[TUHS] Demise of AT&T

responding to the SVID thread: <https://www.tuhs.org/mailman3/hyperkitty/list/tuhs@tuhs.org/message/UFYAOAVWUNJ646WT5YWD6T2HHOKPZWEB/> This is my timeline for AT&T : 1956 - Consent decrees for IBM & AT&T, barring each other from entering the others’ business. 1964 - AT&T engages with MULTICS, to build an "Information and Computing Service”, supplying the customer ‘outlets’ 1969 - AT&T / Bell Labs withdraws from MULTICS 1974 - UNIX announced in CACM 1984 - AT&T de-merges, [0] into 7 Regional Operators, keeping ”long lines” and hardware/software development & manufacture. 1991 - IBM declares first loss, increasing losses for another two years. 1994 - Unix sold to Novell, hardware to NCR 2004 - AT&T acquired by SBC [0], a former Baby Bell that’d understood the Mobile phone market & pricing. Both IBM & AT&T were aware of “Silicon Valley” and the rapid evolution of microelectronics. AT&T even considered a 1966 Terman proposal, post-Stanford, to create “Silicon Valley East”. AT&T validly reasoned they couldn’t create it, there wasn’t the needed culture of Cooperation & Collaboration around them. Gordon Bell at DEC certainly understood the changes in technology and admitted “he didn’t believe it” [ i.e. predictions from his own models ]. It wasn’t just AT&T, IBM & DEC that got run over by commodity DRAM & CPU’s, it was the entire Minicomputer Industry, effectively extinct by 1995. [3] One of the causes of AT&T’s management failure was the “East Coat” management culture, documented by Tom Wolfe in 1983 [2] in “The Tinkerings of Robert Noyce”. What Wolfe seems to have missed is the “East Coast” focus on high Gross Margins (50%-90% for product lines, at both IBM & AT&T), compared to the Silicon Valley focus on “Volume” with implied modest Gross Margins: deliberately sharing cost-savings with Customers. An unanswered question about Silicon Valley is: Why did it happen in California and not be successfully cloned elsewhere? There is something different / special about California that for a century has diversified its economy, consistently grown its population faster than other US states, and grown its economy, over a century, faster than any US state or other nation. [ 7 ] I’ve not seen any definitive explanation of the Californian Miracle, it’s incontrovertible in the long-run numbers, before & after Silicon Chips, presumably due to multiple reinforcing factors, that create & maintain exceptional industries. i.e. virtuous circles, like Silicon Valley remaining an economic powerhouse, even as ’technology’ has evolved from Silicon devinces to chips, to software, to Internet Services. The Silicon Revolution didn’t just crush computer companies - it broke other long-term successful innovators: Kodak invented Digital Cameras in 1972, only to be forced into bankruptcy around 2009 after a century plus of operations, continuing after significant restructuring. <https://en.wikipedia.org/wiki/Kodak#Bankruptcy> =================== The SVID thread touches on the reasons that AT&T failed: <https://www.tuhs.org/mailman3/hyperkitty/list/tuhs@tuhs.org/message/UFYAOAVWUNJ646WT5YWD6T2HHOKPZWEB/> Clem Cole was in the room when Bill Gates said “it’s all about volume” [ implying modest margins ] Others mention ‘lawyers’ came to dominate the firm, making poor decisions. Rob Pike has previously mentioned on-list a lawyer killed the music, in PAC format, they’d arranged to put on Plan 9 distribution CD. <https://www.tuhs.org/mailman3/hyperkitty/list/tuhs@tuhs.org/message/NNKHNKQP2ZE7DD7MYODZTRTDTIGDYFME/> Rachel Chalmers [4] in 1999 quotes Dennis Ritchie on the struggles to get the Version 6 code and Lions Commentary released. The lawyers and Corporate Culture were still hanging on 20 years later, wanting to block public release because they could, not for business reasons. In 1956, AT&T accounted for ~2% of US GDP, equivalent to ~$500B in sales today [1], comparable to Apple’s 2024 earnings of $391B. Tom Wolfe [2] wrote a piece on Noyce at Fairchild, then Intel, and details an “East Coast Management Style”, of opulence, self-indulgence and aggrandisement of ’the ruling class’. Wolfe is excoriating about “East Coast” management, though doesn’t detail the Silicon Valley / California approach in any detail. Noyce & Moore left Fairchild in 1968 with the invention of the Self-Aligned Silicon Gate for