Geek.com · Geek Observation October 13 Intel has been in a downturn in recent years. In the chip market, Intel has lost its technological advantages, its products are very complex, CEOs have been replaced like a lantern, employees have switched jobs, and the closed corporate culture has attracted criticism. Looking at the stock price performance, Intel is also far behind Nvidia and Qualcomm, although Intel’s sales are higher than them.

From another perspective, Intel is doubling down on Israel. Intel imported $8 billion worth of chips from Israel last year, a record high and twice as much as six years ago. Intel alone accounts for about 14 percent of all Israeli technology exports.

In Intel’s revival strategy, Israel plays a key role. At the beginning of the year, Intel appointed Kissinger (Pat Gelsinger) as its new CEO, and he claimed to focus on computer and server chips, AI and automotive chips.

Why is the once-giant revival? Because Intel is in decline. How did it decline? Some think Intel missed the mobile revolution, some think Intel didn’t get into the AI ​​market early, and others blame former CEO Brian Krzanich.

01 Missing Mobile Revolution and AI Opportunities

From the perspective of market size, the current mobile chip market has far surpassed computer and server chips. Some people believe that Intel failed to identify the potential of the new market early, and as a result, Qualcomm, Samsung, and TSMC became big.

Around 2005, Apple CEO Steve Jobs wanted then Intel CEO Paul Otellini to provide chips for Apple’s iPhone. Paul Otellini believed that Apple’s price was too low, and it would take a lot of effort to use Intel’s high-performance CPUs in mobile phones, so Apple developed its own chips and let TSMC manufacture them.

In order to win a piece of cake in the smartphone market, Intel is not without efforts. For example, Intel spent $1.4 billion to buy Infineon and invest in LG. Still, Qualcomm led the pack, leaving Intel far behind.

Intel also missed out on the AI ​​and machine learning wave. It used to be a small market, only for robot and computer game developers, but now it is beginning to penetrate into various industries, such as cybersecurity, automobiles, smart speakers, VR, etc.

A former Intel executive revealed: “Now that AI has become widespread, Intel has not put too much thought into it. The winner is Nvidia. In the eyes of everyone before, Nvidia was just a company that provides GPUs for gamers. It optimizes hardware for high performance. The performance game is on the line. Nvidia’s value has risen because of AI machine learning, and today its market cap is twice that of Intel, even though Intel sells more chips.”

Nvidia developed the software layer for CUDA GPUs, a model that gave it a monopoly with 81% of the AI ​​processor market. While Google, AMD, Xilinx, and Intel compete, Nvidia has an edge over the competition because of its powerful software.

The former Intel executive said: “If you use Intel hardware, engineers need to learn a new software language, and of course they don’t want to. Years ago, Intel won the x86 market through software and became the ruler of desktops, notebooks, and servers. Now Nvidia It worked the same way. In other words, Nvidia beat Intel in Intel’s way and owns the future.”

02 Loss of management

Others blamed former CEO Brian Krzanich for Intel’s loss. Brian Krzanich became Intel CEO in 2013 and stepped down in 2018. At that time, Intel ignored the production line problem, delays and delays became the norm, and ultimately Intel missed the opportunity. During Brian Krzanich’s tenure, Intel failed to release the most advanced chips on schedule, which has never been seen in Intel’s history.

In 2015, Intel launched the 14nm production line, but then the 10nm technology was delayed, initially delayed by 6 months, but finally delayed by 3 years. Dell, a big customer for Intel, cut its sales forecast by $1 billion last year because of chip delays.

Molikhim l’Hatzlakha is an Israeli who revisits Intel’s history in a book. In 1992, a junior Intel employee in Israel sent CEO Andy Grove a fax that changed Intel’s fortunes. At that time, David Perlmutter and two other engineers told their superiors that they should maintain a good relationship with Microsoft and not rewrite the CPU core software. They warn that upper-level policies can hurt the application environment, which is built on Intel architecture. “Fax eventually gave birth to the Pentium processor, which changed the world of personal computing, turning the IBM-compatible PC into a product for every household,” the book says.

Molikhim l’Hatzlakh believes that top-down management later influenced Intel’s development. Many elites fled when Brian Krzanich was in charge of Intel. A former Intel executive said: “You don’t have a heart attack in a day, it’s the result of decades of unhealthy nutrition, and Intel hasn’t invested much in mobile for decades. Even someone like Andy Groe The big guys like my husband did not invest in mobile chips, although the first chip used by Nokia was the Intel 386.”

Another former Intel executive said: “Intel is blinded by success. It refuses to see what competitors see, and over the past few decades, the Intel CEO position has turned Intel into another company. In the beginning At the time it was a group of companies with PhDs in engineering and chemistry who were also entrepreneurs, but over the past 20 years, a lot of Intel CEOs have come from manufacturing or finance. These are people who understand technology or understand finance, but don’t have the vision.”

