The end often comes “gradually, then suddenly”. In Ernest Hemingway’s 1926 novel The Son Also Rises, a protagonist uses the phrase to describe two ways in which he went bankrupt. From literature to business to just about every aspect of human relationships, change tends to take place slowly at first and then all at once.
Over the past five years or so, what was one of the world’s best-known tech icons has been on a slippery slope, gradually losing market share and mindshare. For years, there was still hope that with a makeover, the Silicon Valley pioneer might re-emerge like a phoenix. But Intel, whose microprocessors are found in 75% of all personal computers, or PCs, around the world, now appears to be in its final days as a standalone firm.
Within the next several weeks, Intel is likely to be broken up into at least two, possibly even three or more pieces, as the world’s largest chip-making firm or foundry operator, Taiwan Semiconductor Manufacturing Co (TSMC) and the world’s number two chip design firm Broadcom divide its assets up. TSMC is seeking a deal to control Intel’s foundry business, while Broadcom is seeking to grab control of Intel’s chip design and marketing business, including its microprocessor CPUs, or central processing units that are the brains of the PC that you use, or the server that your company uses to store its data or power the cloud computing data centers. The Donald Trump administration is helping broker the two unconnected deals to help keep America’s lead in the global arms race to design and manufacture the highest-performance chips within its own borders.
Ironically, Intel was once the dominant designer and manufacturer of chips that powered PCs, servers and data centres for cloud computing. It was one of the dynamic duo dubbed “WinTel”, with Microsoft’s operating software Windows being the other that launched the PC era four decades ago. The original WinTel relationship was based on an instruction set architecture known as “x86”. Intel patented the instructions for operating PC hardware, and Microsoft wrote the software that ran exclusively on those instructions.
Not satisfied with being a mere widget-maker, Intel spent billions on an “Intel Inside” corporate branding campaign to differentiate itself from commodity memory chipmakers in East Asia, even though it did not sell to end users.
I wrote about the gradual decline of Intel back in my column in late 2020 when chip giant Nvidia Corp was almost running away with the crown that Intel had worn for decades. Nvidia stock has been up nearly 1,000% since, while Intel stock has been down about 50% despite a 35% rebound over the past two weeks. At the time, my sense was that Intel could be saved even though it would be a shadow of its former self.
See also: Samsung broadens AI phone lineup with US$300 5G Galaxy models
In March 2021, Intel’s board brought back Pat Gelsinger, a former senior executive, to right the ship and restore it to its former glory. The idea was to resurrect its manufacturing operations and strengthen its chip design business to compete head-on with the likes of Nvidia, Broadcom, Advanced Micro Devices and AMD. Trouble is bringing back Gelsinger was too little, too late. Intel also needed more money and more time to get into shape. In December, the Intel board fired Gelsinger, and since then, the company has been “in play,” a euphemism for becoming a takeover target.
Intel actually “invented” the global semiconductor industry. Journalist Don Hoefler, who coined the term “Silicon Valley” in a 1971 story about chip firms in the San Francisco Bay area, referred to Intel as the dominant user of silicon wafers whose business model at the time was being copied by an array of lookalikes. Started by engineers Gordon Moore and Robert Noyce in 1968, it became the predominant chipmaker and an early driver of the US tech industry. The chip industry’s famous “Moore’s Law”, which states that a chip doubles the number of transistors — as well as its power efficiency and performance — every two years, is named after Intel’s co-founder. Long-time CEO Andy Grove, who took Intel to the next level in the 1990s, wrote about how the then-chip giant kept its edge. He titled his book after his motto at chipmaker: “Only the paranoid survive”.
To understand how Intel got to the edge of a precipice, look no further than the transformation of the global chip industry. Back in the day when Intel was the King of chips, the bumper sticker in the chip industry said: “Real men have their own fabs.” If a chip company did not have its own fab, it wasn’t a real chip company. A “fab” is a wafer fabrication plant where sophisticated chips are manufactured. Three decades ago, it took one or two billion US dollars to build a fab. These days, a big fab costs US$25 billion ($33.4 billion). Yes, US$25 billion to build just one factory.
See also: Release of Microsoft’s Majorana 1 chip shows quantum computers are ‘years, not decades’ away
As the cost of building massive factories to make state-of-the-art chips went through the roof, the business model of the semiconductor industry evolved into those with fabs and those that were “fabless” or without plants, essentially chip design companies that outsourced manufacturing to companies like TSMC that did not design chips but only made chips under contract. Nvidia, Broadcom and Qualcomm are companies that have never had any factories. TSMC, South Korea’s Samsung Electronics, Taiwan’s United Microelectronics and US-based Global Foundries, which bought Singapore’s Chartered Semiconductors and China’s Semiconductor Manufacturing International Corporation (SMIC), are contract manufacturing foundries that make chips for specialised design firms like Nvidia.
Intel started as a vertically integrated chipmaker that not only designed its chips but also had factories to make its own chips. The integrated model was the secret sauce of its huge profit streams. However, as chips became more complex, they became harder to make for companies that were doing too many other things besides manufacturing them. Pure play foundries like TSMC had an edge, which has continued to widen over the past decade.