MOSFET to pursue devices made of large, regular arrays of transistors, also, quite particularly, to run the business without interference from the East Coast, able to reinvest their profits and grow the business as fast as they could. Fairchild had passed over Noyce in favour of Lester Hogan from Motorola [5] and his management team. Hogan was given a record renumeration package to move to Fairchild, but didn’t save the business. Intel has lasted, quickly growing to be a significant force and holding a lead. Fairchild never recovered after Noyce & Moore left and it sputtered out, underlining the importance of management style in Semiconductor businesses. Fairchild Semiconductor had, for over a decade, grown to be the dominant silicon device manufacturer, despite the parent company consistently using their profits to invest in vanity projects with low returns or losses. In 1963, Fairchild had announced UHF TV transistors for "$1.05” ( ~$11 now ) after the FAA added UHF band to broadcast TV. To compete with Vacuum tubes, given for nothing to TV manufacturers, Fairchild had to drop prices ten-fold ( vs mil.spec devices ) [ Valve manufactures made their money on replacement valves, not on original parts. ] Importantly, Fairchild’s price was below cost at the time. The engineers who pitched this to Noyce knew they’d sell millions and be able to quickly reduce production costs to make large profits. Noyce & Moore seemed to have used their Maths ability to solve a business / economics problem for them: How to optimise long-run revenue and profits in a dynamic, high CapEx / R&D / Non-Recurring Expenditure technology field? Their answer is what we’ve christened “Moore’s Law”: a) run R&D as hard as you can, shortening the ‘generation’ period to 2-3 years, well below the usual 5-10 years for “cost recovery” b) pass on 100% of cost savings to customers, relying on “Elasticity of Demand” to drive Volumes and increase total revenue. I assume they understood enough Game Theory to know once they adopted a “high volume / low-cost / rapid generations” strategy, they’d force all manufacturers in the industry to follow or be left behind. Kenneth Flamm [6], an economist, has written about the sustained “100% pass-through rate” of Silicon Semiconductors until at least 2010. I’ve not seen him or others discuss the origins of this strategy, unique to Semiconductors. Gordon Bell [3] noted the impact of the CMOS CPU revolution created by Silicon Valley. In 1984, there were 92 US Minicomputer Companies using discrete or Small Scale IC’s. In 1990, only 4 survived: Data General, HP, DEC and IBM [ Bell in [3] notes their fates out to 2002 ] IBM declared it’s first loss [$2.6B] in 1991, deepening to ~$5B and then ~$8B in 1992 & 1993 - seemingly without warning, they were caught by the changes in the market as commodity chips surpassed everything. IBM knew microprocessors were rapidly evolving, knew it’s Corporate Culture was antithetical to developing a product quickly & cheaply, so developed the IBM PC in 1981 in Boca Raton, Florida - away from Manhattan & breaking their traditional fully internal development and high gross margin model. During the 1970’s and 1980’s, IBM accounted for over 50% of Industry revenue - bigger than everyone else combined. That they weren’t able to adapt and respond to the challenges they clearly saw speaks of management failure. Like AT&T, they saw technology evolving and correctly forecast it’s impact on their ’traditional’ business lines, but were unable to changed their corporate culture to adapt. ================================================================================= References ================================================================================= [0] <https://en.wikipedia.org/wiki/Breakup_of_the_Bell_System> 1984: Southwestern Bell Corporation (known as SBC Communications after 1995) In 2005, SBC acquired former parent AT&T Corporation and took the AT&T name, becoming AT&T Inc. =================== [1] How Antitrust Enforcement Can Spur Innovation: Bell Labs and the 1956 Consent Decree <https://assets.aeaweb.org/asset-server/files/13359.pdf> As described in Section I, the Bell System was the dominant provider of telecommunications services in the United States. In terms of assets, AT&T was by far the largest private corporation in the world in 1956. AT&T, together with