深圳市三联盛科技股份有限公司-彩客平台

中文版  english

semiconductor people must see the development and change of ic industry in the past 70 years

2016-11-15

this paper mainly introduces the development track of semiconductor industry in the past 30 years. through this introduction, readers can have a more comprehensive understanding of the semiconductor industry, establish a basic understanding of the semiconductor industry, so as to effectively predict the future development trend by analyzing the current situation of the semiconductor industry
                       .
overview of global semiconductor industry
 
crazy 2015
 
in the first half of 2015, the three major mergers and acquisitions in the international semiconductor market have established that 2015 can be recorded in the history of semiconductors. nxp merged freescale with us $11 billion in march, and avago acquired broadcom (broadcom) with a record $37 billion in late may, which is the highest acquisition case in the semiconductor industry so far, just four days later, intel announced that it would spend $15 billion to buy altera, a big fpga factory. during this period, hundreds of millions of dollars of mergers and acquisitions were also emerging, for example, infineon (infineon) purchased the us analog chip factory ir for us $3 billion. after several rounds of wild price hikes, china's wu yuefeng capital finally beat cypress to win the isi (core semiconductor) with annual revenue of less than 100 million us dollars with nearly 800 million us dollars
 
since this year, mtk, taiwan's leading ic design company, has acquired four design companies at a whirlwind speed. among them, it has purchased taiwan's largest analog chip manufacturer li qi with nearly nt $30 billion, which shocked the industry. at the end of august, riyueyue, the world's largest ic packaging and testing company, announced that it would buy 25% of the world's third largest silicon manufacturer with nt $35 billion, in addition, atmel abandoned china electronics cec's offer and reached a $4.6 billion acquisition agreement with german dialog
 
in early october, before this article was completed, dell announced the acquisition of emc for us $67 billion, which was the largest acquisition in the history of it industry and ranked sixth among the top 10 global acquisitions. before that, vodafone (uk) finally acquired mannesmann, a century old german enterprise, with nearly $200 billion, and the merger of aol and time warner for more than 180 billion us dollars, among them, lucent's us $23 billion acquisition of ascent and hp's $18 billion acquisition of compaq are the top two. however, avago's us $37 billion acquisition of broadcom has broken the record of the entire it and semiconductor industries, the just announced acquisition of emc by dell with 67 billion yuan is a record of it industry
 
   from the global m & a history listed above, the previous mega mergers and acquisitions were basically monopolized by telecommunications, media, pharmaceutical and banking industries. but now it seems that the it technology industry and even the pure semiconductor industry have begun to occupy an increasingly important position in the world business history. the merger of dell and emc belongs to the category of it industry. this paper focuses on the semiconductor industry, therefore, in 2015, the value of m & a transactions generated by the semiconductor industry will be close to 120 billion us dollars for the first time
 
let's take a look at the history of the semiconductor industry
 
   the age of transistor and the birth of ic in 1940s and 1950s
 
   during world war ii, the new technology competition led by the military industry reached an unprecedented level, and because of the needs of the war, scientific and technological inventions around the world broke out, among which electronic technology was regarded as the key development frontier technology by the united states
 
the world's first military electronic computer was born in the united states in 1942. it is composed of numerous electronic tubes, resistors, capacitors and hundreds of thousands of wires. it is larger than an ordinary house and weighs 30 tons
 
in 1947, shockley, known as the "father of transistors", and his two colleagues made the first transistor and won the nobel prize in physics
 
in 1951, ibm delivered the first commercial calculator for the u.s. census. ibm released the first calculator with stored programs in 52, which is what we call the computer
 
in 1958, j. kilby of ti (texas instruments) designed an ic (integrated circuit) based on germanium. the news came that fairchild, composed of eight talented disciples who betrayed shockley, immediately developed silicon-based ic. the birth of silicon ic made it possible for ic to be produced on a large scale, in 1969, the court formally ruled that j. kilby of ti and r. noyce of fairchild were inventors of ic in the legal sense
 
     for the semiconductor industry, this stage belongs to the invention stage
 
the era of ic integrated circuits in the 1960s
 
planar technology developed by fairchild can easily fabricate semiconductor modules on the same side of a silicon chip by means of oxidation, yellow light lithography, etching, metal evaporation and other techniques. in 1960, epitaxy technology was also developed by bell laboratories. at this point, the semiconductor industry has obtained the ability of batch production, and finally stands firm and begins to grow rapidly.
 
