How America Won Back Semiconductors from Japan
our last video on japan's semiconductor industry ended with the industry at the very peak of its powers taking place over the span of 30 years the island country's rise to semiconductor supremacy shook the industrial foundations of the west the japanese semiconductor industry once seemed invincible but what goes up must come down in this video we look at japan's semiconductor decline but first the asionometry newsletter check out this one about neon shortages in the semiconductor industry it's worth a read considering today's current issues the sign up link is in the video description below i try to put one out every week maybe two alright back to the show throughout the 1980s japanese ic production gained fast ground on the americans this was especially the case in markets relating to a specific type of memory chip dram the japanese absolutely dominated three generations in the 1980s 64 kbit released in 1982 256 released in 1984 and finally one megabit released in 1988. japanese market share in these three items was over 70 percent and in the latter two classes it reached 90 percent this decade of dominance was not only because their products were better than the american competition they were significantly cheaper too this was because japanese companies with access to cheap capital through their business group ties invested in acquiring advanced automation tools and processes this resulted in much lower defect rates and higher yields furthermore japanese companies made a savvy technology bet early generations of dram technologies used n-channel mos or nmos until the 1980s complementary mos or cmos was an alternative technology that at the time was seen as more expensive and harder to pull off american companies stuck with nmos while the japanese used cmos this had been a fortunate choice one made mostly due to power concerns at the time as it turns out rapid advances in lithography technologies made the cost of cmos far lower than anyone could have ever anticipated and it became the industry standard the american chipmakers found themselves stuck with an outdated expensive technology it cost them their lead and the japanese soon search ahead then came 1985 a horrific down year for semiconductors a result of a mild recession in the computer market after growing five times over in the previous four years the microcomputer market saw an eight percent decline in 1985. this small decline however had massive reverberations down the supply chain prices collapsed 60 percent and an immense amount of capacity found itself without sales to fulfill intel a dram pioneer noted the biggest crash in orders in over 10 years they shockingly exited the dram industry entirely shutting down eight semiconductor factories and laying off 2 700 people this of course would be the best thing that ever happened to them but let's not get too far ahead of ourselves texas instruments saw revenue shrink 14 percent national semiconductor saw a 17 decline in revenue in addition to a stunning 117 million dollar loss by 1986 the american dram industry had consolidated from 14 producers in 1970 to just three the whole thing is just another indication of how brutal the dram memory cycle is on the way down the industry's decline alarmed everyone including the united states government the no good very bad year of 1985 led to the japanese receiving complaints for predatory and unfair trade practices the accusation had two parts to it one the japanese were dumping semiconductors into the u.s market at below fair market value and two the japanese were not providing foreign chip makers sufficient access to its domestic market at the start not everyone was particularly excited about making this political the state department and the white house were concerned about its effect on overall u.s japan relations a similar attempt in 1982 had been shot down by the u.s trade representative for this reason and the american semiconductor industry as a whole isn't monolithic the anti-dumping laws had always been there but the industry represented by the semiconductor industry association or sia never moved on using them before in fact only small micron would really see benefit from trade barriers actually being imposed computer maker end users like at t and ibm benefited from cheap memory prices and were loath to see them rise the imposition of trade barriers would make america an even more expensive place to make things accelerating the electronics assembly offshoring trend already underway also the american semiconductor industry was doing battle overseas a trade agreement would likely do nothing for japanese dumping into these third country markets in fact it might intensify them and finally the damage had already been done the american memory makers had already left the building and weren't coming back it also seemed unlikely that new companies would enter the market japanese penetration into the us market had actually declined in the two years prior to 1985.