all companies in the Bell system, employed 746,000 people with a total revenue of $5.3 billion or 1.9% of the U.S. GDP at the time (Antitrust Subcommittee, 1959; Temin and Galambos, 1987). =================== [2] THE TINKERINGS OF ROBERT NOYCE How the sun rose on the Silicon Valley Tom Wolfe, Dec 1983 <http://classic.esquire.com/the-tinkerings-of-robert-noyce/> <https://web.stanford.edu/class/e145/2007_fall/materials/noyce.html> A certain instinct Noyce had about this new industry and the people who worked in it began to take on the outlines of a concept. Corporations in the East adopted a feudal approach to organization, without even being aware of it. There were kings and lords, and there were vassals, soldiers, yeomen, and serfs, with layers of protocol and perquisites, such as the car and driver, to symbolize superiority and establish the boundary lines. Back east the CEOs had offices with carved paneling, fake fireplaces, escritoires, bergères, leather-bound books, and dressing rooms, like a suite in a baronial manor house. Fairchild Semiconductor needed a strict operating structure, particularly in this period of rapid growth, but it did not need a social structure. In fact, nothing could be worse. Noyce realized how much he detested the eastern corporate system of class and status with its endless gradations, topped off by the CEOs and vice-presidents who conducted their daily lives as if they were a corporate court and aristocracy. He rejected the idea of a social hierarchy at Fairchild. =================== [3] Rise and Fall of Minicomputers Decline of the Classic Minicomputer (1985-1995) Gordon Bell 2013 <https://ethw.org/Rise_and_Fall_of_Minicomputers#Decline_of_the_Classic_Minicomputer_(1985-1995)> While the demise of classic minicomputers was clear by 1985, companies continued offering them until the early 1990s, when the firms went bankrupt or were acquired by more astute competitors. (See Table 1) Wang declared bankruptcy in 1992. Compaq bought DEC in 1998, and HP acquired Compaq in 2002. EMC turned Data General into a data storage business in 1999. Table 1 <https://ethw.org/File:Bell-MinicomputerTable.JPG> 92 U.S. minicomputer companies, 1968-1985 (created by the author with help from colleagues over many years) The following list includes all firms making general and special-purpose minicomputers for real-time processing, communications, business etc., sold to end users through OEMs, and bundled for process control and testing. It does not include scores of military, AT&T, European, and Japanese computers and processing systems developed for niche markets. 49 started up and retained autonomy or died 2 grew at significant rates and continued to grow: Data General (1969), Prime (1972) 8 grew at diminished or declining rates, or found small niches 39 ceased to manufacture 10 merged with larger companies 8 existing computer companies built minicomputers 2 grew rapidly: Digital Equipment Corporation (1960), IBM (1965) 25 existing companies built minicomputers for their own use 1 formed a division, Dymec, around an acquisition and grew rapidly: HP (1966) =================== [4] Code Critic John Lions wrote the first, and perhaps only, literary criticism of Unix, sparking one of open source's first legal battles. Rachel Chalmers November 30, 1999 <https://www.salon.com/test2/1999/11/30/lions_2/> "By the time the seventh edition system came out, the company had begun to worry more about the intellectual property issues and 'trade secrets' and so forth," Ritchie explains. "There was somewhat of a struggle between us in the research group who saw the benefit in having the system readily available, and the Unix Support Group ... Even though in the 1970s Unix was not a commercial proposition, USG and the lawyers were cautious. At any rate, we in research lost the argument.” ——— Ritchie never lost hope that the Lions books could see the light of day. He leaned on company after company. "This was, after all, 25-plus-year-old material, but when they would ask their lawyers, they would say that they couldn't see any harm at first glance, but there was a sort of 'but you never know ...' attitude, and they never got the courage to go ahead," he explains. Finally, at SCO, Ritchie hit paydirt. He already knew Mike Tilson, an SCO executive. With the help of his fellow Unix gurus Peter Salus