at the end of 1960s, fairchild made ram (random memory). noyce, which left fairchild and intel and ibm founded by moore, put forward commercial and optimized schemes one after another

  in 1964, moore, who was still in fairchild company, predicted that the number of transistors and resistors integrated on semiconductor chips would double every year. in 1975, moore formally submitted a paper in ieee. according to the actual situation at that time, moore's law was amended by changing "doubling every year" to "doubling every two years", this is the most famous moore's law in semiconductor industry

lsic lsic large scale integrated circuit, the appearance of personal computer pc in 1970s
 
in 1971, the world's first microprocessor 4004 (4-bit) was born in intel company. the 4004 cpu has 2300 transistors. today, the core i7 cpu of the same intel company has a maximum of 2.27 billion transistors. after 40 years of development, the performance difference is 1 million times
 
  from the late 1960s to the 1970s, there was a big explosion of semiconductor manufacturing technology. just because of the formation of silicon valley, many semiconductor companies gathered here to set up and launched one technology competition after another. with the development of technology, more and more complex processes have been applied to semiconductor manufacturing, and the cost has been falling, in this way, lsic (large scale integrated circuit) should also be born. with the emergence of lsic, it also laid the foundation for semiconductor from commercial to civil
 
in 1976, jobs founded apple company and designed the first civilian computer, which achieved great success. from then on, the popularity of computer officially entered the civil era
 
    for the semiconductor industry, this stage belongs to the quasi civil stage
 
popularization of pc in 1980s
 
in the late 1970s, although jobs' apple ii achieved success and pushed the computer into the civil field for the first time, it was still a patent of the rich class. until ibm company introduced the first personal computer desktop computer in 1981, the pc that entered the civil field was already 16 bit products
 
in 1984, ibm launched a more optimized pc and adopted the strategy of opening technology. so far, pc began to be popular all over the world. compared with ibm's success, apple, which launched personal calculators earlier, was abandoned by the market because of its closed strategy, and founder steve jobs was expelled from the apple board of directors
 
in the 30 years since the advent of pc, the semiconductor market has basically developed around pc. the two most important components are semiconductor memory and micro processor. the rapid development of memory and microprocessor has brought about the prosperity of pc, what's more, the electronic technology that no longer affects human beings all the time
 
in the 1980s, japan's semiconductor manufacturers adopted the dram based idm business model, which made them in a leading position in the global semiconductor market. in its heyday, it even occupied half of the global semiconductor market, eclipsing the united states, the leader of the semiconductor industry
 
we can clearly see from the table above that 11 of the world's top 20 semiconductor manufacturers were japanese manufacturers in 1988, while the united states accounted for only 5. among them, the world's top three are firmly occupied by japanese manufacturers for a long time
 
    in the 1980s, japan, with its strong economic strength, devoted all its efforts to develop the electronic industry, and created many well-known electronic brands. in the field of pc and consumer electronics (especially consumer electronics), these idm manufacturers hope to manufacture all electronic components themselves as much as possible. they also pursue the perfection of manufacturing, so it also created the global dominance of japanese semiconductor components
 
in 1988, samsung of south korea entered the world's top 20 semiconductors for the first time
 
tsmc was founded in 1987. it initiated foundry mode, which promoted the prosperity of fabless, which occupied nearly one third of semiconductor industry in the future
 

  the maturity of pc and the birth of internet in 1990s

   in the 1990s, the semiconductor industry still follows moore's law. pc is more and more widely used and its functions are more and more powerful. at this time, software plays a decisive role. microsoft's window operating system is a great success and has established its dominant position in pc software, after 30 years, the alliance of wintel (microsoft and intel) still dominates the pc industry
in 1992, intel achieved great success because of its alliance with microsoft. in that year, intel beat nec with revenue of 5 billion dollars and became the leader of global semiconductor industry. since then, intel has been the first semiconductor company in the industry with annual revenue of 10 billion us dollars
 
in 1993, samsung made a great leap forward in ranking, ranking the 7th among the world's top 10 semiconductor manufacturers for the first time, and began to challenge the position of japanese semiconductor companies in dram
 
   in 1994, the global total sales of semiconductors exceeded the $100 billion mark for the first time
 
in 1998, the long-term weak european semiconductor factories were reorganized. sgs thomsone was renamed stmicroelectronics. in 1999, siemes (siemens) spun off its semiconductor business and established infineon
 
