If the chip production in Taiwan stops completely, the industry will need at least 5 years to recover

The meeting of top political leaders of China and the US was designed to ease tensions in the Taiwan issue.
A significant portion of semiconductor components are manufactured in Taiwan, and stability in the region is extremely important for the entire global economy.
Some experts believe that in the case of a complete shutdown of Taiwan enterprises, the semiconductor industry will be able to recover only in five years.Image source: TSМSimilar estimates in his interview with Barron's shares the author of the book \"War for chips: the battle for the most important technology in the world\" Chris Miller - Tufts University professor from the United States.
In his opinion, the U.S.
authorities underestimate the possible consequences of a regional conflict that could put Taiwan`s enterprises out of business.
Not only do the island produces up to 90% of advanced semiconductor components, but also the local production volume of chips is so large that no other region in the world can compete with it so far.
Up to a third of the annual increase in computing power in the world is provided by Taiwanese enterprises, according to expert estimates.
In the professor's opinion, the semiconductor crisis of 2020 and 2021, which paralyzed the work of the global automobile industry, is only a pale shadow of the likely problems that an armed conflict involving Taiwan would cause.
Even the PRC's \"nonviolent\" methods of influencing the island's economy can have a significant impact on the determination of Taiwan's leadership to remain independent.By the way, Chris Miller highlights the role of qualified personnel in the formation of the modern semiconductor industry.
The same founder of TSMC, Morris Chang, according to the author of the book, is the man to whom the whole world should be grateful for the technology we now have at our disposal.
For the United States, however, it is important to maintain a sensible immigration policy, since many of the leaders of major technology corporations are migrant workers or their direct descendants, and their role in shaping U.S.
industry and the economy is hard to dispute.

MediaTek will reduce dependence on TSMC by outsourcing chip production to Intel and GlobalFoundries

Reducing dependence on TSMC and its Taiwanese facilities will be an important trend for MediaTek in coming years, as CEO of this mobile processor developer, Rick Tsai, made clear.
The company's CEO explained that the main factor behind this decision will be geographical diversification, as over-reliance on Taiwan concerns not only the company, but also its customers.
Large consumers of semiconductor products, according to Mr.
Tsai, will begin to require chip suppliers to use multiple sources of production: Taiwan will be combined with the U.S.
or Europe.
MediaTek is preparing for the emergence of such requirements and is gradually expanding the geography of its chips production, not just limited to Taiwan.As over time, MediaTek expects to increase the volume of sales of its chips in the U.S.
market, then the localization of production, it is ready to pay close attention.
When TSMC's chip contract manufacturing facility in Arizona will start operating, MediaTek hopes to become its customer.
In addition, the production of processors of this brand will be engaged in GlobalFoundries, which has enterprises in New York State and Singapore.
However, in this case we are talking about the use of not the most advanced lithography, but it is more important to diversify sources.Intel, which has repeatedly reported on cooperation with MediaTek, from the second half of 2024, as explained by Rick Tsai, will provide it with chips, which will be produced by Intel 16 technology in the enterprise in Ireland.
These components will find use in TVs and Wi-Fi routers, and do not require advanced lithography to produce them either.
This market segment is quite large, so the company considers cooperation with Intel in all seriousness.
The head of MediaTek personally observes progress in this area on an almost monthly basis.

Chinese semiconductor industry accused the US of discrimination due to new restrictive measures

Chinese authorities have already formally protested the adoption of a package of laws by their US counterparts, which provides companies with subsidies to build facilities in the US in exchange for, among other things, a promise not to develop production in China for ten years.
In any case, vice chairman of the China Semiconductor Industry Association Yu Xiekang said without going into details at an event in Nanjing that parts of the US package of laws \"clearly discriminate against China\" and that the US initiative generally aims to support China's competitors in the foreign policy arena.
According to the industry association, some clauses of the US government's subsidy program for the national semiconductor industry contradict the fair trade principles supported by the WTO.According to Yu Sekan, the Chinese authorities should continue to support the national semiconductor industry in order to eliminate bottlenecks that hinder its further technological development.
Others at the event added that the U.S.
package of laws introduces more uncertainty into the development environment of the entire global industry, and Chinese authorities, educational institutions and companies should counteract this together.

