Intel Core: We have no interest in CPU market share

In the past two years, under the strategic planning of NVIDIA, Qualcomm, AMD, Intel, Huawei and other technology companies to accelerate the layout, the competition in the chip field has intensified. Now, with the advent of the era of artificial intelligence, new vitality is once again stimulated for the chip market, and at the same time, this means that a new round of chip wars is about to start.

Intel’s complex “core thing”

One word to describe Intel in the Moore era is that it has strength inside and strong support outside.

This made it almost no heavyweight opponent in the processor market at the time. Yes, AMD didn’t deserve to be compared to Intel at the time. Some experts even believe that if not for the constraints of anti-monopoly laws, Intel would probably have knocked down or acquired AMD.

And the time turned back to 2019, and the style of painting changed suddenly. It is believed that in the long run, Intel will lose its absolute monopoly in the global processor market. In the end, Intel and AMD’s market share even reached a balance point of 50% each. What’s even more mysterious is that this scene is already happening. In the Korean regional market, AMD has actually reached 53.58%, and it has begun to get rid of the identity of the “Millennium Second”. AMD’s counterattack is as embarrassing as Intel’s decline. And carefully study Intel’s helplessness, there are two key nodes are worth noting.

One is the “lost and hard to find” x86 CPU market.

Although Intel restarted the Pentium G3420, it has become an indisputable fact that the x86 CPU market has gradually been captured by AMD. AMD’s processors have a whopping 78 percent share, and have held steady at that level since July.

And just a few years ago, Intel held more than 90% of the x86 CPU market.

Interestingly, this is what Intel “actively” gave up.

The second is the new AI fortress that cannot be conquered for a long time.

Taking the initiative to give up the country of the CPU, what is Intel going to do? The answer is to hit 30% of the market share in the “all-silicon” field.

To put it simply, it is attacking the GPU market, the AI ​​market, and the FPGA market.

That’s why Intel has gone into a “shopaholic” mode over the past few years. In 2016, it acquired Movidius, an edge and vision AI startup. In the same year, it won Nervana for $350 million and entered the deep learning training chip market. In 2017, the acquisition of Israeli self-driving company Mobileye for $15.3 billion made the industry even more astonished. If the acquisition of Habana Labs is completed, Intel’s combination of AI products can be said to cover edge computing, data centers, and various business supports from CPU to ASIC.

Could this help Intel win new markets? It has to be said that Intel, which entered the game late, can only be said to be a “brother” in the field of artificial intelligence chips.

In the past few years, Nvidia has become the absolute darling of the capital market by virtue of its early investment in the field of artificial intelligence, and has also become a leader in this field, occupying an absolute advantage. In the list of global artificial intelligence chip companies in 2018, NVIDIA firmly holds the top spot.

Intel Core: We have no interest in CPU market share

However, Intel, which has spent a lot of money, is not weak. As early as 2016, it launched the Nervana AI chip dedicated to deep learning training and deep learning inference. At 2019CES, it also launched the Nervana Neural Network Processor (NNP-I) . Its accumulation in related technologies is also comprehensive, including OpenVINO, a tool suite for AI edge computing development, which can be used for inference chips, as well as vision processors such as 16nm Myriad X VPU, Stratix 10 FPGA cloud chips, and more.

At present, it seems that Intel has laid out in multiple silicon tracks, as well as 5G, cloud, edge computing and other fields, but in the rapid changes and iterations, it cannot determine the winner in one game.

The most direct reason is that Intel, which used to be a “dugu seeking defeat”, is now facing challenges from Google, Facebook, Microsoft, Amazon, as well as Chinese manufacturers such as Huawei and Cambrian, who are constantly bringing new technologies and products to the market. , and each has its own merits. According to the New York Times, there are more than 40 startups entering the field of AI chips.

The competition among powerful competitors and whether Nervana has given up on its previous achievements all make Intel’s future in the AI ​​hardware landscape even more unclear. The most direct impact is that Intel’s performance in high-performance computing is not as good as imagined. Because competitors do not have the high profile to hand over the market. Not long ago, Nvidia struck a deal with Microsoft, Intel’s former owner, to deploy one of the cloud’s largest GPU-accelerated supercomputers on Azure. AMD Epyc processors also have new customers such as Amazon and Microsoft Azure.