tsmc, which was established only 10 years ago in 1998, has entered the top 15 of global semiconductor enterprises
 
   in 1999, the former overlord hitachi, nec and mitsubishi electric separated memory dram departments to jointly form elpida, which is also a powerful anti japanese samsung
 
internet bubble and mobile communications in 2000~2010, pc era
 
bamboo shoots after a spring rain in the late 90s, the internet changed the way people knew about technology. internet companies sprung up like mushrooms and crazy chase of capital markets. finally, at the time of the millennium, the bubble burst and everything returned to calm. the growth rate of the whole semiconductor in 2000 reached unbelievable growth of over 30%. but with the collapse of the bubble, a cliff shaped fall occurred in 2001, with an astonishing 30% decline, not even in the next few years, until the global semiconductor market began to resume growth in 2004.
 
   network communication ic has entered thousands of households with the popularity of the internet. data is transmitted from wired transmission to optical fiber transmission and then to ultra-high speed wireless transmission. the sky of our home or city is full of endless data moving. all this makes new applications appear in the semiconductor industry which has been developing around the pc for decades, although pc is still a big brother, but the movement of plates is already happening quietly
 
in 2007, with apple's iphone and google's open android mobile phone system, smart mobile devices began to occupy every corner of the world. after the introduction of various tablet computers, mobile communication devices appeared blowout
 
   the emergence of intelligent mobile devices also marks the extreme development of the semiconductor industry, with extremely high efficiency, extremely small size chip and extremely fast transmission speed wireless communication technology, but all these are far from the end or only the beginning
 
in 2001, tsmc, after only 10 years' efforts, confirmed the success of foundry model and entered the global top 10 for the first time
 
hyundai, which acquired lg semiconductor in 2001, was officially separated from hyundai group and renamed hynix. hynix has become a strong competitor in dram field
 
mitsubishi and hitachi announced the separation of their semiconductor business to form renesas
 
in 2003, with a revenue of $2.4 billion, qualcomm became the first pure ic design company to enter the global top 20 semiconductor industry, and began to play an increasingly important role in the semiconductor industry
 
in 2003, motorola, the first generation semiconductor overlord, spun off its semiconductor business and established freescale, which was 100% acquired by blackstone group in 2006
 
in 2005, spansion was completely separated from amd and amd gave up the flash memory market
 
infineon (infineon) separated its memory business in 2006 and established qimonda. the company had no resistance to samsung's dram market and declared bankruptcy in 2009
 
in 2006, lsi acquired agere (geer system) separated from lucent, and lsi continued to exist
 
in 2006, european semiconductor giant philips divested semiconductor business and established nxp (nxp)
 
amd, which has been competing with intel for a long time in 2006, announced that it would give up all invalid business, only keep cup and logic business, start from a new start, and acquire ati to start the transformation road
 
   in 2008, the revenue of tsmc exceeded the 10 billion us dollar mark for the first time, and squeezed into the world's third largest semiconductor enterprise. it formed a tripartite situation with intel and samsung, and gradually opened the gap with other semiconductor enterprises, forming the three giants
 
   the global semiconductor output value exceeded 200 billion us dollars for the first time in 2000, and exceeded 300 billion us dollars in 2010
 
facing the post pc era represented by mobile communication, many former semiconductor overlords have fallen or reorganized in the past 10 years, which makes people feel sorry
 
2010 ~ the arrival of big data and the formation of internet of things (iot)
 

  with the rise of mobile communication, the huge data received or sent by wired or wireless intelligent terminals need to rely on larger scale storage devices. more and more enterprises also need to establish more powerful databases. whether it is traditional banks, telecom operators or emerging network sales, or even physical retail or individuals, they need more and more access capacity, with ubiquitous data access, the world really needs more advanced storage devices

   advanced means miniaturization and strong efficiency. we can't meet the demand by building endless large computer rooms, but we can give it to semiconductors who are good at it. semiconductor enterprises based on memory have made a lot of money in the years of big data explosion
 
   chips with more and more efficient performance and high-speed wireless data transmission have given birth to mobile personal intelligent terminals such as smart phones and tablet computers. subsequently, tv will enter our life again as a new role of home intelligent terminal. it will meet all the needs of our family life. the tv products of apple and google will be on the market soon, what's more, the original cold car will become the ultimate product of personal mobile terminal. it will become the intelligent mobile fortress for us to communicate with the world, rather than just a means of transportation. apple and google have made great efforts to develop smart cars
 
