Tiger does not speak, thought it was a sick cat.
In the face of media long-term voices of doubt, Intel global vice president and president of China Yang Xu finally issued a cry, face excited but slightly more urgent.
This scene took place at an Intel event on Sept. 19. Sitting next to Yang Xu was Yang Xu, a voice-laden voice that stalled at Stacy Smith, executive vice president and president of the manufacturing, operations and sales group, who first responded Come up, raised his hand pat Yang Yang's shoulders, smiling.
In recent years, Yang Xu's days are not good .A person close to Yang Xu told Sina science and technology, for a long time in the past, the company has always been in the development of passive, customers, media, colleagues and even government officials, on many occasions Proposed to Yang Xu a similar question: Such a technology-renowned multinational corporations, in the present moment of technological change, how quiet movement so small? Really lagging behind? ......
Wrong, perhaps not in Yang Xu, but in the company's higher level of decision-making, and even the global strategic layout.Data show that 7 years ago, Intel also occupy the entire semiconductor chip market nearly 80% share, and last year, The data dropped dramatically to 14%.
Intel, the blue giant who has dominated the industry for 40 years, fell on the mobile Internet that has emerged in recent years and is facing multiple encircleations from Qualcomm, Nvidia, Samsung, AMD, etc., Dilemma.
Time to run the second half of this year, Intel ushered in a few good news, Yang Xu, who may see a turning point in the third quarter revenue this year, net profit of 4.516 billion USdollars, creating the best profit level over the same period, in the computer business , Data center business, Internet of Things business, etc. In addition, there are legal disputes between Apple and Qualcomm, the news that Apple is considering the use of Intel chips instead, which will undoubtedly make Intel brand and production capacity to further enhance.
Say that the ship is hard to turn around, Intel ship carrying 100000 staff warships turned for three years, whether it has been heading, whether to continue Yang Fan it?
Lost to the fall
October 3 this year, Intel officials sent a bad news: Former CEO Paul - Otellini died in his sleep, at the age of 66.
Otellini is Intel's fifth CEO, and at his helm eight years, Intel's financial performance has soared: from 34 billion U.S. dollars in 2005 to 53 billion U.S. dollars in 2012.
At the same time, that is eight years, Intel missed the mobile Internet.
Time goes back to May 17, 2013. On that day, Otellini announced its departure and 'confession' was a mistake: "Intel chips could have appeared on Apple mobile phones, but I rejected Apple's proposed transaction at that time."
Jobs had found Otellini that year, put forward the demand for chips and gave a quote, but the offer is lower than the cost of Intel's budget, Otellini rejected.
But now it seems that the budget cost was wrong at that time, "Otellini said in an interview that no one knew what it could do until the launch of the iPhone. In fact, it shipped more than anyone imagined Higher than a hundred times.
On the other hand, on June 27, 2016, Marvell acquired the Intel Communications and Applications Processor Group for $ 600 million in cash and won the Intel XScale product line.
IT analysts told Sina Technology, this move means that Intel really gave up the mobile market.Although it started in the late 90s to produce mobile phones and other handheld chips, but eventually lost to competitors such as Qualcomm, has not been published for many years Performance in this area, the possibility of perennial losses.
It is worth mentioning that, Intel's mobile processor development is always facing the problem of slow progress, 'to seize the era of the 4G market, QUALCOMM and MediaTek at the earliest release of flagship models every six months, and Intel's SoFIA mobile phone chips remain at 3G technology, even if the market is not competitive '.
The person believes that Intel is giving up the telecommunications sector is helpless move.On the one hand, comparing the data from the parties that year, Intel in the era of 4G seriously behind, courageously catching up not only need to continue to invest a lot of money, but also the introduction of a large number of talents and technology, and On the other hand, when the smart phone market tended to be saturated, various mobile phone manufacturers had already made a choice, Intel not only did not have the new customer, at the same time also lost the existing customer, 'For example, Asus's ZenFone, reduced Intel's order'.
In fact, from 2016 Intel's strategic layout, the company has indeed given up the mobile era, but focused on 5G. 5G era is recognized as the Internet of things, all devices will be networked, then the amount of data will multiply by geometric multiples, Need more computing power and processing power, which is undoubtedly the best Intel.
Gui </s> and resistance
And take the initiative to give up the mobile era is different, in Moore's Law, Intel has always been sticking, and even can be said to maintain.
The reason why Yang Xu 'cry, driven by 10nm process technology announced by Intel, the technology for' Moore's Law is dead 'gave a positive response to the remarks.
Moore's Law is not a natural law. Instead, it is proposed by Gordon Moore, one of Intel's founders, that when the price is constant, the number of transistors that can be accommodated in an integrated circuit will be doubled every two years or so , Performance will be doubled.That is, the same cost, performance will be more than double every 2 years.
