This article first appeared in The Edge Malaysia Weekly on June 17, 2024 - June 23, 2024

Artificial intelligence (AI) is a game changer that is exponentially enhancing human/machine capability, which raises dynamic competitive pressure on both the supply and demand side of market services and products. The market is digesting which hardware and software companies are likely to benefit most from the biggest revolution since the arrival of the internet. The largest US tech companies, Nvidia and Apple on the hardware side and Microsoft and Google on the software side, have become trillion-dollar companies since the arrival of ChatGPT.

Taiwan demonstrated its influence on the chip industry when almost all the top semiconductor and AI leaders flew to Taipei to attend Computex on June 4. Those present included Intel CEO Pat Gelsinger, Qualcomm President/CEO Cristiano Amon, Advanced Micro Devices (AMD) Chairman/CEO Lisa Su, Mediatek CEO Rick Tsai, NXP Semiconductors chief technology officer Lars Reger and Delta Research Center president Tzi-cker Chiueh. Nvidia CEO Jensen Huang, being the most valuable AI chip designer, hosted a pre-Computex conference, where he laid out how Nvidia accelerated computing by running specialised central processing unit (CPU) chips and graphics processing units (GPU) in parallel that can generate data at 100 times the current speed, increase energy needs by a factor of three times, but at an extra cost of only 50%. This speed and processing acceleration have enabled AI processing of Large Language Models and Generative AI applications to be adopted, ensuring a captive market of new AI models running on Nvidia chips, which explains the phenomenal rise in Nividia shares.

Su opened Computex by explaining how her company, known for Ryzen chips that gained a sterling reputation for gaming processors, now uses AI to improve chip speed, computing power and energy efficiency. The new Ryzen chips are ideal for machine learning and data centres that can deliver 50 trillion operations per second. Thus AMD, like Nvidia, is betting on the use of AI chips not only in the infrastructure (data centres), but also in a new generation of AI-based mobile devices, laptops, desktops, servers and electric vehicles. Qualcomm, the No 3 US fabless chip producer, has designed a Snapdragon chip second to Ryzen in speed, but consumes 22% less power. Intel, the largest in-house manufacturer, has also designed a Lunar Lake chip that has comparable functionality to AMD and Qualcomm.

Computex showcased AI personal computer (PC) chips by leading original equipment manufacturers such as Acer, Asus, Dell, Hewlett-Packard, Lenovo and Microsoft. For many years, laptops, PCs and server chips were based on Intel and AMD x86 architecture, whereas Apple, Qualcomm and Samsung used the UK-founded ARM’s architecture, which has driven most hand-held devices’ software platforms. These x86 processors are powerful, but they are energy hungry and generate more heat. Because the new generation of AI chips are effectively a powerful computer on a chip, they will disrupt the older x86-based PC market.

What is an AI PC? Microsoft defines it as a mix of CPUs, GPUs and neural processing units, which together with Microsoft Copilot software and command key can produce a laptop or desktop that can deliver AI capabilities. For example, Acer already produces a laptop that incorporates Copilot, so that the users get an AI-powered assistant that helps to solve complex workflows.

In other words, AI PCs usher in a whole new range of smart hardware plus user-friendly software that can help individuals, companies and governments to use AI to increase productivity.

Two competitive issues arise for AI PCs. First, economies of scale reduce marginal cost of production that enables output to be produced faster, cheaper and more conveniently. However, relying solely on speed and cost efficiency does not guarantee resilience. If one part of the supply chain is disrupted, it could halt production along the whole chain.

How did America achieve technological superiority? The US technology makers recognised that the larger consumer market has a significant scale and cost advantage over the smaller-scale “specific requirement” defence market. Big profits from the consumer market were reinvested into research and development (R&D) to create a continuous product improvement cycle and competitive advantage. Leading technology companies (either software or hardware makers) with brand captive markets (like Microsoft or Apple) can generate high profits that are reinvested in next generation technology, putting pressure on followers who have neither scale nor enough money to allocate to AI R&D.

Going forward, the competitive game will be which tech company or hardware maker is able to incorporate AI features that will be quickly adopted by the mass consumer market.

Chip War author Chris Miller urges the US and its allies to upgrade R&D and contain rivals such as Russia and China to maintain tech superiority. Sanctions and tariffs mean that access to AI intellectual property, chips, hardware and lithographic machines, as well as critical materials may not be available to competitive rivals. However, the strategy of containment through sanctions cannot work forever, as Chinese tech companies develop alternative solutions that can either mimic the nano-chip performance or do so at lower cost because of economies of scale.

In recent months, major companies such as Nvidia, Amazon, Google, Microsoft, Alibaba, Tencent and Huawei have been building data centres in Southeast Asia, because the region’s aggregate income is close to US$4 trillion (RM18.8 trillion). Moreover, the region has a lot of cheap energy in terms of hydropower, solar, water, and oil and gas reserves that are vital to powering AI chips and data centres. This is the opportunity for Malaysia to enable it to catch up in key technologies because the semiconductor supply chain is global in nature. No country can make everything that semiconductor chip manufacturing needs. The chip design market is exploding because everyone is designing chips for specialist needs. Tech competition is ultimately competition for talent at all levels of the production chain.

In sum, the technology game has shifted up one notch with the arrival of AI, robotics and due to geopolitical rivalries. Investing in the new AI technology sphere requires a clear-eyed understanding of how the global tech ecosystem has been changing dramatically since 2022. Picking the right company means understanding how its corporate strategy fits into this new, dynamic complex environment. Some subsidies help, but it really is about the ability to deliver what the market needs over the long term.

Tan Sri Andrew Sheng writes on Asian global issues. Loh Peixin is a research associate at the George Town Institute of Open and Advanced Studies, Wawasan Open University. The writers are engaged in a major study of the tech industry in Penang. 

https://www.theedgemarkets.com/node/715798