what happened in 1985 was a demand problem not a competition one but eventually micron forced the issue in june 1985 the small memory maker filed an anti-dumping complaint against japanese exporters of 64k dram intel amd a national semiconductor soon followed doing the same for the eprom market ibm declined to stand in their way thus the sia finally threw its weight behind trade sanctions with the sia aligned they went to dc and found the political atmosphere much more amenable to their message the reagan administration even filed a complaint of their own covering 256k dram a very unusual move internally the administration was concerned about the growing trade deficit reagan was generally a free markets guy but he was also grappling with a surge in protectionist sentiment within congress there was a bill in congress at the time proposing sanctions against every country running a trade surplus against the united states the administration wanted to keep this from turning into law this semiconductor suit could help head off this surge in protectionist sentiment and kill the bill at the same time it would also show the public that the administration was strong on opening up foreign markets while protecting domestic trade from the japanese the americans saw to remove obstacles to market entry and grow american semiconductor share in the japanese market 20 to 30 percent was the target they wanted every japanese firm to stop dumping into the american market not only a few of them and they wanted a separate monitoring body to help enforce all of this things were changing in japan for a long time midi had guided the country's industries and shepherded its growth many outsiders saw midi's bureaucratic guidance as crucial to the country's economic success the puppeteer holding all the strings thus americans saw the organization as a channel through which to restrain the country's semiconductor practices and in fact the japanese had done something similar for the automobile industry a few years before starting in the spring of 1986 us and japanese negotiators discussed the matter eventually they announced an agreement the 1986 u.s japan semiconductor agreement the americans got everything they wished including the japanese agreeing to grow american market share in japan to slightly above 20 percent in five years it was seen as a big win despite some murmurs about the death of free trade or something like that but what the americans did not know was that the relationship between midi and the japanese industrial apparatus was changing it wasn't 1955 anymore japanese companies were experiencing massive export success around the world tiny companies like sony have grown up into billion dollar conglomerates they had more than enough capital to fund their efforts without power of the purse midi found itself left only with the instrument of consensus many bureaucrats could talk until the cows come home about implementing these new trade practices to head off sanctions but they had no real way to force the companies to fall in line many japanese companies felt that their government caved and betrayed them they had hoped for a simple fine but having to submit data to a third party as part of a big intricate price monitoring mechanism ticked them off to meet its anti-dumping guidelines midi imposed a voluntary export restraint asking its companies to meet a price rather than volume goal to meet the 20 market share target it could only ask japanese companies to source more from the united states as you might expect these efforts failed but to the west it seemed like midi didn't want to implement the agreement because midi was so powerful right so reagan found himself forced to impose a 100 tariff on 300 million dollars worth of japanese goods in april 1987 a move that made headlines at the time relations improved a few years later and a new more lenient agreement was struck in the 1990s combined with the 1985 plaza agreement which revalued the japanese yen the u.s japan semiconductor agreement lifted the pricing pressure in the memory market unlike with other industries the american chipmakers refused to cede the industry to the japanese instead the remaining american companies began adopting japanese characteristics namely a renewed emphasis on manufacturing they switched away from the outdated nmos technology and adopted cmos they refocused on total quality management techniques and ditched human labor it helped that their plants weren't unionized they also implemented automation into their production lines notably they started buying from the best japanese semiconductor equipment makers putting pressure on their legacy american competitors to improve intel in particular responded well new head of manufacturing craig barrett broadly adopted japanese practices he sent factory managers for extended stays at japanese chip factories and combed the japanese press for information about how their fabs worked barrett also first implemented the famous copy exactly mythology the idea is that the high volume production plant would fully replicate the pilot manufacturing plant it saw initial resistance but rapidly scaled output while maintaining good yields barrett would later serve as the company's ceo from 1998 to 2005.
by the early 1990s the american semiconductor industry had cut its defect rates by a factor of 10 utilization rates tripled from 20 percent in 1984. yields at american fabs improved dramatically from 60 in 1986 to 84 in 1991. over the same period of time the japanese had gone from 75 percent to 93 percent a gap still existed but it had shrank from 15 points to 9. the remaining gap was largely due to economies of scale american fabs were simply smaller and they lacked the size to match up with the japanese americans also clearly noticed what the japanese had done to gain manufacturing competency in the early 1980s they closely studied the vlsi project where the japanese government brought together a number of competing companies to share knowledge and advance the industry the americans wanted to adopt this model as well in 1984 they passed the national cooperative research act which modified anti-trust law to allow certain cooperative research initiatives between competing firms they also revise laws to allow partnerships between national research institutes and private companies based in the united states furthermore private companies can retain the exclusive rights of whatever they learned in this partnership for five years or longer this represented a significant break with the japanese approach where the government owned whatever came out of those partnerships one can definitely argue that publicly funded research should be available to all but i doubt such an arrangement would have gotten far off the ground with these the legal pathway was paved for a new type of consortium in 1987 14 american semiconductor companies and darpa joined together to found cemetec semitech would eventually find its place as a facilitator of relationships within the american semiconductor industry and a navigator for its future road map it has its criticisms but through cemetec the american semiconductor industry fostered more openness between its various players for the benefit of the industry as a whole it would help make the way for the american resurgence to come intel was improving their manufacturing processes and seeing results but the company quickly realized that they lacked the necessary economies of scale to compete in memory thus their 1985 exit from dram it had been shocking and painful but it also allowed them to focus their resources onto logic chips chips that take in and process information at the time a faster growing market to cement the strategic change of direction the company decided to single source its 386 microprocessor this was very unusual at the time but in doing so they were able to control the product's pricing buoyed by close relations with american pc makers like ibm intel soon came to dominate this high margin sector of the semiconductor industry everyone else followed