Samsung Electronics wants to dominate cutting-edge chipmaking

SAMSUNG ELECTRONICS (SE) is a behemoth. The South Korean tech company is the crown jewel of the mighty Samsung chaebol, as the country’s conglomerates are known. It makes more smartphones than any other company in the world, as well as home-entertainment systems and appliances. It dominates the manufacturing of memory chips, which are used to store data on electronic devices and whose price has been pushed up by the global semiconductor shortage. SE’s annual revenues of $200bn are not much lower than those of Apple, the most valuable firm in history, and it is sitting on a cash pile of $100bn.

Listen to this story

Enjoy more audio and podcasts on iOS or Android.

Now both SE and its parent group, whose name means “three stars”, are entering a critical new chapter. In August Lee Jae-yong, the scion of the family which founded Samsung in 1938, was released from prison, where he spent two stints after a conviction for his involvement in a bribery scandal. He is finally taking full control of the empire from his late father, Lee Kun-hee, who died last year. Succession was complicated first by the elder Lee’s six-year coma, then by his son’s bribery conviction, linked to SE’s efforts to win the government’s backing for a merger of two Samsung subsidiaries that would cement his control.

Free at last, Mr Lee has grand plans for the company, which he wants to become as dominant in cutting-edge logic chips, used for processing information, as it already is in memory and smartphones. That will pit SE head-to-head with chipmaking powerhouses such as TSMC of Taiwan and America’s Intel, and thrust it into a fierce global contest over one of the world’s most strategic industries.

On October 7th SE confirmed it will manufacture some of the world’s most advanced logic microprocessors, based on its novel “gate-all-around” architecture with transistors measuring three nanometres (billionths of a metre), in 2022. It also surprised analysts by announcing a plan to mass-produce two-nanometre chips from 2025. It is forecast to invest an eye-watering $37bn or so in capital expenditure across its businesses this year. And it is winning new customers, such as Nvidia, an American chip designer, and Tesla, an electric-car maker.

The outcome of Mr Lee’s gamble will have profound consequences—and not just for Samsung. It matters to South Korea, whose president justified Mr Lee’s parole as being in the national interest, given the chaebol’s importance to the economy. And it will influence the global semiconductor industry, the critical nature of which has been underscored by the worldwide chip shortage. To ensure success, the man whom acquaintances describe as shy, decent and astute must also summon a degree of ruthlessness.

SE is a complex corporate creature with a strategic challenge and underwhelming stockmarket performance. It is best understood as divided into two main businesses. The first makes “sets”: smartphones, televisions and household appliances. The second produces “components”, which go into Samsung’s own sets, as well as being sold to external customers like Apple. Samsung splits its sets business further into two divisions: TVs and appliances such as washing machines, and digital devices (chiefly smartphones). The component business, meanwhile, comprises semiconductors and displays.

The sets business is not a growth engine. In Mr Lee’s hierarchy of SE operations, say people close to the company, home appliances sit at the bottom, below the TV unit with similarly low margins but a bigger role in reinforcing SE’s valuable brand. Next comes the handset business, which in the early 2010s contributed over half of profits. Although its obituary has been written several times before, it continues to generate lots of cash and, thanks to a new fast-selling range of phones with foldable screens, some fresh optimism.

Atop the hierarchy sit semiconductors. Historically, SE has focused on memory chips, where it has 44% of the global market for DRAM chips (used for temporary storage in desktops) and 36% in NAND devices (used for permanent storage in mobiles). The memory business brings in just over 20% of revenues but nearly half of operating profits (see chart 1). Everything else is potentially expendable in the service of its juicy margins. If a “set” business has a disagreement with a components unit over pricing or other terms, insiders say, the component business takes precedence. According to the company, its unique ecosystem benefits from having diverse businesses which allow internal innovation while providing stability through the ups and downs of industry cycles.

Late-onset memory loss

Analysts reckon that SE’s memory-making has plenty of life left in it. Because such chips are critical for storing data across industries, it is “only going in one direction: up”, says Nicolas Gaudois of UBS, a bank. Omdia, a research firm, predicts that the global memory-chip market will expand at double-digit rates each year between 2020 and 2025. It is now less cyclical thanks to surging demand from data centres and, on the supply side, consolidation in the industry where ever more extreme miniaturisation means that rivals can no longer step up production as easily as before. SE says that it has proven an ability to innovate and extract value in established businesses. Internally, though, certain SE executives worry that memory is a mature operation. And some investors fret that demand for memory chips may soften towards the end of the year.

