Intel's share went to the White House: what do U.S. officials fear losing?

On Friday, August 22, Intel officially confirmed that the U.S. government received 9.9% of the company and became its largest shareholder. Measures of this kind are usually taken in crisis situations, when the collapse threatens entire industries. This was the case, for example, with General Motors and Chrysler in 2009 - then the Obama administration bought their shares to save 1.9 million jobs. Intel employs 96,000 people and will have 75,000 by the end of the year - a significant reduction, but not a national disaster. Yes, the computer giant has been making serious losses lately, but it has already raised $2 billion from Softbank and is now looking for other big investors. The point seems to be different: Intel is virtually the only U.S. company capable of producing state-of-the-art computer chips, and this production is exactly what the government does not want to lose.

To understand the essence of the problem, let's go back more than 30 years and thousands of kilometers to the east, to the island of Taiwan. In 1987, Morris Chan, an American engineer of Chinese descent, founded the Taiwan Semiconductor Manufacturing Company (TSMC) there. Chan was 56 years old at the time and had 25 years of experience at Texas Instruments, one of the largest technology companies of the time, where he had risen from a mere engineer to senior vice president, leading the global semiconductor business.

He was promoted to CEO, but something went wrong. Instead of the top job, he was put in charge of consumer product sales, which Chan, an engineer to the core, didn't like, according to the authors of the technology podcast Acquired, who compiled a detailed history of Chan and TSMC. So he left Texas Instruments and eventually accepted an offer from the Taiwanese government, which was eager to launch high-tech computer chip manufacturing on the island.

In general, it could have been something like a dusty retirement position for him, but Chan had an idea. At that time, it was believed that every self-respecting creator of computer technology should have its own production of chips. However, this production is very complex and expensive, even in the 80's to launch it required hundreds of millions of dollars. Many promising startups never took off, because the founders could not raise so much money. While still working at Texas Instruments, Chan saw similar situations more than once.

And he came up with the idea of creating a factory without a product, that is, producing chips exclusively to order. To put it briefly, the idea worked. It was very convenient for technology entrepreneurs to design new chips (which is relatively inexpensive) and then order their production from Paul.

In particular, if it weren't for Chan, we might not even have the current investor favorite, Nvidia, a company with a capitalization of over $4 trillion. In 1997, Nvidia founder Jensen Huang sent a paper (can you believe it?) letter to Chan in Taiwan, asking for a meeting because TSMC's U.S. office wasn't making contact, considering Huang's startup too insignificant a client. Chang was interested and agreed to the meeting.

"He told me that Nvidia was struggling financially, but the chip he now wanted to order would not only save the company, but would make Nvidia a major customer of TSMC. That was a pretty bold statement. Our turnover was over a billion dollars, and to become a major customer, he needed to generate at least $50 million a year in revenue for us... It was a gaming (graphics) chip. Of course, it turned out to be successful. His prediction came true," Chan recalled.

The partnership continues to this day. When we read that Nvidia "made" a new ultrafast chip for AI, it means that the company designed it, but in fact its products are made by TSMC factories. It's a similar situation with other major computer manufacturers - AMD, Apple, Broadcom, Qualcomm and many other tech industry leaders entrust the production of their chips to a company founded by Chan.

By the end of 2024, TSMC earned a net profit of $36.52 billion (plus 36% YoY) on consolidated revenues of $90.08 billion (plus 30%). The company's capitalization exceeded $1.2 trillion, and 94-year-old Morris Chan is a billionaire with a fortune exceeding $5 billion, according to Forbes.

But that's not what's most important - according to TrendForce, TSMC accounted for nearly two-thirds of global chip production in Q4 2024, with its nearest competitor, South Korea's Samsung, lagging far behind with 9.1% of the market.

The problem is that China considers the island its integral part. And there are periodic fears that Beijing will try to seize Taiwan by force, and the"cold technological war" in the form of various restrictions and duties between the United States and China could turn into a hot phase. And if at the same time South Korea, which, as we remember, has a very restless northern neighbor, also goes up in flames? Overnight, the world could lose three quarters of its supply of chips for smartphones, computers, servers, cars, airplanes, and everything else.

Bloomberg even calculated what China's attack on Taiwan and subsequent retaliatory measures could cost the world economy. It turned out that Taiwan would lose about 40% of GDP, China - 16.7%, the U.S. - 6.7%, and the world as a whole - about $10 trillion, which is more than 10% of global GDP. In economic terms, it will be such a bomb, compared to which the covid crisis (remember the disruption in the supply of new cars due to a shortage of computer chips?) will seem like a child's matinee.

Naturally, the US realizes this and would like to back it up.

Back in 2022, U.S. President Joe Biden signed the CHIPS and Science Act, a law that provides, among other things, government grants and tax incentives for the development of advanced chip manufacturing within the country.