Former president of Arm left the board of directors of Chinese SMIC

The worsening controversy between the PRC and the United States, provoked by the recent visit to Taiwan of American Parliament Speaker Nancy Pelosi, touched the activities of those companies that work in the segment of semiconductor components production.
After nine years on the board of directors of SMIC, former Arm president Tudor Brown was forced to announce his departure from the position.Image source: Ceres Power HoldingsThis move by the industry veteran, who was president of British processor architecture developer Arm until May 2012, was reported by Bloomberg with reference to his personal comments on the social network LinkedIn.
Tudor Brown admitted that he regretted leaving the board of directors of Chinese contract chip maker SMIC, where he had served for the previous nine years.
He held the post of president of Arm from 1990 to 2012, but left it even before the British developer bought the Japanese corporation SoftBank.According to Tudor Brown, international controversy in recent times only intensified, which forced him to leave the board of directors of the Chinese company SMIC.
It became known recently that the U.S.
export control rules were tightened and now China's SMIC will not be able to receive from any supplier lithographic equipment using U.S.
technology suitable for the production of chips at 14nm and coarser, although previously the border was held at 10nm.
According to independent experts, this has not prevented SMIC from setting up last summer's supply of components with characteristics close to the 7nm products of Taiwan's TSMC.

Micron, Intel and NVIDIA are not going to reduce the number of personnel in the near future

Micron Technology had to deal with contradictory statements in the past days.
On the one hand, it expressed its willingness to invest $40 billion to expand its memory production in the U.S.
On the other hand, it had to admit that demand for memory failed to meet expectations last quarter, proportionally reducing revenues.
Hire new employees in such circumstances Micron will not be able to, but will not cut staff, and it will be joined by Intel and NVIDIA.Image source: Micron TechnologyAt least, that's what Bloomberg representatives report, citing the companies' proprietary information.
In fact, Micron Technology will not suspend hiring even in this difficult situation, but will significantly limit its scope.
Micron executives aren't thinking about cuts commensurate with the level of trouble in the economy, either, as Bloomberg explains, citing its own sources.At a general staff meeting at Intel last week, management assured the team that it won't be cutting staff, as Bloomberg adds.
NVIDIA made similar statements this week, already after admitting to a slump in revenue.
In fact, company founder Jensen Huang even promised to raise employee pay to counter inflation.
NVIDIA's management is confident in the market opportunities open to the company and the technology it is developing.Analysts at Citigroup, however, do not share the optimism of these market participants, who are convinced that the correction in demand is temporary in nature.
According to experts, the semiconductor components market is entering its worst crisis since 2001, and the downturn will be observed in all market segments.

Chinese manufacturers of RAM lag 5 years behind Korean competitors

In contrast to the production of solid-state memory such as 3D NAND, in which the Chinese company YMTC lags behind the world leaders by literally one generation, the production of memory chips such as DRAM is given to Chinese companies with great difficulty.
According to estimates by South Korean experts, China's CXMT lags behind its Korean rivals in this area for a couple of generations and five years.Image source: SMICFor example, as explained by Business Korea, the leading DRAM manufacturing company in China, CXMT has declared readiness to start production with the technology standards of 16 to 17 nm this year, while the Korean SK hynix and Samsung Electronics in comparable terms are going to switch to 12-nm and 13-nm technology combined with EUV-lithography.
Each generation of lithographic technology in memory chip production is separated from the previous one by about two or two and a half years, so Chinese DRAM manufacturers lag behind their Korean colleagues by about five years, according to experts.The same CXMT, according to their data, suffers from a low yield of good products.
Even having mastered the production of memory using 19-nm and 18-nm technologies in 2019, it now receives only three-quarters of good chips from the silicon wafer.
In the next generation, CXMT's yield rate of good products does not exceed 40%.
As a result, the company's share of the global market does not exceed one percent.
According to Korean analysts, the Chinese memory chip industry will not be able to match foreign competitors as dramatically as it did in the display industry.The lag in solid-state memory technology from Chinese companies is noticeably smaller, but still reaches a couple of years.
Korean companies are going to mass produce 3D NAND chips with more than 200 layers by the end of this year or early next year, while China's YMTC is unlikely to do so before 2024.
The company can be kept from further convergence of YMTC with Korean competitors by the introduction of specific sanctions from the U.S., but so far this factor does not limit the development of this segment of the Chinese industry.In the segment of contract production of more complex microchips China is behind South Korea by about five years.
Samsung Electronics is already able to produce 5-nm and 4-nm products for its customers, while China's SMIC has not progressed beyond 14 nm - largely due to U.S.
sanctions, by the way.
Further development of advanced lithography without switching to ultra-hard ultraviolet light (EUV) seems difficult, but this is where U.S.
sanctions against SMIC come into play.