The law of the decades after the birth of the law, the semiconductor process technology basically in accordance with the law of development, the most obvious is reflected in the naming of the previous generation of 0.7 times the process of naming the next generation products, so Intel's process named 90nm, 65nm, 45nm, 32nm, 22nm, etc., basically follow the rules of upgrading a generation every two years.
However, Intel is slow again.
In the 22nm up to 14nm, 14nm up to 10nm process, the cycle more than two years.Especially the latter, Intel was technically challenged, the Tick-Tock strategy shut down, one year to upgrade the process to enhance the structure The pace is broken.
After a lapse of 3 years, Intel released the 10nm process, which is nearly a year later than its competitors. At the same time, it leads to doubts about the failure of Moore's Law.
In fact, in many media view, Samsung and TSMC are the promoters of this law in recent years, and the two introduced the 10nm process at the end of last year.When talking about Intel in the industry, the usual argument is that technology To the bottleneck, 'gas is exhausted'.
Of course, Intel does not agree with this view, for a long time silent senior officials finally sit still.
At a technology exchange, Intel chief executive Mark T. Bohr pointed out that the merchants were nailed on by name and determined that the merchants had taken a process-based digital marketing game in order to snatch the market.
Mark T. Bohr bluntly, some companies deviate from Moore's Law on the naming of the process technology. Even if the transistor density increases very little, but still continue to promote the use of a new generation of process node naming, 'This also led to the process node name can not correctly reflect this Where the process is located on the Moore's Law curve. "
Intel has been asking itself to look forward to three generations of manufacturing process, which means that to see the technology seven years in advance, will lead the business for many years, Stace Smith's statement was significantly tougher.
Intel internal staff disclosed to Sina science and technology, the company very much do not agree with the name of the Friends of the Samsung and TSMC obviously stir-fry 7nm process, this year also speculation 5nm process, '7nm not mass production time, 5nm why talk about it? '.
The staff further explained that from a technical point of view, multiplying the gate height by the height of the logic cell would reduce the area and increase the density. "We figured out that the 14nm to 10nm miniature technology is stronger." In other words, last year's 10nm technology and Intel's 14nm technology similar, 'the same 10-nanometer process, we are twice as big than the density of your business.'
A PR manager came to share a story on a brand night at Intel on November 27. She said just as hotel staff saw her carrying an Intel badge and said to the 'chip company' Sure, 'I was trying to tell him (service staff) that Intel was not just a chip company, we also had AI, driverless and 5G, it was a home data company!'
Her story won the applause from the audience of executives, who, from top to bottom, was ashamed to rephrase itself as a chip maker and even canceled the IDF (Intel Developer Forum), a 20-year-old IDF.
In the eyes of Intel, IDF has been around a group of companies that have been in contact with the PC for years, and Intel's center has moved from PC to data and these new businesses can not find a partner at the IDF conference. Instead, Engage in small forums to replace IDF with activities in different fields, such as AI Day.
'Data is a new oil, with new data processing capabilities in mind,' said Song Qijiang, president of Intel China Research Institute, in an interview with Sina Technology said that robotics, communications infrastructure and visual technology will be The future direction of the company.
SongJijiang said Intel's data strategy is quite clear, that is, with the cloud and data centers, Internet of things, storage and 5G form a virtuous circle of growth, driving cloud computing and everything connected.
Although the strategy is clear, but in practice, Intel can be described as 'self-defeating' in all fields.For example, automatic driving, Intel spent 15.3 billion US dollars in March this year, the acquisition of Mobileye, the plan will try to increase Intel's market share and competition in the field of automatic driving Force, and the money accounted for 70% of Intel's cash reserves.
But it faces four old rivals: NVIDIA, Google, Qualcomm and Apple, and NVIDIA, in particular, began investing in more than a dozen companies a decade ago and has increased its autopilot and AI spend in the last two years It is difficult to say whether Intel, a company that spends too much money, can compete with it.
And for the current hot artificial intelligence field, Intel also worked hard.'We did not talk too much in the past, but our research is very cutting-edge ', Song believes that the current stage of artificial intelligence market is not, but in the next three years Growth rate may be 12 times today, 'Intel's current investment more timely, in the future to produce commercial value, to ensure that our products already exist.'
Indeed, technologies such as artificial intelligence and the Internet of Things have been given high hopes by Intel, finding other revenue-generating growth points in the face of the PC downturn. It is even more important to focus on the industry direction and market rhythm.
Encourage everyone to do the prototype early, to fail early on failure, "said Song Jiqiang Intel Institute currently a total of 800 people, the main focus of the four directions: the circuit and architecture; systems and software; 5G communications and security technology, China is the United States outside The largest research center, full-time research staff of more than 60, the total number of research institutes less than 100.
In this total of less than one thousand team, every year to have more than 200 projects submitted, some of which will be Intel practice and application of these projects, representing the forefront of Intel's technology, but also carries its' turning over dream '.