Intel may not be as competitive as expected in the field of high-performance computing.

transformation logic

Intel is best known for its personal computer processors. But starting in 2016, the company has chosen a new track for itself, intending to become a data company.

In 2013, the rise of artificial intelligence made AI chips become a “net celebrity” in the global chip industry. AI chip startups appeared frequently, and large Internet companies or cloud computing companies such as Google, Amazon, Baidu, Ali, Huawei, etc., have developed AI by themselves. Chips are still in the ascendant.

This means that the demand for computing power in the future intelligent world has changed significantly. Although general-purpose computing chips will still be the mainstream chips in data centers, large-scale data centers increasingly require diversified and customized computing power. Process larger data and execute more complex algorithms.

Not only that, the superposition of 5G and IoT technologies not only brings about the explosive growth of data, but also makes data computing and storage ubiquitous.

This round of Intel’s transformation is based on the judgment of the trend. The so-called data company has two meanings for Intel. First, this does not mean that Intel wants to create data, nor does it mean giving up chip design and manufacturing, but Intel believes that the future world is based on data, and computing will be ubiquitous and diverse, and Intel has to meet this demand. Second, Intel not only satisfies the chip manufacturing itself, but also needs to penetrate more links in the industry chain, so that industry users can better calculate, store, transmit, and utilize data.

It makes sense to look back at Intel’s first transformation at this point. Founded in 1968, Intel Corporation is the world’s largest semiconductor manufacturer and one of the most representative companies in “Silicon Valley”. The two founders of Intel, Robert Noyce, who invented the integrated circuit, and Gordon Moore, who proposed “Moore’s Law”. The company’s third employee, Andy Grove, was the core driver behind Intel’s first transformation and one of the world’s masters of corporate management.

Intel’s original product was a semiconductor memory chip. In 1969, Intel launched its first batch of products, the 3101 memory chip, followed by the 1101 and 1103. This cheap and high-quality product was very popular and in short supply. Its birth officially announced the demise of magnetic core memory. In 1971, Intel’s stock went public.

Due to Intel’s continuous innovation in memory chips, the global computer industry has undergone a revolutionary evolution. At that time, Intel looked at the world in this field, with a market share of nearly 100%. But in the 1980s, Japanese companies suddenly rose up and quickly devoured Intel’s memory chip market share with huge investment and amazing efficiency.

Grove, who was Intel’s CEO at the time, was known for his “paranoia,” and after meeting after meeting and endless bickering, he made the decision to shift the line. This was the first major transformation in Intel’s history, and at the time, it was the equivalent of giving up one forest and rebuilding another, which was unacceptable, both inside and out.

Grove decided at the time that the computer market was likely to grow rapidly, but as long as Intel still shared its designs with other chipmakers, it would be left as an ill-fated parts supplier, beholden to IBM, which was 60 times larger than it. . Intel must make itself the sole source of microprocessors to dictate its own destiny.

However, conservatives believe that Intel can further build a deep moat in this already-established forest and try to defeat new entrants. After all, the overall market for memory chips has not declined. Compared with switching tracks, this choice is safer.

Grove eventually insisted on rebuilding a forest. Beginning in the fall of 1985, Intel devoted all its efforts to the development and research of microprocessors. A year later, it launched the 386 microprocessor, which was of trans-generational significance at that time, activating the market in one fell swoop. Taking this as a node, Intel, the king of memory, gradually disappeared, and the more powerful microprocessor empire gradually stabilized.

By 1992, Intel’s sales reached $5.8 billion, and its profits exceeded $1 billion for the first time, making it the world’s largest semiconductor company. Intel and Microsoft replaced IBM and gradually became the leaders of the entire computer industry.

Before and after the comparison, the external environment of the two transformations is completely different. In the first transformation, Intel’s existing track has not shrunk, but it has entered more powerful competitors; this time, Intel’s track has been greatly expanded, and new opportunities need to be seized.

The same is that the purpose of Intel’s two transformations is to seize a broader market. The last round of transformation helped Intel create a new era of PCs; if this round of transformation is successful, Intel will not only stand behind the scenes in the digital market, but become the protagonist. Navin Shenoy said the biggest opportunity in Intel’s history lies ahead, and Intel wants to seize it.