Intel has had a long history of manufacturing challenges. It encountered problems when making 14-nanometre (nm) chips in 2014. By the way, circuit widths on chips are measured in nanometres, which are one-billionth of a metre. The smaller the width of the circuit, the better. When Intel moved to 10nm chips in 2017, manufacturing issues reportedly became even more acute. Then, in 2020, at the start of the Covid-19 pandemic, Intel announced further delay in the rollout of 7nm chips. Rival TSMC was already on nm chips by then. TSM later moved to 3nm chips and is currently transitioning to 2nm chips. By late 2027, the Taiwanese chipmaker expects to be rolling out its first 1.4nm chips. China’s top foundry, SMIC, makes 7nm chips because it does not have access to Dutch chip equipment giant ASML’s sophisticated tools and machinery due to a Western ban on exports.
So, who will buy what as Intel is dismantled? For starters, Intel has two subsidiaries to sell before it starts to auction off the family jewels — its foundries and its core CPU chip design and marketing unit. Altera, a designer of programmable logic devices and Mobileye, which develops vision-based, self-driving cars and advanced driver-assistance systems providing warnings for collision prevention and mitigation, will be the first assets to be sold. Intel purchased Altera in 2015 for US$16.7 billion and paid US$15.3 billion for Israel-based Mobileye in 2017. It is reportedly looking to sell Altera to a private equity firm, Silver Lake Management, for US$9 billion. Or a US$7.7 billion loss. Intel re-listed Mobileye on the Nasdaq in 2023 and currently has an 88% stake in the firm. Based on Mobileye’s current share price, Intel’s stake in the firm is valued at US$12.9 billion, or a loss of US$2.4 billion over eight years.
In September, cellular chipmaker Qualcomm notified the Intel board that it was interested in an acquisition. Unfortunately, it wasn’t a serious bid, and Qualcomm has since pulled out of the race. That leaves two of the three largest chip firms on Earth: manufacturing behemoth TSMC and chip design giant Broadcom, which also invests in software firms. Top chip firm Nvidia remains focused on AI chips and, as such, has no interest in Intel.
For its part, TSMC had initially ruled out a deal with Intel, but after Gelsinger’s sacking and President Trump’s return to the White House, it has been reportedly roped in to help Intel’s foundries. TSMC is reportedly interested in running Intel’s foundries if it can get a controlling stake in the manufacturing unit. Some of Trump’s advisers want the best foundry operator on earth to run US foundries. Others say a foreign company like TSMC should not be allowed to gain control of the foundries or allowed to run them with a minority stake. Their view is that only US companies should be allowed to run US foundries.
Analysts expect a lot of haggling before a final deal is hammered out. Former President Joe Biden’s US Chips and Science Act allocated US$52 billion to help the American chip industry expand its domestic footprint. The biggest beneficiary was Intel, which announced a US$100 billion investment in chip fabs in Chandler, Arizona and Columbus, Ohio. It has already received US$7.9 billion. TSMC, which recently opened a large fab outside Phoenix, Arizona, also benefitted from the Chips Act receiving US$6.6 billion, while Samsung Electronics, which is building its own fab, received US$6.3 billion from the US government, which wants to beef up local chip-making capabilities.
Sink your teeth into in-depth insights from our contributors, and dive into financial and economic trends
So, how much will Intel raise for the sale? Pierre Ferragu, hardware analyst at New Street Research, estimates that Broadcom could pay anything between US$135 billion to US$200 billion for Intel’s core x86 business. Can Broadcom afford to pay for that kind of dough? “Broadcom applying its usual recipe for success for mature franchises; focus on core products, over-invest in R&D and product roadmaps, cut everything else,” would be the formula to make it work, Ferragu argues. “The deal would be a small moving part for Broadcom’s US$1.1 trillion market cap”, he notes. “Its balance sheet can support an additional US$80 billion in debt, while the remaining equity portion would dilute shareholders only by 5% to 10%.”
Ferragu believes Intel’s foundries could fetch anything between US$50 billion to US$100 billion. Intel Foundries has US$25 billion in cash on its balance sheet. Whether it gets a 50.1% stake or a 49.9% stake, TSMC will need Trump’s White House to give it a free hand in building a decent AI chip manufacturing business in the US. Even if you take the low end of the range, Intel shareholders could get US$310 billion from the break-up of the firm. That’s US$44 per share Intel shareholders split between some cash, some shares of Broadcom, some shares of Mobileye and Altera, and shares in a recapitalised Intel foundry, the New Street analyst notes. Not everyone likes it. “What’s taking place now is ‘a fire sale’, not unlocking of shareholder value at Intel,” Intel’s former lead silicon engineer and legendary chip designer Jim Keller noted in a tweet over X (formerly Twitter). Keller believes Intel assets are probably worth close to US$1 trillion. With the Silicon Valley pioneer in deep trouble getting US$44 for a stock that was trading at just over US$18 two months ago, it looks like a bargain.
Assif Shameen is a technology and business writer based in North America
