The Firms Behind the World’s Most Critical Technology

As with any investment, your capital is at risk.

In 1985, Taiwan effectively gave Morris Chang, an American businessman of Chinese birth, a blank cheque to establish the island’s semiconductor industry.

He bet on a revolutionary idea, a business dedicated to building others’ chips, not its own.

Today, Scottish Mortgage has 10 per cent of its portfolio invested in the semiconductor industry, and its holding TSMC is one of the world's most valuable firms. Several other portfolio companies, including NVIDIA, Tesla and Amazon, depend on it, while another investment, ASML, is one of its closest partners.

As part of Scottish Mortgage’s inaugural digital conference, Chris Miller spoke about the semiconductor chip industry’s development and the economic, geopolitical and technological forces that continue to shape it.

Miller is associate professor of international history at the Fletcher School of Law and Diplomacy at Tufts University and the author of the 2022 book Chip War: The Fight for the World’s Most Critical Technology.

TSMC: from zero to hero

Miller describes TSMC as “indispensable and the world’s most important chipmaker”. Before its founding, it was the norm for companies to design and physically produce their own chips. The few who didn’t had to risk exposing their intellectual property to a rival.

However, Chang realised that “specialisation was key” to creating the efficiencies that would counteract the increased cost of making the ever-smaller transistors (tiny on-and-off switches) required to make more complex chips. So, he focused on pioneering the foundry – a factory dedicated to manufacturing others’ designs.

That paid off. Today, TSMC is one of the world’s 10 most valuable listed companies, and it is critical to progress in smartphones, electric vehicles, AI and more. Only Samsung has been able to keep up with making the most advanced architectures, but it lags far behind in market share.

TSMC recently said it produces 99 per cent of the chips used to train artificial intelligence, known as AI accelerators or graphics processing units (GPUs).

“[That’s] about as close to the definition of monopoly as you can get,” says Miller.

“It's striking that some of the commentary from Wall Street was confusion about why TSMC hadn't hiked prices more. But TSMC's leadership would say they're not thinking about this contract or the next.

“They're thinking about relationships that have already lasted decades and that they hope will last decades into the future. That requires a level of trust in manufacturing and, I think, trust in pricing with their customers.”

From the Cold War to national priority

National interests have been central to the semiconductor industry’s evolution. Miller’s initial interest in semiconductors was sparked by “their central position in competition between countries for national power”.

Some of the US’s first commercially produced chips were used in nuclear missiles’ guidance systems during the Cold War against Russia. Today, China’s access to high-end technology preoccupies much of the US’s defence thinking.

China is the largest importer of semiconductors, spending more money each year on them than on oil over the past decade, and the US has been trying to limit its access to the most advanced technology.

“Both China and the US have identified artificial intelligence as… a general-purpose technology that each government believes will have broad economic, technological and strategic ramifications.

“Indeed, we already see both governments beginning to use AI, … to sift through intelligence that their spy satellites are collecting, … or to guide drones more accurately as they fly through the air.

"Everyone has access to algorithms that are largely open-sourced. Talent disperses widely across the world. But computing power is produced by a tiny number of companies.”

The US has prevented TSMC from building state-of-the-art chips for Chinese clients, including limiting sales to Chinese customers, such as Huawei.

“But that only underscores the extent to which TSMC is a company with unique capabilities that no one, in China or anywhere else, can match,” says Miller

ASML: betting on the long term

No country or company can do it alone, but one company may come close to being indispensable.

Today, the most complex chip-making tools are photolithography machines. The Dutch firm ASML produces the most advanced equipment capable of “carving the most intricate patterns in silicon.” It has a 100 per cent market share of the leading equipment and harnesses a complex supply chain to do so.

Miller puts its dominance down to management’s long-term mindset. In the 1990s and 2000s, the company made a strategic “bet on producing the next-generation [extreme ultraviolet lithography (EUV)] tool that today is central for advanced chipmaking.”

The company persevered despite EUV being seen “as harder than bringing a man to the moon.”

According to Miller, the machines “require the flattest mirrors humans have ever made and one of the strongest lasers ever deployed in a commercial device”.

