This article first appeared in Forum, The Edge Malaysia Weekly on September 9, 2024 - September 15, 2024
Intel is one of the most famous semiconductor chip producers. Founded in 1968, it was the legendary chip foundry that had its “Intel Inside” almost every personal computer and server in the 1980s and 1990s. But its success in that sector missed the rise of DRAM (dynamic random access memory) and NAND (NOT-AND logic circuit) flash memory chips, today dominated by South Korean chipmakers; GPU (graphics processing unit) chips led by Nvidia; and ARM (Advanced RISC Machine) reduced instruction set computer architecture chips used widely in mobile phones.
In other words, while Intel reaped the benefits of dominating general purpose chips, it missed the rise of specialised chips used in gaming machines, mobile phones (such as the M1 chip used by Apple) and crypto-asset mining computing equipment, as well as the artificial intelligence (AI) chips used in big data processing centres. Its asset-heavy vertically integrated foundry fab model has been challenged by asset-light but design-intensive (fabless) chip companies such as Qualcomm, AMD and Nvidia, which rely on specialist makers such as Taiwan Semiconductor Manufacturing Co (TSMC) to do high-quality fabrication.
Intel’s legendary co-founder Gordon Moore coined Moore’s Law, which predicted that the number of transistors in integrated circuits (ICs) would double every two years, generating speed, scale and scope in computer processing capacity. Meanwhile, co-founder Andy Grove was the ruthlessly focused engineer and corporate captain who drove research and development (R&D), production efficiency and product branding, particularly the famous 86-series chip.
Intel was the darling of the tech market in the run-up to the Nasdaq boom of 2000. Thereafter, the company was led by marketing and financial engineers who slowly lost focus on how to evolve chip design and production in a situation where the design and manufacturing of smaller, faster and energy-efficient chips were costing more and more. Consulting firm McKinsey estimated that the design cost of bringing a 65nm (nanometer) chip to production in 2006 was US$28 million, whereas this rose to US$540 million for a 3nm chip by 2020. The cost of each advanced ASML lithography machine essential to producing such advanced chips is now more than US$378 million (RM1.6 billion) each. The financial cost of capital and capacity expansion rise with each new generation of chips, making it tough not to keep investing for the long term, but companies must do so at the right cycle.
Intel became the top chipmaker because for half a century, US companies like Intel, Motorola and Texas Instruments (TI) were vertically integrated as they designed, manufactured and marketed their own chips. In the 1980s, Motorola and TI were much bigger than Intel, but they were more conservative in R&D.
As the cost of R&D rose, from 2019 to 2023, Intel spent US$101 billion on increasing plant and equipment (P&E) capacity and US$75 billion on R&D. But it also lavished shareholders with US$30 billion in stock buybacks and US$25 billion in cash dividends, which together absorbed 79% of its net income. Intel’s distribution to shareholders has been far greater than that of TSMC and Samsung, which distributed 67% and 38% respectively over the same period.
Despite the vast sums that Intel had aggressively indulged on its shareholders, it consistently allocated 30% of its revenue to P&E and 22% to R&D. The other major global integrated device manufacturer (IDM), Samsung, had a big gap of 13% between P&E and R&D compared to Intel. In contrast, TSMC, which is a pure foundry that manufactures to client needs (such as Apple), has allocated a significant 45% of its revenue to capital expenditure while its spending on R&D remains low at 8%. In other words, TSMC lets its clients focus on R&D while it focuses on production excellence. But being involved in different chip sectors, the broad skills and knowledge intensity of its engineers are impressive.
During the pandemic, even though Intel made huge allocations to catch up with its competitors, its financial performance began to weaken when its net income declined more than 50% after 2021. To maintain its share price, the company generously provided a dividend payout to shareholders averaging 129% of net income between 2022 and 2023. At the same time, Intel increased borrowing, bringing its total debt to around US$45 billion. For a financial analyst, this looked more like financial engineering to produce profits through leverage, rather than paying attention to real excellence and cutting edge engineering.
Despite getting more than US$8 billion from the US government under the CHIPS Act to help onshore chip manufacturing return to the US, Intel is beginning to experience both financial and operational headwinds.
On Aug 1, the company announced 15,000 job cuts and suspended dividends. The market reacted harshly after it released its 2Q2024 earnings report, resulting in around US$30 billion being wiped off its market cap. Market rumours swirled after Lip Bu Tan — a former CEO of Cadence Design Systems, which produces cutting-edge software tools to design advanced chips — resigned from Intel’s board, with unconfirmed disagreements on the company’s strategic direction.
Intel’s 2Q2024 earnings report showed that operating losses had increased by US$948 million compared to last year. Key factors contributing to the huge losses included the higher cost of producing smaller processors such as Meteor Lake, and increased construction charges on new AI fabs in the US and supporting facilities around the world. Our calculations of economic-value-added based on Intel’s financial accounts suggest that value-added declined to negative-value-added of US$11.5 billion from the previous year.
Intel has shown the classic strategic choice for market leaders, which is to keep milking profits from legacy winners but lose focus on keeping the R&D edge over strong competitors like AMD, Nvidia and Qualcomm, which offer comparable AI chips with better performance or price. If you do not cut out the fat earlier with short-term impact on quarterly profits, the market will punish you when you take belated action. TSMC today has eight times the market cap of Intel.
Beleaguered CEO Pat Gelsinger has called in Wall Street advisers like Goldman Sachs and Morgan Stanley to advise Intel on how to move forward. Can financial engineers fix real engineering strategic issues, other than to temporarily calm impatient investors? There are options such as spinning off subsidiaries like Altera or splitting the company into different listed companies. The risk is that Intel, which has a total market cap of less than one-tenth of that of any of the Magnificent Seven tech giants, will be bought up as their manufacturing arm. As Grove used to say, only the paranoid survive. The question is whether the present Intel leadership is paranoid enough to survive that cruellest of market tests.
Tan Sri Andrew Sheng writes on global issues that impact Asia. Loh Peixin is a research associate at the George Town Institute of Open and Advanced Studies, Wawasan Open University. The authors are engaged in a major study of the tech industry in Penang.
Save by subscribing to us for your print and/or digital copy.
P/S: The Edge is also available on Apple's App Store and Android's Google Play.