and Berny Goodheart, Ritchie brought pressure to bear. "Mike himself drafted a 'grant of permission' letter," says Ritchie, "'to save the legal people from doing the work!'" Research, at last, had won. =================== [5] Wikipedia on Fairchild <https://en.wikipedia.org/wiki/Fairchild_Semiconductor> Sherman Fairchild hired Lester Hogan, who was the head of Motorola semiconductor division. Hogan proceeded to hire another hundred managers from Motorola to entirely displace the management of Fairchild. The loss of these iconic executives, coupled with Hogan's displacement of Fairchild managers demoralized Fairchild and prompted the entire exodus of employees to found new companies. =================== [6] Moore's Law and the Economics of Semiconductor Price Trends Flamm, 2018, NBER <https://www.nber.org/system/files/working_papers/w24553/w24553.pdf> ——— Flamm posits semiconductors [ DRAM & Microprocessors at least ] have maintained a 100% cost “pass-through rate” since the 1960’s. He’s been an expert witness for courts many times, as well as writing reports for the NBER and publishing academic papers. ——— In short, if the historic pattern of 2-3 year technology node introductions, combined with a long run trend of wafer processing costs increasing very slowly were to have continued indefinitely, a minimum floor of perhaps a 20 to 30 percent annual decline in quality-adjusted costs for manufacturing electronic circuits would be predicted, due solely to these “Moore’s Law” fabrication cost reductions. On average, over long periods, the denser, “shrink” version of the same chip design fabricated year earlier would be expected to cost 20 to 30 percent less to manufacture, purely because of the improved manufacturing technology. How would reductions in production cost translate into price declines? One very simple way to think about it would be in terms of a “pass-through rate,” defined as dP/dC (incremental change in price per incremental change in production cost). The pass-through rate for an industry-wide decline in marginal cost is equal to one in a perfectly competitive industry with constant returns to scale, but can exceed or fall short of 1 in imperfectly competitive industries. Assuming the perfectly competitive case as a benchmark for long-run pass-through in “relatively competitive” semiconductor product markets, this would then imply an expectation of 20-30% annual declines in price, due solely to Moore’s Law. To summarize these results, then, though there are substantial differences in the magnitude of declines across different time periods and data sources, all of the various types of price indexes constructed concur in showing substantially higher rates of decline in microprocessor price prior to 2004, a stop-and-start pattern after 2004, and a dramatically lower rate of decline since 2010. Taken at face value, this creates a new puzzle. Even if the rate of innovation had slowed in general for microprocessors, if the underlying innovation in semiconductor manufacturing technology has continued at the late 1990s pace (i.e., a new technology node every two years and roughly constant wafer processing costs in the long run), then manufacturing costs would continue to decline at a 30 percent annual rate, and the rates of decline in processor price that are being measured now fall well short of that mark. Either the rate of innovation in semiconductor manufacturing must also have declined, or the declining manufacturing costs are no longer being passed along to consumers to the same extent, or both. The semiconductor industry and engineering consensus seems to be that the pace of innovation derived from continuing feature-size scaling in semiconductor manufacturing has slowed markedly. ————————— [ submission to a court case ] Assessment of the Impact of Overcharges on Canadian Direct and Indirect Purchasers of DRAM and Products Containing DRAM Submitted to: The Honourable Ian Binnie 28 June 2013 <https://www.cfmlawyers.ca/wp-content/uploads/2012/05/DRAM-Ross-Distribution-Protocol-Report.pdf> VI. Pass-Through in this Case 48. Based upon my review of the information described above, I am of the view that there would likely be a very high degree of pass-through in DRAM distribution channels, all the way down to final consumers. This conclusion is based principally on a review of the