Vietnamese Intel learned how to prepare processor substrates themselves

As you know, the COVID-19 pandemic led to a shortage of all sorts of microchips.
Logistics went down and demand for electronics went up.
Intel in 2021 did a lot to ease the shortage of its main product, processors.
A significant step along the way was the expansion of processor substrate preparation at the company's assembly plant in Vietnam.Image source: IntelThe Ajinomoto build-up film (ABF) substrates are a combination of about 10 layers of glass fiber with contact groups on the bottom and top and multiple conductors inside.
The substrates serve as a base to mount the processor chip and as a safeguard to protect it from damage.
The substrate at the bottom carries a group of pins to fit into the processor socket on the motherboard, and the top is covered by a heat dissipating cover.
Also, surface mount capacitors are soldered on both sides of the substrate to reduce noise and to stabilize the processor power.Traditionally, Intel bought processor substrates with capacitors already soldered on one side.
At the Vietnam plant where the processors are assembled and tested, capacitors were soldered to the other side during final assembly of the processors.
But supply disruptions caused by the pandemic forced the company to take on the task of preparing the substrates for the assembly of processors, namely to unsolder capacitors on both sides of the substrate.In 2021, the company purchased a hundred pieces of special equipment for the plant in Vietnam and organized there a complete process of preparing the substrates for final assembly.
Since then, the company does not need intermediaries to perform such work, and this allowed it not only to maintain production volumes, but even to produce products in excess of the plan, even in the face of logistical problems and shortages.
According to Intel, it was able to produce millions of additional processor units in just over a year.
\"This initiative is a terrific example of how integrated manufacturing is at the core of Intel's success.
Our global network of factories and ecosystem of suppliers directly enables a more adaptable and sustainable supply of products.
Last year, when the industry experienced a substrate shortage, our ability to leverage internal capacity generated over $2 billion in revenue growth, allowing us to respond quickly to dynamic customer demand,\" said Keyvan Esfarjani, executive vice president and chief global operations officer at Intel.In just 15 years of operations in Vietnam, Intel has invested $1.5 billion in manufacturing.
By the end of 2021, the facility had delivered over 3 billion processors to market.
Jobs are provided for 2,800 employees and about 4,000 more specialized jobs have been created around the factory infrastructure.

Qualcomm has made it into Samsung's top five clients by virtue of its chip contract

In the first quarter of this year, Qualcomm managed to make it into Samsung Electronics' top client list for the first time, according to Korea's Yonhap News agency.
This was helped by an increase in revenue the Korean contractor receives from contract manufacturing services for mobile processors for the U.S.
developer.
Together, the top five customers account for about 14% of Samsung's revenue.Image source: Samsung ElectricsIn the last quarter, Qualcomm swapped places with Verizon on this list, although German telecom operator Deutsche Telekom retained its position among Samsung's largest customers.
Traditionally, competing with the Korean giant Apple is also among its largest partners, because it depends on Samsung in the supply of components for smartphones - the same displays, for example.
Best Buy and Supreme Electronics round out Samsung's top five sources of revenue.In December, Qualcomm executives confirmed that Samsung's 4nm technology will be used to produce the Snapdragon 8 Gen 1 mobile processor.
Now rumors attribute Qualcomm's rejection of the Korean contractor's services in the release of the Qualcomm Snapdragon 8 Gen 1 Plus processor, which will be released this Friday.
It is believed that it will be produced by rival TSMC.
Qualcomm's history of tossing from TSMC to Samsung and back is quite extensive in its content, but in terms of business strategy is intended to diversify the risks of dependence on a single processor supplier.According to TrendForce data for the fourth quarter of last year, TSMC controlled 52.1% of the global market for semiconductor contract manufacturing services, while Samsung Electronics was content with 18.3% of the market.
Last month, representatives of the Korean giant said that there is a high demand for the core services of the company, as well as the reduction of defect rates in the use of advanced technological processes.
Rumors about Samsung's intentions to raise prices for its services by up to 20% have not yet officially commented, but the Korean company's management in recent months is not tired to repeat that the pricing policy should be realistic and take into account the need to invest in expanding and modernizing production.