Inside every machine, “a tiny ball of tin is pulverised into a plasma. The temperature of this explosion is 40 times hotter than the sun's surface.

“Humans have made no tools more complex than these. You can understand why it seemed highly implausible for a long time that a machine with an explosion like this inside would be usable in a factory context.”

ASML succeeded. Its latest machines cost over $300m per machine and contain hundreds of thousands of components; each must be extraordinarily reliable and precise. Misplacing even a single atom can disrupt the process when manufacturing at the nanometre (billionths of a metre) scale.

Miller suggests it will be almost impossible for anyone to catch up with the company in the foreseeable future.

“By the time you’ve learned to replicate it, ASML will have already produced its next-generation tool, and so [its competitors] will still be behind.”

NVIDIA: all-in on GPUs

Miller views technology leadership as critical for establishing longevity: "Ultimately, the only way you’ve got a defensible market position is by having better technology than your competitors.”

NVIDIA has done this through the foresight of its chief executive, Jensen Huang, who invested heavily in the developing technology and software now critical to AI models.

The first large-scale generative AI tool to capture widespread public attention was OpenAI’s generative AI chatbot, ChatGPT. About a month after Chip War’s publication, ChatGPT launched, and demand for NVIDIA's graphics chips soared as others rushed to build and train their own AI models.

But over a decade prior, Huang had realised his firm’s graphics processing unit (GPU) chips could be used for more than just rendering computer graphics. He understood that their parallel processing capabilities could be used to solve complex scientific problems and accelerate computational tasks.

His company invested deeply in building software tools called CUDA to make GPU programming more accessible to developers. When AI researchers adopted the technology, NVIDIA seized the opportunity, giving CUDA extra functionality and optimising the GPUs’ architectures to help.

At the time, AI was seen as “far out and implausible”, at least as a business for NVIDIA.

Miller highlights Wall Street’s discomfort on: “first, building chips for an industry that didn't exist, and second, giving away the software for free.”

Huang understood that repurposing NVIDIA’s chips would make them “vastly more efficient than the existing class of processors, called CPUs (central processing units)”.

“NVIDIA bet that if it could make its chips efficient for AI, it would drive down the costs of AI and, therefore, blow up in a much larger market than anyone else foresaw.”

Miller contrasts that with Intel, which was once the world’s largest chipmaker and still plays a dominant role in the design and manufacture of the CPUs that are still important for personal computing and data centres.

He points out that it “flirted” with AI in various formats but “could never really convince itself to dive in in a way that Jensen bet the company on AI being its future.

“He deserves much credit for NVIDIA's decisions,” says Miller of one of the Fortune 500’s longest-serving chief executives.

“Significant bets were taken, not based on short-term profit, but instead on long-term technological shifts that, it's now quite clear, NVIDIA bet correctly.”

Patience is a virtue

It’s a reminder that technology moves fast, but perhaps not as quickly or clearly, nor always in the same direction, as we might anticipate.

As Miller highlights, companies need to determine “the proper use case, the correct form factor and the suitable business model” and that can take decades.

He points out that it was a decade from the invention of the first transistor to the first integrated circuit. Then, it took another decade for mainframe computers to become widespread in large corporations. Two decades later came the first personal computer, and another couple of decades later, the first smartphone.

“Each of these steps was ‘transformative’ and produced companies that were among the most valuable in the world at the time,” Miller says, suggesting that AI still has a long road ahead.

"We have yet to conceptualise [its] application in many spheres because we’re in the very earliest stages of a technological shift that will take decades to play out.”

Progress will not be a straight line, but we can be confident that the advances in artificial intelligence and other emerging industries, such as space travel and the energy transition, will demand more advanced chip designs and complex chip manufacturing.

For that to happen, companies need access to capital from patient investors prepared to back visionary companies and leaders through the ups and downs of change. That is Scottish Mortgage’s forte.

Risk factors

The trust invests in overseas securities. Changes in the rates of exchange may also cause the value of your investment (and any income it may pay) to go down or up.

The Trust invests in emerging markets, which includes China, where difficulties with market volatility, political and economic instability including the risk of market shutdown, trading, liquidity, settlement, corporate governance, regulation, legislation and taxation could arise, resulting in a negative impact on the value of your investment.