let's observe the changes in the semiconductor industry through the evolution of global top 20 semiconductors in recent years
 
in 2011, ti acquired ns for us $6.5 billion
 
   in 2011, google acquired motorola mobile business unit. following the separation of semiconductor business in 2003, motorola, a generation of overlord, disintegrated and completely withdrew from the mobile phone market, leaving only enterprise solution companies. after absorbing a large number of patents it wanted in 2014, google sold some useless patents and trademarks to lenovo
 
in 2012, mtk acquired morningstar, its largest rival, for nt $115 billion and built it into a global leader in tv chips
 
in 2012, micron acquired and bankrupted elpida, a memory company composed of nec, the former dram king, hitachi. after being transfused by the japanese government for hundreds of billions of yen in 2009, it also failed and finally declared bankruptcy
 
in 2012, asml invited intel, tsmc and samsung to share 15%, 5% and 3% shares of asml respectively, and provided additional funds for asml's research and development of ultra ultraviolet (euv) micro imaging technology in the new area, so as to accelerate the development of chip manufacturing process below 10 nm
 
   in 2013, ziguang group acquired china's second and third ic design companies, spreadtrum and redico with us $1.8 billion and us $900 million respectively
 
avago acquired lsi, a memory chip manufacturer, for us $6.6 billion in 2013
 
in 2014, cypress and spansion announced a merger to maintain their leading position in norflash and sram
 
in 2014, global foundries acquired ibm semiconductor business. this acquisition obtained ibm's semiconductor patent, technology and two wafer plants. in addition, ibm also obtained 1.5 billion us dollars. gf was asked to take over these assets
 
   in 2014, china's 100 billion integrated circuit industry investment fund was officially established, and intel invested us $1.5 billion in ziguang 20%
 
   as of october 2015, the situation described at the beginning of this paper has appeared in the semiconductor industry. there has been an unprecedented wave of mergers and acquisitions or restructuring in the semiconductor industry, which may also mark the beginning of a new era
 
   from 2015 to now, nxp and freescale have merged, adi acquired linear, avago acquired broadcom, intel acquired altera, softbank acquired arm, renesas acquired intersil, until now, it is widely rumored that qualcomm acquired nxp. the semiconductor industry is accelerating integration.
 

  looking back on history

   the development history of semiconductor industry is basically the development history of the whole electronic information industry. although semiconductor does not face consumers with terminal products or brands, it is also because of the advancement of semiconductor technology that our science and technology can keep developing
 
bell lab, founded by at & t in the 1920s, has successfully developed transistors, solar cells, lasers and communication satellites in history. eleven nobel laureates have emerged, which can be said to be the greatest company in the global science and technology industry, and has led the electronic information industry for more than half a century
 
in 1996, bell lab and at & t equipment department established lucent (lucent technology) in the face of the communication technology in the new area, until the advent of the network era and a series of strategic errors
 
alcatel acquired lucent with $13.4 billion in 2006, and bell lab seemed to be dead in name. in april this year, nokia, which got rid of mobile phones and returned to china, bought alcatel lucent for $16.6 billion. alcatel and lucent officially entered history
 
motorola, a company comparable to bell lab, invented the radio in the 1930s, successfully realized mobile communication in the 1940s, and finally invented the first mobile phone. in 1969, apollo 11 used motorola radio to transmit the first sentence from the moon
 
before japanese manufacturers dominated semiconductors in the 1980s, motorola had been dominating the semiconductor industry. in the 1990s, moto, which was heavily damaged by nokia in the global mobile phone market, finally disbanded its semiconductor business with a price of 200 million dollars in 2003 to establish freescale. in 2006, blackstone group acquired freescale for $17.6 billion
 
however, motorola, the parent company, sold its mobile phone business to google for $12 billion in 2011. so far, motorola, the overlord of the generation, officially withdrew from the mass consumer market, leaving only the enterprise solutions company. in 2014, after absorbing a large number of patents it wanted, google sold the remaining useless patents and trademarks of moto mobile phone business to lenovo
 
early this year, nxp announced that it would buy freescale for $11 billion. freescale, which has a trace of moto's blood, will be eliminated. the great moltrola was destroyed
 
fairchild is different from the above two it giants. it is a pure semiconductor company and i think the most legendary semiconductor enterprise so far
 