Since the outbreak of the new crown, the advantages of Nvidia and AMD over Intel have further expanded. Nvidia has become the leader in AI chips, not only that, but its share in the data center market is also increasing, and Nvidia may also take ARM.

In the PC chip market, AMD has caught up with Intel, and the technology is even ahead. Market experts generally believe that AMD’s new server chips will outperform Intel. In the field of chip manufacturing, TSMC has surpassed Intel. TSMC is producing 7-nanometer and 10-nanometer chips, and will launch 3-nanometer chips next year. On the other hand, Intel is working hard to build a 7-nanometer chip factory, and a 4-nanometer factory is in the planning stage.

03 The wishful thinking of the new CEO

Now that Pat Gelsinger is the new CEO of Intel, he wants to start a new era, and the first is to take back the manufacturing advantage and focus on the focus. Intel has launched some small acquisitions that have nothing to do with Intel’s core businesses, such as sports imaging technology developer Replay Technologies, 3D sensor printed circuit board developer InVision Biometrics, and Kissinger’s approach to these businesses is simple: shut down.

In terms of manufacturing, Kissinger is willing to transfer some unimportant chips to competitors, while Intel also produces chips for other companies. This latest move is in line with a principle offered by the legendary Andy Grove: “Intel has a goal: Our factories must always be running at full capacity, because a factory that doesn’t use 100% of its capacity is a money-losing business. .”

Look at Kissinger’s investment again, investing $100 billion to upgrade buildings and build new factories, of which $20 billion is for the construction of the Arizona factory (2) and $35 billion to upgrade the New Mexico factory. Intel is also preparing to build a Kiryat Gat factory in Israel. In Europe, Intel is also preparing to invest $100 billion in local factories, as well as in the Israeli automotive chip sector. In short, Intel wants to be the EUVL leader.

Intel will launch 18A chips before 2025, and it will not be evaluated in nanometers, but Angstroms (equivalent to one-tenth of a nanometer). Kissinger is also preparing to buy chip makers, GlobalFoundries is a potential target. GlobalFoundries is valued at about $30 billion, but it appears to have no intention of selling, but rather to go public.

04 Data Center AI Chip

Intel may be the last chip maker in the western world, and the geopolitical significance is very important. A source said: “Kissinger is very aggressive, he will not tolerate Intel to maintain the status quo. Kissinger is good at decision-making, action, either save or kill, that is his style. How will Kissinger in the future with Dancing with Intel is something we can look forward to.”

Compared with its competitors, Intel employees have more seniority, but because of the large scale of the company, sometimes it is impossible to move forward quickly. In order to recruit talents, Intel’s salary for new employees is very competitive.

Intel also needs to adjust its acquisition strategy. For example, in Israel, Intel closed or sold many acquired small businesses, including Replay Technologies, RealSense, Oplus Technologies and DSPC. A former Intel executive said: “Intel’s approach is similar. Immediately after the acquisition, employee training is arranged to understand Intel, understand the corporate culture, the office system of the new employee is replaced, and then the employee is put into the unified Intel cubicle. Months. Afterwards, Intel asked them: How can I make the output as it was before the acquisition? That’s why the founders of Mobileye and Habana Labs asked to remain independent after the acquisition, at least for the first few years.”

Habana Labs is very important to Intel, and it is an important part of its AI strategy. At the end of 2019, Intel wants to enter the machine learning market by acquiring AI chip developers. Many people think that Intel’s acquisition of Habana Labs is a bit late and expensive. Habana Labs has not launched any products so far, and it will take time to integrate Habana Labs technology into Intel server boards. But with Facebook likely to preempt Habana Labs, Intel had to move fast and buy Habana Labs for $2 billion, knowing that the company hasn’t made a single penny of revenue.

Unlike Intel, Habana Labs’ technology is based on ARM, but Intel is convinced that Xeon server chips can be integrated with motherboards powered by Habana Labs chips to improve data center performance. Data centers are shifting to the cloud, and Nvidia and AMD are competitors to Intel.

Before Intel’s failure in the mobile and IoT chip market, Kissinger has a new strategy. He wants to focus on high-performance PC server chips. Such chips are much stronger than mobile chips and can bring in 40 billion in annual revenue. Dollar.

Although Intel has acquired Mobileye for many years, Mobileye is still an island within Intel. Mobileye still operates independently. Mobileye accounts for a very low proportion of Intel’s entire revenue, only $1 billion per year, but Kissinger values ​​Mobileye very much. He believes that if Intel wants to gain a share of the smart car semiconductor market in the future, it can only rely on Mobileye. By 2030, Intel expects the automotive chip market to account for one-fifth of the high-end automotive component market and one-tenth of the global chip market. Other chip companies don’t have the automotive chip business, Intel has it, and that’s an advantage.

Intel’s revival is still a mystery, but at least Kissinger is working on it!