This situation has created a difficult situation for Pat Gelsinger, the CEO of Intel. He took the helm of the big American chipmaker three years ago, but Intel’s problems date back decades.

Intel, whose name was once synonymous with “Silicon Valley”, had gradually lost its lead in the production of semiconductor products to foreign rivals such as Taiwan’s TSMC before Gelsinger returned to the company in 2021. The blue team is now spending billions of dollars every quarter to reclaim the playing field in a risky effort.

According to CNBC, Intel shares have been the worst performer among tech companies on the S&P 500 list, with a 37% price drop in 2024. On the other side of the field is Nvidia, the American chipmaker and super microcomputer, which is experiencing astonishing growth as demand for artificial intelligence servers increases.

Intel CEO Pat Gelsinger

Intel, long the most valuable U.S. chipmaker, is now not only one-sixteenth of Nvidia by market capitalization, but dwarfs all other competitors—Texas Instruments, Qualcomm, Broadcom, and AMD. Let’s not forget that the company was the world’s largest semiconductor company by sales for decades, but has now posted seven consecutive quarters of declining revenue. The biggest chipmaker of the late 20th century fell behind Nvidia last year and is now lost in the dust of the green team.

Gelsinger has gone for a risky business model. He is leading Intel to make chips for other companies such as Nvidia, Apple, and Qualcomm, in addition to producing processors for itself. In this direction, Intel faces strong competitors such as TSMC and Samsung, and its success in attracting customers depends on its ability to regain the leadership of manufacturers.

Semiconductor companies want an alternative to TSMC so they don’t have to rely on just one supplier as the shadow of war hangs over Taiwan. US political leaders call Intel the American chip champion and say the company is an important part of the US processor supply chain.

The blue team has lost the rhyme for years. In fact, after the release of the iPhone in 2007, Intel lost the mobile chip market, and even now, although it has a great desire to enter the field of artificial intelligence, it has not yet been able to convince companies such as Meta, Microsoft, and Google to use its products instead of Nvidia chips. to order How did the world’s largest chip maker come to this day?

The iPhone could have an Intel chip in its heart. In his 2011 biography of Steve Jobs, Walter Isaacson revealed that when Apple was developing the first iPhone, Steve Jobs met with former Intel CEO Paul Otellini to discuss using Intel chips in the (yet-unreleased) iPhone. The first generation of iPhone was released with MacOS operating system and it was logical that Apple wanted to use Intel chips in it, because the best computers of the time used these chips.

But Jobs abandoned the Intel chips for two reasons: first, Intel was very slow in supplying them, and second, Apple did not want those chips to be sold to its competitors. According to Isaacson, the two tech giants could not agree on the price and intellectual property of the chips. Thus, the deal was stopped and Apple asked Samsung for help.

Steve Jobs, the founder and former CEO of Apple introduced the iPhone 3GS

The first iPhone with a Samsung chip was released. Apple bought PA Semi in 2008 and unveiled its first chip in 2010. In just 5 years, the annual sales of the iPhone exceeded 100 million units. In 2010, the sales of smartphones, both Android and iPhone, surpassed the sales of personal computers. The era of mobile monotony had begun.

Chips based on the ARM architecture have improved rapidly due to the huge production volume and the mobile industry’s need for faster performance and new features every year. Apple placed huge orders with TSMC, the first of which was the order to make the beating heart of the iPhone 6, the A8 chip, in 2014. Cupertino’s orders have flooded TSMC with cash. The Taiwanese company used this liquidity in the best possible way to upgrade its equipment annually and gradually overtook Intel.

By the end of the 2010s, flagship phone processors were challenging Intel’s desktop and laptop chips in some benchmarks, while consuming much lower power. In 2017, Apple and Qualcomm began adding neural processing units (NPUs) for artificial intelligence capabilities to their chips to gain another advantage over Intel processors. The blue team joined this arena late and released its first processor equipped with NPU late last year, thus gradually ceding a large part of its share in the PC chip market to ARM’s revolutionary chips 7 years ago.

In 2020, Apple removed Intel processors from its desktop and laptop computers. Since then, Macs have only come with chips based on the ARM architecture. This year, some flagship Windows laptops, including Microsoft Surfaces, are going to be released with ARM chips. Low-cost laptops equipped with Google’s Chrome OS are also increasingly turning to these chips. Gartner analyst Mikako Kitagawa says: “A large part of Intel’s market share decline is due to Apple saying goodbye to its processors. Cupertino alone owns about 10% of the world’s personal computer market.

Chips get faster as their number of transistors increases. More transistors allow the processor to perform more calculations. The first Intel microprocessor, named Intel 4004 and presented in 1971, had about two thousand transistors. Some of today’s chips have tens of billions of transistors.

As semiconductor companies shrink chips, they put more transistors on them. The size of the transistor indicates the level of chip lithography technology. In this way, the smaller it is, the more powerful and less power-consuming the chip will be. Intel’s 4004 chip used a 10 micrometer (10,000 nm) manufacturing process, while today’s latest TSMC chips use highly advanced 3 nm lithography. The latest Intel products have 7nm lithography.

