SPECIAL FEATURE
w.media Editor Jan Yong interviews Robert Quinn on his analysis of the global chip ecosystem with an emphasis on Asia. Robert is an expert in the semiconductor industry, with manufacturing experience spanning Applied Materials, Samsung, Intel, Texas Instruments, and GlobalFoundries.
- With AI scaling up exponentially, how is this impacting the semiconductor industry? Can chips catch up with the speed of AI advancement?
I’ve spent considerable time studying this AI industrial revolution that’s racing towards us like a tsunami. In the United States, we’re building data centers across the country at an unprecedented pace. Currently, we have the capacity to produce the chips, the GPUs, and the servers. We even have the electricity generation capability. But here’s what might surprise people: the biggest bottleneck to the industry right now isn’t silicon. It’s transmission lines. We simply cannot build electrical transmission infrastructure fast enough to deliver power to these data centers.
One promising solution emerging is SMRs: small modular nuclear reactors. I recently toured Texas A&M University, which has partnered with Last Energy to build a small nuclear reactor for testing purposes. Historically, one of the biggest obstacles to nuclear deployment has been the lengthy federal approval process. However, I believe the federal government is accelerating the pathway for small nuclear reactors across the United States. We’re already seeing this with projects like Texas A&M’s partnership with Last Energy, which was selected for the Department of Energy’s New Reactor Pilot Program and is moving through authorization on an accelerated timeline, with testing targeted for summer 2026.
The second part of this equation involves how AI fundamentally changes the hardware requirements for our IoT devices. We are going to need to upgrade virtually everything: our phones, our computers, our laptops, our tablets, our watches. All of these devices will require AI edge inferencing hardware built directly into them. This is where I see a powerful second wave of chip demand emerging from the advent of artificial intelligence. The first wave is data center buildout. The second wave is the complete refresh of consumer and enterprise edge devices.
“The first wave is data center buildout. The second wave is the complete refresh of consumer and enterprise edge devices.”
- How will geopolitics affect the supply chain?
With the current administration in the United States, I do not anticipate dramatic shifts in geopolitical dynamics. We will continue to see countries engage in erratic posturing: pushing back against tariffs, raising tariffs, dropping tariffs. It’s a constantly shifting landscape.
What most people don’t realize is how remarkably delicate the semiconductor supply chain truly is. The cell phone you carry in your pocket has crossed 12 intercontinental borders before it ever reaches your hands at the store. Consider the journey: minerals mined in Africa, rare earth elements extracted from other regions of the world, smelting operations conducted in China, board processing work completed in Malaysia, intellectual property originating from the United States, and EUV lithography equipment manufactured in Europe.
This industry is truly a collaboration of the entire world. I recently wrote a piece called “The Three-Headed Dragon,” examining Asia, the Americas, and Europe. We are that three-headed dragon, and we share a single heart. If one were to sever the head of any dragon, we would all perish. The interdependence of the global semiconductor ecosystem is far more fragile than most realize, and companies are finding it increasingly challenging to build sustainable chip supply chains.
Just-in-time (JIT) delivery is dead. We learned that painful lesson during COVID. Today, there are emerging solutions for building sustainable supply chains, including purchasing and storing chips in semiconductor vaults held within free trade zones. It’s no longer a matter of JIT. It’s a matter of JIC: just-in-case. It takes only one pandemic, one geopolitical event, or one natural disaster to bring the entire house of cards crashing down.
“The [semiconductor] industry is truly a collaboration of the entire world.”
- Due to the US-China AI race, will Southeast Asia reap the greatest benefit in terms of semiconductor companies relocating there or parts of the work being done there? Among SEA nations, which ones are best placed to scale up higher on the supply chain?
To understand the chip industry, you must first understand that chips are like apples and oranges. There are vastly different types of chips serving vastly different purposes.
You have 256-nanometer chips that quietly run the world. You have 3-nanometer chips that power cell phones and data centers. The chips that truly keep our modern civilization functioning are legacy node chips built on older architectures, and the majority of these are manufactured in Asia. I personally don’t see this manufacturing supply chain relocating to the United States, nor should it.
Where the U.S. is focusing its efforts is on EUV lithography and advanced small-node chips that power artificial intelligence, smartphones, and other high-end devices. These chips are extraordinarily difficult to manufacture. It costs billions of dollars to build a single factory capable of producing them. Yields remain challenging, and only a handful of companies in the world possess the capability to manufacture them.
I believe we will see both sectors growing in parallel. For legacy node production, I expect countries like India and other emerging markets to enter this sector and scale significantly. Legacy chips will continue to scale predominantly across Asia.
For advanced EUV chips, I expect significant growth in the United States, while Taiwan, South Korea, and Japan will continue expanding their EUV capabilities. Eventually, we may see EUV production scale in Europe as well.
But here’s the critical point: legacy chips are the chips that actually run the world. We will need far greater volumes of legacy node chips than advanced EUV chips. Both markets are essential, but the scale of legacy demand is often underestimated.
Robert Quinn, ‘semiconductor ambassador’
“Legacy chips are the chips that actually run the world. We will need far greater volumes of legacy node chips than advanced EUV chips.”
- Indonesia and Vietnam are ramping up their semiconductor industry by offering more incentives to investors. Singapore, despite higher operating costs, continues attracting foreign investment. Malaysia, currently one of the world’s largest ATP (Assembly, Test, and Packaging) hubs, wants to start designing chips as per their deal with Arm Holdings plc. Is this a preview of where SEA is heading in terms of moving up the supply chain?