China's YMTC began sending out samples of 192-layer 3D NAND flash memory chips

Yangtze Memory Technologies (YMTC), the largest Chinese manufacturer of 3D NAND flash memory chips, announced that it has already begun sending out samples of the new 192-layer 3D NAND flash memory chips to customers.
A full-scale release of the chips could occur by the end of the year, according to DigiTimes.Image source: YMTC is producing 128-layer 3D NAND flash memory chips in volume production, providing strong competition for manufacturers in Korea, Japan and the United States.
Experts praise Chinese chips for their high density and throughput, which the company has managed to achieve during its transition to the new architecture Xtacking 2.0.
Xtacking architecture is a proprietary YMTC development.
It is the basis for the production of 64-layer NAND chips.
Last year, the Chinese manufacturer switched to Xtacking 2.0 technology, which has provided a number of advantages for the family of 128-layer products.DigiTimes reports that YMTC overcame all the difficulties associated with the mass production of 128-layer memory type 3D NAND and set up its production.
The company has increased its monthly silicon wafer processing capacity to 100,000.
Insiders say that the NAND memory from the company YMTC interested even in Apple, which will consider using them in their products.Chinese manufacturer has set a goal by the end of 2023 to win 7-8% of the world market of NAND memory production.
In terms of volume, this means processing 200 thousand silicon wafers per month.

Micron announced the development of the world's first 232-layer 3D NAND flash memory - 128 GB in one chip

American company Micron announced the development of the industry's first 3D NAND flash memory chips, consisting of 232 layers.
Mass production of the novelty based on TLC cells will start at the end of 2022.
The first products based on the new memory will be available in 2023.
The increased write density promises lower power consumption, higher speeds, and lower costs, which will be converted into either Micron's profit or lower prices for SSDs.Image source: MicronIn order to achieve the record number of layers, Micron used a previously proven method - it placed one 3D NAND chip on top of another, electrically connecting them and placing them on a common substrate (substrate and controller).
In the process of splicing the crystals some of the layers were inevitably lost, so from the two 128-layer crystals came not 256-layer 3D NAND chip, but 232-layer, which is also good.
It should also be noted that Samsung presented a 256-layer 3D NAND back in 2020, but mass production of such chips has not started.
Therefore, the previous record-holder can be considered a 176-layer flash memory chips, the production of which both Micron and Samsung began in the second half of 2021.
Also recall the Chinese company YMTC, which is about to appear in the arsenal of 192-layer 3D NAND.Another advantage of new products Micron can be considered as moving the control electronics under the chip with an array of cells.
This is a CMOS under array (CuA) technology, which all 3D NAND manufacturers have to a greater or lesser extent today.
Micron does not disclose detailed specifications for 232-layer chips.
It is known that the capacity of chips is 1 Tbit (128 GB) and in general they have become faster, which promises the emergence of even more efficient SSDs.

Micron announced the development of the world's first 232-layer 3D NAND flash memory - 128 GB in one chip

American company Micron announced the development of the industry's first 3D NAND flash memory chips, consisting of 232 layers.
Mass production of the novelty based on TLC cells will start at the end of 2022.
The first products based on the new memory will be available in 2023.
The increased write density promises lower power consumption, higher speeds, and lower costs, which will be converted into either Micron's profit or lower prices for SSDs.Image source: MicronTo achieve the record number of layers, Micron used a previously proven method - it placed one 3D NAND chip on top of another, electrically connecting them and placing them on a common substrate (substrate and controller).
In the process of splicing the crystals some of the layers were inevitably lost, so from the two 128-layer crystals came not 256-layer 3D NAND chip, but 232-layer, which is also good.
It should also be noted that Samsung presented a 256-layer 3D NAND back in 2020, but mass production of such chips has not started.
Therefore, the previous record-holder can be considered a 176-layer flash memory chips, the production of which both Micron and Samsung began in the second half of 2021.
Also recall the Chinese company YMTC, which is about to appear in the arsenal of 192-layer 3D NAND.Another advantage of new products Micron can be considered as moving the control electronics under the chip with an array of cells.
This is a CMOS under array (CuA) technology, which all 3D NAND manufacturers have to a greater or lesser extent today.
Micron does not disclose detailed specifications for 232-layer chips.
It is known that the capacity of chips is 1 Tbit (128 GB) and in general they have become faster, which promises the emergence of even more efficient SSDs.