structure of the various markets together with an application of standard economic theory. I also draw on pass-through analyses done by a number of experts in the U.S. action, recognizing that they are clearly contradictory on some key points. Evidence from the U.S. Case 49. Let me begin with a brief review of the work done by economists for various parties in the American action. a. In his report Michael Harris (July 10, 2007) estimated pass-through from top to bottom rather than just at one stage. That is, he looked to see if increases in the price of base DRAM were passed all the way down the distribution channels and increased computer prices. In one test he estimated that there was more than 100% pass-through of base DRAM price increased to final computer purchasers. In a second test he studied the effect of increases in spot market DRAM prices on aftermarket DRAM prices. Again, he found more than 100% pass-through.21 b. Professor Kenneth Flamm, in his report (December 15, 2010) provided estimates of the pass-through from retailers to final consumers using data from four major U.S. retailers (Best Buy, Office Depot, CDW, and Tech Depot) for various categories of computer products (desktops, notebooks, servers, memory, printers and graphics). Most pass-through estimates were in a range between 90% and 113%, the desktop and notebook rates were all within a few points above or below 100% with one exception. =================== [7] Links on California, population and economic growth from 1900 to 2000 ————————— Wikipedia <https://en.wikipedia.org/wiki/Economy_of_California> The economy of the State of California is the largest in the United States [ in 2024 ]. It is the largest sub-national economy in the world. If California was an independent nation, it would rank as the fourth largest economy in the world in nominal terms, behind Germany and ahead of Japan. As of 2024, California is home to the highest number of Fortune 500 companies of any U.S. state. As both the most populous US state and one of the most climatologically diverse states the economy of California is varied, with many sizable sectors. ————————— California since c. 1900 <https://www.britannica.com/place/California-state/California-since-c-1900> population in 1900: 2 M. now largest US state by population & GNP. ————————— California’s Economy. [ fact sheet 2025 ] <https://www.ppic.org/publication/californias-economy/> California is an economic powerhouse, nationally and globally. • In 2023, California’s gross domestic product (GDP) was about $3.9 trillion, comprising 14% of national GDP ($27.7 trillion). Texas and New York are the next largest state economies, at 9% and 8%, respectively. • California’s economy ranks fifth internationally, behind the US, China, Germany, and Japan. On a per capita basis, California’s GDP is greater than that of all of these countries. • California’s economy has grown relatively slowly in recent years, averaging 2.3% per year between 2020 and 2023, compared to 3.9% on average over the previous four years. By comparison, Florida (4.6%) and Texas (3.9%) grew faster than California since 2020. • Over the long term, California’s economy has grown faster than the nation overall (111% vs 75% over the past 25 years) and faster than other large states except for Texas (128%). On a per capita basis, California’s economic growth outpaces all other large states over the long term. Growth in jobs and businesses have powered the state’s economy. • California’s labor market grew by 4.2 million jobs (30%) between 1998 and the second quarter of 2024; over the same 25-year period, the number of businesses with employees grew more than 72%. Both outpaced population growth (18%), leading to robust gains in economic output. ————————— California’s Population [ fact sheet 2025 ] <https://www.ppic.org/publication/californias-population/> One in eight US residents lives in California. • With just over 39 million people (according to July 2024 estimates), California is the nation’s most populous state - its population is much larger than that of second-place Texas (31.3 million) and third-place Florida (23.4 million). ————————— =================== -- Steve Jenkin, IT Systems and Design 0412 786 915 (+61 412 786 915) PO Box 38, Kippax ACT 2615, AUSTRALIA mailto:sjenkin@canb.auug.org.au http://members.tip.net.au/~sjenkin