suit thus throughout the late 1980s and early 1990s the american semiconductor industry transitioned from making memory products to making logic and integrated system products texas instruments for instance built its comeback on making digital signal processors and small startups seized upon the vlsi design revolution to create amazing new products like the graphics card profits rebounded to new highs japanese companies failed to see memories declining status within the broader computer ecosystem a critical error the reasons why are multifaceted the japanese semiconductor industry lacked the industry openness the americans had in my previous video i talked about the trouble the japanese government had in getting competing companies to work together that never went away and the japanese weren't open to trying new tools and processes even when they were literally in their own backyard for instance the founder of samco international invented a new thin film chemical vapor deposition tool in his backyard in kyoto not a single japanese company tried it until the americans first did additionally ties between japanese research institutions and industry were very weak note that it was a japanese scientist hiro kinoshita working at the japanese telecom ntt who did the foundational science work on what would eventually become euv lithography kinoshita first published his work in 1986 but it took a grant from the u.s department of energy and later the u.s chipmakers of euv llc in the 1990s to eventually get the ball rolling japan wouldn't start their euv efforts until 2002 many years thereafter and finally the japanese were focusing too hard on being the first to advance to the next generation of dram memory on one hand it makes sense such focus got them to where they were now why change but this tunnel vision blinded them to the possibility that dram could be replaced by something else it would also set the stage for japan to lose its grip on the dram industry altogether in 1991 the japanese real estate bubble burst and the company entered an economic malaise the real estate bubble had inflated due to the yen's appreciation a trend kicked off by the 1985 plaza accord it might be tempting to attribute the fall of japan's semiconductor dominance to political and macroeconomic happenings like the plaza accord and the bubble but to do so i think would be a drastic simplification this is because japanese chip makers continue to show healthy growth numbers through the late 1980s into the mid 1990s about 10 a year which is in line with the rest of the world market it makes sense a domestic slowdown wouldn't really affect export success or failure the biggest thing i think would be korea in the 1990s samsung started to pull ahead of the japanese chip makers in the dram industry the company had been investing in semiconductors since 1974 when it acquired a 50 stake in the struggling korea semiconductor company then in 1983 samsung received a technology transfer from micron for a 64k dram unit they sent eight korean engineers to train at micron and then lured 20 more korean immigrants back from the united states for additional talents the koreans followed the same strategies pioneered by the japanese in 1986 the korean government organized the national consortium of their own with samsung working with lg and hyundai to research and develop 4 megabit dram this succeeded and the korean product hit the market just 6 months after the japanese korean market share rose from 4 in 1986 to 17 by 1990 this was aided by favorable exchange rate trends for the korean wan samsung took another page out of the japanese book when they decided to work on multiple generations of product simultaneously easier said than done but they pulled it off in 1991 and 1992 samsung released the 16m dram at the same time as nec and fujitsu then it was just samsung and hitachi who first brought 64 megabit to the market with the former just barely ahead and the korean giant had 256 megabit and one gigabit already in development the once invincible japanese were now finding themselves losing share to this new insurgent but as they worked to keep up at the leading edge they found themselves suddenly attacked from an unexpected direction when it was least expected the americans came roaring back micron technology had been one of the few surviving dram producers to emerge from the 1980s the expectation had been that micron was relegated to the backwaters of the industry but micron survival allowed it to refine their manufacturing techniques in doing so they succeeded in drastically reducing the number of masks needed in dram production as much as 50 at the same time the company reduced the size of its chips and implemented new techniques like chemical mechanical polishing or cmp to improve the yield as a result costs of production improved enormously so 10 years after the no good very bad year of 1985 micron decided to give the japanese a taste of their own medicine they began selling older dram chips maybe a half generation behind the leading edge at a cost of four dollars while the japanese were still selling them at six microns soon took over the low end of the dram industry and triggered a massive price collapse across the market as a whole the japanese press referred to this as the micron shock the micron shock set into motion the fall of the japanese semiconductor manufacturing industry chip makers found themselves caught in a pincer movement between micron at the low end and samsung at the high end 128 megabit dram prices took a free fall collapsing 50 percent in a single year the japanese dram makers had not consolidated in many years but when they finally did it came very quickly fujitsu exited the dram market in 1998.
hitachi and nec merged their operations together to create el pida in 1999. they later took over mitsubishi's division in 2002 this company was later acquired by micron technology toshiba pulled out of the dram business in 2001 they pivoted over to their flash memory business the company was later spun out into its own entity and renamed themselves to kiouxia there still remains a japanese semiconductor industry their semiconductor equipment industry is still pretty strong tokyo electron cannon and nikon are big players though none of them can be said to be leading edge and there are some significant japanese semiconductor manufacturers operating in their own niches sony for instance is one of the larger producers of image sensor chips the aforementioned kioxia has significant share in the more advanced nand flash memory industry and renesas produce specialized chips for the automakers the leading edge japanese semiconductor manufacturing has left the island and the country is now an afterthought in the semiconductor world south korea is the dominant memory maker with samsung reigning supreme and the united states is the leading logic chipmaker with the 200 billion dollar intel in the number one slot the america of the 1980s saw japan as invincible back in 1986 the president of micron technology remarked to the new york times we've got a problem in this country we have this idea that as soon as the japanese target an industry you are automatically dumber than a post to be in it i think the american semiconductor industry responded to the japanese challenge largely in the right way yes they lobbied washington and applied international trade pressure onto japan but all that can ever do was to buy time they used that time wisely they shored up their manufacturing weaknesses by copying their challengers they built institutions that sought to build on long-term trends and they created a new richer market that they capitalized upon in the end they became more japanese before the japanese can become more american the effort paid off in spades and it has left the japanese wondering what could have been had things turned out just a little bit differently alright that's it for tonight thanks for watching subscribe to the channel sign up for the newsletter and i'll see you guys next time
2022-06-27 15:07
