Chinese semiconductor industry

ansy1968 said:

A realization but its to late now.... from hyber horse of China military forum

Huawei Plays Matador to Trump's Bull | Opinion
David P. Goldman , Deputy editor and columnist, Asia Times
On 10/14/20 at 7:00 AM EDT

Bulls usually lose bullfights not because they fail to charge, but because they charge at the matador's cape rather than at the matador himself. President Trump has no qualms about charging at China's national champion Huawei, but he is charging at the equivalent of the matador's cape.

Consider Huawei's handset business, which depends on high-end chips manufactured by Taiwanese foundries. Because the Taiwanese use some American equipment to manufacture the chips, the U.S. has imposed an effective blockade on Huawei's outsourcing, and China does not (yet) have the domestic capability to make high-end chips. A similar ban blocks Huawei's access to key American chip-design software.

Huawei will lose market share in the low-margin handset business, and Chinese competitors like Xiaomi will pick up the slack (which explains why the latter's stock price has jumped this year from HK$8 to HK$22). Huawei will have little difficulty using older chips for 5G base stations, installing the lion's share of the six million units that China has ordered to build out its 5G network over the next two years.

Huawei's most important business, though, will emerge unscathed from U.S. sanctions. American strategists think that Huawei is a telecom equipment company. Among other things, Huawei is the world's largest telecom equipment manufacturer, with a 31 percent market share during the first half of 2020—more than the combined share of the second and third position companies, Ericsson and Nokia. Huawei, though, is first and foremost a big data and artificial intelligence (AI) company—the most inventive and imaginative one in the world. Let's call this combination "BD/AI," for short.

BD/AI has already had world-moving consequences in health applications. For example: China will test all nine million inhabitants of Qingdao for coronavirus over five days, after medical authorities detected nine cases of infection. The United States, meanwhile, was reporting an average of about 50,000 cases per day as of earlier this week. Notably, the U.S. has done far more testing than China per capita, with more than 357,000 tests per million of population to date, compared to just 111,000 in China, according to Worldometers.info. Apart from selective, short-term lockdowns, life in China has returned to normal, with more than half a million tourist journeys booked during last week's mid-autumn holiday—all while much of daily life in America remains paralyzed by the pandemic.

The difference lies in data and use of AI to analyze the data. China required far fewer coronavirus tests per capita because its AI systems identified prospective clusters of infection. Small amounts of selective forensic testing guided by AI forecasts preempted larger outbreaks—and where smaller outbreaks occurred, massive testing extinguished the spread of the virus.

If AI is the engine of the Fourth Industrial Revolution, data is the engine's fuel. Gathering and porting data into a usable framework, as Huawei Chief Technology Officer Paul Scanlan explained in my recent book, is the "control point" in today's world economy. There's no secret about this, no recondite conspiracy, no subtle design. Huawei proclaims its intentions at conferences streamed live on the web, and encapsulated in YouTube videos like this one.

Oil determined much of the geostrategy of the past century. Hitler sent the German army south toward the oil-rich Caucasus, rather than east towards Moscow in 1942, and the British fought Rommel in North Africa to protect the Suez Canal route for its oil tankers. Russia and the United States wrangled over the oil-rich Middle East throughout the Cold War. The Arab oil boycott of 1973 sent the U.S. into recession and near defeat in the Cold War; the collapse of oil prices in the 1980s hastened the fall of the Soviet Union. All the old geopolitical models arose from concern over trade routes as control points for the transshipment of oil. They are now as obsolete as Macedonian phalanx tactics in the 3rd century B.C.E., or Napoleonic artillery manuals.

Data is now the control point. A central premise of American policy toward China is that Huawei intends to steal Western data. On the contrary: China is the main collector and provider of data in the fields where it counts the most—for example, medicine. Every European pharmaceuticals company of importance has a BD/AI joint venture in China, because China has digitized the health records and sequenced the DNA of hundreds of millions of its citizens, and can correlate this data to real-time vital signs of hundreds of millions more people. To be sure, Chinese citizens have no protection by privacy laws that have made it impractical to digitize health records in the United States. This is a soluble problem, at least in theory. The technology exists—for example, through block chain accounts—to anonymize individual records while permitting data collection.

Ant Financial's 900 million customers comprise the largest consumer finance database in the world. The company's pending $40 billion initial public offering, the largest in history, reflects a multi-hundred-billion-dollar valuation that stems from Ant's capacity to analyze consumer and small-business creditworthiness in a way that may put traditional banking out of business. TikTok's two billion downloads gives its parent, ByteDance, the largest data set in the world for analysis of consumer preferences and behavior. China's "smart cities," the first large-scale 5G application, will match tens of millions of riders and packages to the location of vehicles, drastically cutting the waiting time for personal transportation and package deliveries.

As I report in my book, Scanlan explained what AI would do for manufacturing. "Let's take robotics today," he told me. "5G changes everything. Typically, 5G is spoken about in terms of download speed, but that's not the most important advantage. For industrial processes, autonomous vehicles and other applications, the latency—the time it takes for one device to acquire and respond to a signal from another device—is more important. On a factory floor today, Robot A does its instruction, passes the bit to Robot B, and it's the same thing. Now, if we put very low latency inside each of the robots—and they can be robots from different manufacturers—and put them in a room and give them the rules, like Go or like chess, to enable them to connect in real time—milliseconds, lightning fast—and then put a bit of plastic in view and say, I want you to make a plastic cup, the robots will organize themselves much better than we would have thought."

Click to expand...

machupicu said:

We shall see +1 yr from now.. rumor has it that Huawei will sell its Honor smartphone division for about $4 billions so that hw can concentrate on high-end smartphones! So what it means: Huawei is Confident about solutions to making 7nm n 5nm chips, no?!



"Thus, the company might be able to outfit its current generation of a flagship with 5nm chips, but the same might not be said for future launches in the coming year. At the moment, a possible supplier for chips was China’s largest semiconductor based firm, SMIC. But unfortunately, the company’s most advanced solution at the time was 14nm processes. A 5nm process has about 171.3 million transistors per square mm, on the other hand, a 14nm process only has about 43 million transistors in the same space. So the higher the number of transistors, the chip is more powerful and energy efficient.

Please, Log in or Register to view URLs content!

Click to expand...

ansy1968 said:

A realization but its to late now.... from hyber horse of China military forum

Click to expand...

machupicu said:

Innosilicon and SMIC collaborated for years and the contract chip designer says that it has helped dozens of SMIC’s customers to develop chips using both mature (e.g., 55 nm, 40 nm, 28 nm) and FinFET-based process technologies (14 nm, 12 nm, 7 nm), so it is not particularly surprising that Innosilicon is the first company to try SMIC’s N+1.

Even if SMIC and Innosilicon have taped out the first commercial N+1 chip, it might take a year before that ASIC will be mass produced. In any case, the tape out means that at least some IP for N+1 is now available and Innosilicon knows how to use the technology.

SMIC’s N+1
SMIC’s N+1 fabrication technology is the company’s next major node that will follow the foundry’s 14 nm and 12 nm manufacturing processes.

Please, Log in or Register to view URLs content!

Click to expand...

WTAN said:

Also note that the 100% Local 14nm Production line will be ready next year. The 7nm Semi Equipment is being currently developed and some companies have already successfully developed 7nm Semi Equipment.

Click to expand...

horse said:

Hi ansy1968, you really deserve to watch this video if you have not seen it before. Lifted from same internet, a few months ago.



Aug 13, 2020

Click to expand...