Absolutely. What we’re witnessing across Southeast Asia is a deliberate and strategic ascent up the semiconductor value chain.
Indonesia and Vietnam recognize that incentives alone won’t differentiate them in the long term, so they’re investing in workforce development and infrastructure to attract not just assembly operations, but higher-value manufacturing. Singapore has always understood that its competitive advantage lies in intellectual capital, advanced R&D, and its position as a regional headquarter for global semiconductor companies. The higher operating costs are offset by stability, talent density, and access to capital.
Malaysia’s evolution is particularly fascinating. The country has been a powerhouse in assembly, test, and packaging for decades. The Arm partnership announced in March 2025, a US$ 250 million deal over 10 years for access to seven of ARM’s high-end chip design blueprints, signals a clear ambition to move beyond ATP into chip design and architecture. This marks the first time Arm has partnered with a national government rather than a private company for this type of IP access. Malaysia aims to establish 10 local chip companies with combined annual revenues reaching US$15 to US$20 billion. This is exactly the kind of strategic leap that transforms a country from a manufacturing destination into an innovation hub.
Southeast Asia is not simply benefiting from geopolitical tensions between the U.S. and China. The region is actively positioning itself to capture higher-margin activities across the semiconductor value chain. The nations that invest most aggressively in education, infrastructure, and ecosystem development will emerge as the winners in this next chapter of the industry.
“What we’re witnessing across Southeast Asia is a deliberate and strategic ascent up the semiconductor value chain.”
- With data centers consuming the bulk of chips (estimated 70 percent) and their relentless demand for more advanced chips, will this mean the chip industry is becoming increasingly dependent on AI’s growth?
Let me add some important context to that 70 per cent figure. That statistic specifically refers to memory chips, where data centers are projected to consume up to 70 per cent of global memory production by 2026. For the broader semiconductor market, data center chips are expected to account for over 50 per cent of the total semiconductor market by 2030, nearly doubling from current levels.
Consider this: Apple’s COO Sabih Khan announced at the Amkor groundbreaking in October that factories across the United States are producing 19 billion chips for Apple this year alone. That’s one company.
This brings us back to my earlier point: chips are apples and oranges. Yes, there are high-end chips powering the processors inside phones, the GPUs in data centers, and the CPUs in laptops. These advanced chips command premium prices and receive most of the media attention.
But then there are the chips that power your doorbell, your microwave, your dishwasher, the transmission in your car. These 140-nanometer and larger legacy chips run virtually everything in our daily lives. The demand for these chips is enormous, though the cost per unit is dramatically lower than data center and processor chips.
Both sectors will grow in parallel. AI is certainly driving significant demand for advanced chips, but the proliferation of connected devices, electric vehicles, industrial automation, and smart infrastructure, is driving equally significant demand for legacy chips. The industry isn’t becoming dependent on AI alone. It’s becoming essential to every aspect of modern life.
- Some say ASML’s EUV machines are the biggest structural bottleneck, as without them, no advanced chips can be manufactured. Your take? [ASML is the only company in the world that produces EUV lithography machines, which are required for manufacturing the most advanced chips (7nm, 5nm, 3nm, and below)]
ASML’s EUV lithography machines are indeed critical infrastructure for advanced chip manufacturing. The U.S. holds significant intellectual property stakes in EUV technology through early research partnerships at national laboratories in the 1990s, and uses export control pressure along with the Foreign Direct Product Rule to prevent this technology from reaching China and certain other regions. For those countries unable to access these machines, ASML absolutely represents a structural bottleneck.
However, for other parts of the world, different bottlenecks exist. In the United States, one of the most significant bottlenecks is actually packaging. The majority of chips manufactured in U.S. fabs are shipped directly back to Asia for packaging. Advanced node packaging, including 2.5D and 3D packaging technologies, is a completely different discipline than packaging older legacy node technology. The expertise and infrastructure for advanced packaging have historically been concentrated in Asia.
The U.S. is actively working to address this gap. Amkor Technology broke ground on a US$7 billion advanced packaging campus in Arizona in October 2025, which will be the first high-volume advanced packaging facility in the United States. This facility will complement TSMC’s (Taiwan Semiconductor Manufacturing Company Limited) nearby wafer fabrication, creating an end-to-end domestic supply chain.
Apple will be Amkor’s first and largest customer, with Nvidia also signed on. We’re also seeing new advanced packaging facilities being built by companies like SkyWater Technology in Florida, which has established a subsidiary there focused on 2.5D and 3D advanced packaging capabilities.
So while ASML is a bottleneck for some, the reality is that bottlenecks vary by region and by position in the supply chain. The industry is a complex ecosystem where multiple chokepoints exist simultaneously.
- Outlook for the semiconductor industry, especially in Asia.
I believe Asia will remain the dominant force in legacy chip production for the foreseeable future. The everyday chips that run our dishwashers, microwaves, refrigerators, doorbells, and countless other devices will continue to be manufactured predominantly in Asia, and demand will only grow as more devices become connected.
On the advanced node front, I expect Samsung to continue pushing the boundaries of EUV technology and emerge as an even more significant player in advanced chip production. Samsung has the scale, the technical capability, and the strategic commitment to compete at the leading edge.
Asia’s semiconductor future is not a story of either/or. It’s a story of both/and. The region will continue dominating legacy production while simultaneously competing for leadership in advanced nodes. The companies and countries that can execute across both segments will be best positioned for the decade ahead.