Engineers at semiconductor companies prided themselves on designing smaller processors every year. One of the managers who worked at Intel in the 1980s said that lithography engineers are the crown jewels of this company. The prediction of Gordon Moore, one of the founders of Intel in the 1960s, which later became known as Moore’s Law, has been a cornerstone of the technology world for decades. This law says that the density of transistors on chips, and thus their computing power, doubles roughly every two years, and their prices become cheaper.

Moore’s Law means that Intel’s software partners such as Microsoft can count on the next generation of PCs or servers being more powerful than the current generation. This rule worked for Intel for almost 50 years. The expectations for continued development in the blue team were so high that they even gave it a nickname and called it Tic Tactic Development. Every two years, Intel released a new chip with a new lithography (tick) and improved its technology and design a little the following year (tick).

The way was opened for further delays at Intel. This company missed its deadline for the next process, i.e. 7nm lithography, and ate it after the publication of the 2020 financial reports. Intel’s stock plummeted as the year’s earnings slump became apparent, paving the way for the former CEO to be ousted and replaced by Gelsinger, who had previously been an engineer at the company.

As Intel struggled to maintain its legendary pace of growth, one of its historical rivals in the PC and server chip market, AMD, made the most of the opportunity. AMD is more of a chip designer than a chip maker. The company designs its chips in California, and then manufacturers such as TSMC or GlobalFoundries produce them for it. Because TSMC, unlike Intel, did not face any particular problem when transitioning to 10 and 7 nm processes, AMD chips reached the level of Intel’s products in the second half of the 2010s and even surpassed the rival team’s products in some specific applications.

AMD, which had a small share of the server processor market a decade ago, has now begun to take over Intel’s share in the industry. The red team will make up more than 20% of all server processors sold in 2022, and according to last year’s estimate by Counterpoint, sales will grow by a whopping 62% that year. Finally, at the end of that year, AMD’s market share was higher than Intel’s.

Graphical processing units, or GPUs, were originally designed to run heavy computer games, but computer scientists realized that these products were also ideal for running parallel computations of artificial intelligence algorithms.

After launching the controversial chatbot ChatGPT in 2022, Nvidia managed to triple its sales in just one year. Expensive server chips have once again caught the attention of companies, and tech giants have started developing expensive servers again. Today’s AI GPU-based servers sometimes pair up to eight Nvidia GPUs with just one Intel processor. So while in old servers, the Intel processor was almost always the most expensive and most important part of the system, now Nvidia chips are the most expensive equipment in GPU-based servers.

Jensen Huang, the CEO of Nvidia introduced the company’s graphics processors

Apart from an AI chip called Gaudi 3, Intel doesn’t have a product in the pipeline to compete with Nvidia’s AI accelerators. In 2018, Blue Team acquired Habana Labs to make its technology the basis for Gaudi chips, thus beginning a serious focus on AI for servers. Gaudi 3 is produced with 5nm lithography, which Intel has not yet reached, and relies on a foreign manufacturer for its production.

Intel expects to pocket $500 million from sales of its Gaudi 3 chips this year, mostly in the second half. By comparison, AMD expects to make $3.5 billion a year from its AI chip. Analysts polled by FactSet expect Nvidia’s AI GPU business to generate $57 billion in revenue for the green team in the second half of this year.

However, Intel has not stopped trying and has recently started looking for a new story in the world of artificial intelligence. If Intel succeeds, it could replace TSMC for other companies and become the US maker of smart chips even for Nvidia. The US government is subsidizing Intel’s giant factory in Columbus, Ohio, as part of an $8.5 billion package of loans and grants to make chips on American soil. In a conversation with reporters a month ago, Gelsinger announced that the Ohio factory will offer leading products to the market in 2028 and will even produce artificial intelligence chips from other companies.

Since 2021, when Gelsinger took the helm, Intel has been grappling with the fallout from past failures and has been working hard to catch up to TSMC through a roadmap it calls “four ninety lithographies in five years.” . Achieving such a goal is not easy. Gelsinger considers the responsibility of leading Intel to regain market leadership so heavy that he once referred to 2022 as a “death march.”

Gelsinger says the final destination is still two years away, but it seems the death march has begun and Intel has reached its first stop. Until then, TSMC will also release its 2nm chips. Intel has announced that by 2025, it will launch its chips based on 18 Angstrom technology, which is equivalent to 1.8 nm.

Pat Gelsinger introducing the company’s new lithography

Intel

Traveling this way has not been cheap for the American giant. Blue Team’s sales reached $4.4 billion, mainly in the US market, but it had an operating loss of $2.5 billion. This number reflects Intel’s huge investment in advanced chip manufacturing facilities and infrastructure.

“We have a lot of these investments in the pipeline and we want to get them through the temporary loss write-down,” Gelsinger told CNBC. This year’s operating loss is the lowest we consider.” Although many companies have not yet officially applied for Intel products, the blue team has so far signed contracts worth $15 billion with foreign companies to produce chips. One of these companies is Microsoft, which has announced that it will use Intel’s future technology to produce its server chips.

If Intel takes the lead again in making the world’s smallest transistors, what Gelsinger dreams of will happen and the blue giant will reawaken. He recently announced that demand for server chips based on Intel’s 3nm technology has picked up this year, and customers who were attracted by competing companies are gradually returning to Intel’s side.

“We’re regaining the trust of our customers,” says Gelsinger. They’re looking at us now and they’re saying, ‘Look, Intel is back!’