At the end of 2024, Intel, essentially the only U.S. company to retain advanced chip manufacturing capacity, was guaranteed $7.86 billion in payments under the act. "With the Intel 3 already in mass production and the expected release of the Intel 18A next year, advanced semiconductors are once again being made on American soil," Pat Gelsinger, Intel's CEO, cheerfully reported on November 27, 2024.

And on December 1st, he was fired. Because producing and selling are not the same thing.

Intel's problems did not start yesterday and are well known. The company missed out on the smartphone chip market by canceling its contract with Apple, Chan, who managed to get that customer for TSMC, recalled in an interview. By focusing on CPUs, the company left the graphics card market to competitors like Nvidia, and when it turned out that it was GPUs that were best suited for AI, Nvidia took off and Intel was left out again.

Eventually, the company began to lose ground in its core market as well. "Back in 2019, Intel controlled 84% of the global PC chip market and 94% of the server market. By 2024, those numbers had fallen to 69% and 62%, respectively. AMD, using the x86 architecture pioneered by Intel, has created better chips. Cloud giants such as Amazon, Google and Microsoft, which once depended on Intel, are now developing their own processors using the work of Arm, a British company owned by Japan's SoftBank. In December, Amazon said that half of the server capacity it had added in the previous two years was using its own processors," The Economist writes.

"Investors have naturally been disappointed. Over the past five years, Intel's stock has fallen about 65%, while AMD's stock has more than doubled, TSMC's nearly quadrupled, and Nvidia's more than 16 times," the WSJ completes the picture.

Intel's annual report for 2024 was signed by new CEO Lip-Bu Tan, and there was little good news for investors. The company as a whole made a net loss of $18.8 billion, while the manufacturing division made an operating loss of $13.4 billion.

Not surprisingly, many people began to ask a logical question - is it necessary for the company to preserve the architecture of the 80s, combining chip design and production? The main competitor AMD, for example, separated and sold its manufacturing division back in 2008, and it seems to have benefited the company - its capitalization is now about 2.7 times higher than that of Intel with all its advanced but unprofitable chip factories.

Gelsinger, who has spent about $90 billion on these factories, has tried to make them more profitable by bringing in contract manufacturing from other companies, but has not succeeded, The Economist notes.

According to the annual report, out of $17.5 billion revenue of Intel's manufacturing unit, third-party orders brought only $385 million. Success is hindered by two fundamental problems: first, Intel factories are "sharpened" to work primarily with the design departments of the company itself and have almost no experience in working with third-party customers.

Secondly, how likely is it that, for example, the same AMD or Nvidia will order their chips from a direct competitor, asks WSJ.

Apparently guided by this logic, Intel Chairman Frank Yeary, according to the WSJ, earlier this year was working on a plan to get Intel out of the manufacturing business entirely and sell it to a consortium of investors or even TSMC.

That's where he clashed with new CEO Lip-Boo Tan, who argues that Intel's manufacturing business is integral to its success and is necessary to keep the U.S. from becoming dependent on foreign semiconductor companies such as TSMC and Samsung, the newspaper's sources said. While TSMC and Samsung have committed to building more factories in the U.S., critics argue that their advanced research and development work will remain concentrated elsewhere, the newspaper said.

Perhaps the U.S. government's sudden decision to convert CHIPS funds earmarked for Intel into a 9.9% equity stake is political, and aimed precisely at preventing the division of a strategically important company for the country.

The statement said the government will not interfere in management and will vote in solidarity with the board of directors, but "with certain exceptions". What those exceptions are is not specified.

In addition, the government reserves a five-year right to acquire another 5% of the company at $20 per share (a discount of about 20% to the current market value) if Intel loses 51% of its manufacturing business. Apparently, this is another "stopper" against a possible separation.

So Lip-Bu Tan will have to solve a very difficult problem: how to get Intel out of the hole without using the obvious business decision to split into two parts. He has already said in his letter that the company will now approach the development of new chips and the building of new fabs very cautiously, investing only when there is guaranteed demand. "Going forward, our investments will be based on confirmed customer commitments. There will be no more unconditional financing. Each investment must be economically justified," he wrote.

The only thing left is to find these customers and then persuade them to switch from decades-old TSMC to Intel.

"Instead of his predecessor's thesis, 'Build and they will come,' Tan is essentially saying, 'We'll build if they come.' Now the question is who will come," summarizes Barron's in a piece eloquently titled "Intel has a lot of problems. The government's share won't help."

And here's where we don't know. If U.S. President Donald Trump, already famous for his unorthodox methods of "manual control" of major technology companies, asks them to listen to Intel's proposals, it might work. However, Tulane University professor Walter Isaacson calls such methods "crony capitalism" and argues that they will only breed corruption and will not lead to a revival of American manufacturing.

TSMC, by the way, to which the White House promised to allocate $6.6 billion under the CHIPS Act, prepared to return this money if they demanded to give up their shares in return. However, a White House spokesman assured Reuters that no such measures are planned for the Taiwanese company, which has already promised to invest $165 billion in its U.S. plants.

This article was AI-translated and verified by a human editor