fairchild was born in 1957 by eight talented students with the greatest invention of the century, shockley called them 8 traitors. in 59, they designed silicon-based ic (integrated circuit) based on transistors, sharing the honor of ic inventors with kirby (based on germanium) of ti
 
the birth of silicon-based ic makes it possible to manufacture a large number of ic's. later, xiantong company developed planar technology, which can easily manufacture semiconductor components on silicon wafer by means of oxidation, yellow light micro shadow, etching, metal evaporation and other techniques. the semiconductor industry has embarked on a fast changing road
 
   these achievements are mainly attributed to n. noyce, the leader of fairchild company, and r. moore, another legendary figure of traitor 8, is the famous proposer of moore's law
 
the great success of xiantong fell into crisis in the late 1960s. because of the discord between the eight founders and the major shareholders, these talents left one after another to start their own businesses. in 67, the general manager of xiantong company, spock, came to ns (national semiconductor) and made ns the sixth largest semiconductor enterprise in the world, in august, 68, two of the most famous traitors, noyce and moore, founded intel, the dominant semiconductor company in the semiconductor industry. in 69, jerry sanders and seven employees founded another semiconductor giant amd company
 
the legend of silicon valley is created by xiantong. applied materials, which is strongly supported by noyce, is now the world's largest semiconductor equipment company. when jobs founded apple, the first investor was fairchild. an executive who left fairchild founded sequoia capital, which became a global model of technology and capital cooperation, cisco, cisco and linkedin are from him
 
there are more than 2000 companies supported by fairchild and its eight co founders, including yahoo, google, youtube, whatsapp and pixar (pixar animation), of which 92 are listed companies with a total market value of $21 trillion
 
at present, fairchild focuses on power supply analog ic, and the market value of nasdaq is only $2 billion. now on and infineon are both interested in acquiring fairchild. maybe in the wave of mergers and acquisitions in 2015, xiantong will go into history forever
 
as apple's steve jobs said: fairchild semiconductor company is like a mature dandelion. when you blow it, the seeds of this entrepreneurial spirit will float around in the wind
 
he has made a contribution to the whole world.
 
bell labs, motorola and fairchild are the greatest technology enterprises in this century. they all bid farewell to the world in different ways in 2015. the rest may be the logo trademarks that once swept the world, leaving a sigh
 
current situation of semiconductor industry
 
  from the current semiconductor top 20, the semiconductor giant of that time is now only ti (deyi), which still holds the leading position of global analog ic, and its younger brother intel still dominates the world's no.1 position. they are the current old american brands
 

  at present, the top three semiconductor companies (intel, samsung, tsmc) have been widening the gap with other semiconductor companies by virtue of their original technical advantages and capital investment of about 10 billion us dollars per year. they are competing for the most basic manufacturing technology

intel entered 45nm, 32nm and 22nm process nodes in 2007, 2009 and 2011, respectively. according to their standard moore's law, it should enter 14nm in 2013. however, it was delayed by one year until 2014. samsung also started mass production a few months later, while tsmc began to ship 16nm about half a year later, this is the process node with the smallest time gap in mass production
 
after intel's first fall, intel tragically announced in september that 10nm would not be successfully mass produced until 2017, and moore's law was skipping tickets. in order to compete for apple's orders, tsmc and samsung have made every effort to tackle the 10nm process. now it seems very likely that intel, tsmc and samsung will reach the 10nm process at the same time
 
it seems that the fierce fighting among the top three will widen the gap with other pursuers. however, more and more people choose to be spectators, and even look like they have nothing to do with themselves. it turns out that these things are not things because they are fabless. it seems that foundry has greatly changed the pattern of the semiconductor industry, because it was unimaginable before, now they just care about how to design chips better, not how to make them
 
traditional analog ic factories are also lack of interest in the most advanced process. they don't need to enter the advanced process like memory companies. fablite is basically their consensus. they have their niche market. the rise of iot makes them like fish in water because they don't need advanced process. ti, st, infineon and nxp have their own niche markets, however, in order to occupy the favorable position of iot, the new super application market, they have formed alliances or ganged up, resulting in many mergers and acquisitions since this year
 
no matter sanbi95ba is the leader in the field of semiconductor industry for several years, it is hard to challenge the position of samsung and nadisk in the field of memory
 
since the acquisition of elpida by micron, the cooperation with intel has been accelerated, and hynix has also built the world's largest semiconductor factory this year. after decades of separation, separation and integration, the memory industry has finally become the top four, with clear barriers. after these years of good times, we have greatly increased our strength, and we will no longer have more losses than we did before. how samsung will defend its position remains to be seen
 
