processor was a collection of about 100 TTL chips working together. This was standard for the time, and one reason Intel had reservations about the idea of a microprocessor. Intel was concerned about selling fewer chips. Intel engineer Ted Hoff estimated that using Intel’s CPU would have saved CTC about $50 per machine.
CTC’s implementation was mostly compatible with what became the Intel 8008, which makes sense, since Intel built it to CTC’s specifications. It had 2 KB of memory, expandable to 16K and two tape drives for storage. It sold for about $5,000, which is over $32,000 in 2018 dollars. The Datapoint 2200 was a real computer and it fit on a desk, but it certainly wasn’t priced for hobbyists.
General Mills was the first customer for the Datapoint 2200.
Why the Datapoint 2200 is the first desktop computer
Poor and Pyle and their programmer, Jonathan Schmidt, ostensibly designed a programmable terminal. That is, their device was a terminal that loaded various emulations off the included tape drive. But they did nothing to ensure a terminal program was the only thing it could load and execute. What they built was a general purpose machine with 2 kb of RAM, an Intel 8008-compatible processor, and a tape drive. It bore an incredibly strong resemblance to a late 1970s home computer.
Whether the first desktop computer was an accident, or whether they designed a computer and sold it as a terminal to avoid spooking investors is always a matter of debate. Computer engineers often build features in and then see what end users will do with it. But CTC’s employees insist the company designed the Datapoint 2200 as a computer in disguise. And in 1972, CTC filed patents on the machine’s design, pitching it as the first personal computer.
The price certainly wasn’t low by today’s standards, but in 1970, a small general-purpose computer that sold for $5,000 was interesting to a large corporation. It wasn’t long before companies started using Datapoint 2200s as general-purpose computers, and the peripherals CTC offered for sale with it, including 8-inch floppy drives, 2.5-megabyte hard drives, and printers, certainly encouraged their use beyond just connecting them up to larger computers to serve as a keyboard and display. It proved popular. In 1976, CTC sold $72 million worth of the machines.
What happened to the Datapoint 2200’s inventors
Victor Poor went on to help CTC develop ARCNET, an early computer networking standard. He stayed with CTC until 1984, when he retired. By then the company had changed its name to Datapoint and fell victim to a 1980s-style corporate raid. Poor had accumulated stock in the company over his 15 years of employment with them, the sale of which gave him enough money to retire on. Poor died of pancreatic cancer on August 17, 2012. He was 79.
Harry Pyle also helped in the development of ARCNET and also stayed with CTC until the mid 1980s. From 1985 to 1996 he worked at a company called Image Data, which specialized in the field of remote radiology, allowing physicians to read X rays remotely. In 1996 he worked on the user interface for Bill Gates’ estate, and in 1999 joined Microsoft as a software engineer. He died of cancer on April 6, 2013 at the age of 63.
Jonathan Schmidt also worked on ARCNET and stayed with CTC/Datapoint until the mid 1980s. After leaving Datapoint he founded a new company called Perftech.
As for the company founders, Gus Roche died in a car accident in August 1975. He was 46. He had left Datapoint within the last year and started two new companies, R&R Research Corp. and Mnemonics Ltd., which were developing computer memory systems.
Phil Ray died of cancer in Bethesda, Maryland on August 13, 1987. He was 52.
What happened to Datapoint
The company that gave away what became the x86 architecture went on to become a Fortune 500 company, reaching #447 on the list in 1983. But Datapoint’s sales model lost effectiveness as the 1980s wore on. Then in the early 1980s, Datapoint endured an accounting scandal. In 1984, a federal judge reprimanded the company for overstating its 1981 earnings by $22.1 million.
In late 1984, corporate raider Asher Edelman took over Datapoint, which led to a number of key people leaving. During the process, Datapoint laid off hundreds of workers. Edelman spun off the services division named Intelogic Trace. Then Edelman used Datapoint’s and Intelogic Trace’s financial holdings to buy other companies. Datapoint limped on and survived until May 2000, when it filed for bankruptcy and broke up into several companies. One of the resulting companies specializes in call center equipment and the other maintains Resource Management System, a proprietary operating system developed by Datapoint that dates back to the 1980s.
Datapoint’s story is long and complex. It’s a key part of computer history, even if it’s easy to overlook. The modern computer industry was born in the United States. But it’s easy to argue its birthplace was San Antonio, Texas, rather than Silicon Valley.
Lamont Wood chronicled the company in his 2012 book Datapoint: The Lost Story of the Texans Who Invented the Personal Computer Revolution.