the battle of the three strong
 
intel has been the world's semiconductor leader for 23 years. up to now, there is no possibility of the throne changing in five years. the main reason is that the position with the old 2 samsung is not close to the level of flesh and blood
 
however, what will be the trend in the future? we can see that intel's growth is slowing down, it still continues to dominate the pc cpu market, and after being firmly blocked by arm from mobile communication, we can see intel's struggle
 
tsmc claims that they don't have to catch up with intel in 7nm nodes, but can surpass intel in 10nm in 16 years. if this is the case, intel's old cpu may be broken by new rivals arm or amd by tsmc technology. in my opinion, intel, the overlord, will be eliminated in 5 years
 
samsung semiconductor entered the top 10 semiconductor companies in the world in 1993. it swept through all the memory semiconductor factories in europe, america and japan, and achieved dram's dominant position today. all the big memory factories suffered a lot because of samsung's existence
 
samsung's magic power is to constantly promote new technology and let others lose money. when others are still using the previous generation of technology, samsung will enter the new technology, so that its competitors can only keep losing money, while samsung can still maintain a small profit. the most cruel thing is that it only needs a small profit, rather than how much to make, because it is more concerned about making competitors lose money, money to earn together is not allowed for samsung, annihilation of 1000 enemy, 800 self injury, samsung has gone through the place of grass
 
the advent of mobile communication and big data has created a large demand for access. with the reduction of competitors, samsung has ushered in a big harvest. the trend of big data has been set. infinite data transmission and access will not change in the next few years. samsung has all the conditions to challenge intel
 

but does samsung really have no shortcomings? i think it will be samsung's problem to be too big. it seems that it is a good idea to have the cutting-edge manufacturing capacity of all products, which makes it look very powerful, but it is not invincible

if one product has problems, it needs the profit of another product to make up for it. it is difficult for us to take into account all the products and face the challenges of all the top companies in the world. if one or two products fail, it will cost twice or even three times to recover. what if the problem is the main project, for example, terminal products?
 
if there is a problem in the sales of mobile phones or consumer electronics at the end, what will samsung face? we have given us the answer 10 years ago, such as sony
 
mr. zhang zhongmou founded tsmc and launched foundry, a brand-new semiconductor business model. at that time, all semiconductor enterprises had their own wafer factories. what a fantastic idea! who needed foundry wafers?
 
zhang zhongmou may have known everything for a long time, but in fact, simply think, this information it industry will be very, very huge in the future. it can't only be played by some of your giants. when the industry matures, there must be large and small participants, and the oem may have a promising day. but if tsmc can be so brilliant today, i don't know if mr. zhang has predicted it?
 
the foundry model has given birth to the ic design industry, so that companies with the ability to design chips no longer need to build a fab. according to moore's law, the future fab will be a huge investment
 
with foundry, semiconductor companies without factories have sprung up like mushrooms. these fablesses companies have also provided endless orders for tsmc. even the original idm factories have begun to follow the route of fablite. they will no longer spend huge funds to develop new technologies or build new generation factories that will never be completed, a factory with billions of dollars has to be built in a few years, and it is becoming more and more expensive. now a new factory is more than 10 billion dollars. maybe sony is killed in this way
 
mobile communication and iot in the future will give more opportunities to many big and beautiful or small and fine fables. in addition, idm factories will rely more and more on foundry's advanced technology. after their backward factories are eliminated, they may become a problem. this will be the foundation for tsmc to continue to be strong, and its trend is good
 
however, what's the problem with tsmc? in the face of the determined beauty, foundry will be the place for competent people to compete. it's not so easy for tsmc to eat alone again. intel is highly interested, but his cpu has misgivings about fire in the backyard, so it's hard to make a move now. samsung dominates dram and nand, and everything must be in the development of new battlefields, because the battle front is too long, samsung needs a longer front (product line) to create space. foundry is basically a battlefield it can enter directly. especially this year's battle between apple processor tsmc and samsung can be described as brilliant and winding
 
 if the challenger can seize the first opportunity for two generations in a row, the world's semiconductor leader will be able to change ownership, and the defenders will not have to worry too much. if the first opportunity is lost once in a while, it may still be saved, but if it is lost continuously, it will have to pay a price
 
as long as moore's law does not die, we will still be able to overcome the difficulties we can see at present, and they will change with each passing day