One option would be to follow Apple and develop a services business, which has grown from 8% of the iPhone-maker’s revenues in 2012 to a fifth. However, despite a few successes, notably in payments and health apps, SE’s efforts to add software and services to its world-beating hardware have been sporadic.

This is partly because SE’s hardware-first approach is deeply rooted in its culture. It will probably be reinforced by Mr Lee’s character and experience. “His disposition is very cautious and conservative, more so than his father,” says a former SE executive. This innate conservatism may have been strengthened by his first big endeavour after attending Harvard Business School. In the late 1990s, at the height of the dotcom bubble, he invested in eSamsung, a venture-capital firm. Watching the subsequent bust left Mr Lee sceptical of Korean software engineers, says the former executive; eSamsung was shut down.

Going big on services could also jeopardise SE’s long-standing and successful partnerships with software giants such as Google and Microsoft. In 2014 SE launched a music-streaming service called Milk Music, which despite its success was scrapped two years later. “Google viewed Samsung’s software efforts as fragmenting the Android ecosystem and felt threatened,” recalls a former executive. “I feel pretty sure that Samsung has given up on software and services,” he sighs. He worries about a big missed opportunity. Even if the firm talks about making another run at it, he adds, this would probably be merely to keep Google and other partners honest.

Another problem is China. The country is an important source of demand for both memory and logic chips. To help satisfy it, SE is finishing its second memory-chip plant in the western city of Xi’an this year. Despite rising tensions between China and the West, in particular America, neither SE nor any other South Korean chipmaker is likely to give up on their giant neighbour, which is likely to remain a big buyer for many years (especially for the technically more complex DRAM chips). This means that SE must walk a fine line to keep Chinese clients while not relinquishing American customers.

This array of complications and risks helps explain SE’s underperformance relative to other giants, both in consumer technology (Apple and Xiaomi of China) and chipmaking (TSMC and Intel). Because it combines several relatively distinct businesses, the company suffers from a conglomerate discount. It is listed only in Seoul, where limits on exposure to individual stocks have in the past pushed local investors to sell SE, which accounts for nearly a fifth of the KOSPI stockmarket index, whenever its share price spiked. And SE’s enormous cash pile depresses returns.

As a result, despite solid operating performance SE’s shares have traded between one and one-and-a-half times forward book value, far below its peers. Increasing its dividend from 22% of net profit in 2018 to 78% in 2020 helped more than double SE’s market value in the two years to January. But Apple’s nearly trebled in the same period. A strong outlook for semiconductors and lower cyclicality in memory chips have yet to translate into a richer valuation. Having surged by nearly half in late 2020, SE’s market capitalisation has declined by 13% since the start of the year, while New York’s tech-heavy NASDAQ index and a basket of global chipmakers have made gains (see chart 2).

The logical move

Mr Lee’s bet on cutting-edge logic chips is designed to reverse the underperformance. The idea is to win a big slice of a fast-growing and lucrative market for non-memory chips, which account for 70% of the $550bn global semiconductor market. Mr Lee has set out a goal to match SE’s roughly 40% market share in memory in the “foundry” business of manufacturing processors for customers.

The Samsung scion has his work cut out. SE’s foundry division took a hit in 2016 when Apple moved all its business for the A-series processor for the iPhone to TSMC. That shock offered a stark example of how SE’s complex structure throws up possible conflicts of interest with key customers. Half of SE’s foundry output goes to its sets divisions, with the rest supplying outside customers. Apple preferred TSMC, a pure foundry firm, to SE, with which it competes in smartphones.

So far progress towards Mr Lee’s ambitious target, first signalled a few years ago, has been slow. The firm has around 15% of the foundry market, compared with more than 50% for TSMC, which plans to spend $100bn over the next three years on new capacity. SE’s non-memory chip revenues make up only 7% of total sales (though that is up from nothing in 2005 and the company also makes some other specialised processors for sensors and the like). The share of profits is even lower.

Perceived conflicts of interest are not its only challenge. Although the memory and logic businesses share some commonalities and overheads, they differ in important ways. Producing memory chips is chiefly about speed, volume and economies of scale. Making high-end logic processors is much more complex technologically, with engineering done at nanoscopic scales and customers increasingly desiring silicon customised for their purposes.

On technology SE (and, to be fair, just about everyone else) has fallen behind TSMC in at least the last two generations of cutting-edge processors. Part of that may be down to sensible caution. But reticence can further complicate relations with customers, many of which are reluctant to place orders unless they can get capacity guarantees, says a semiconductor executive at another firm. Rather than anticipating their needs, SE has been reactive, says the executive.

Cognisant of these problems, Mr Lee clearly wants to accelerate SE’s transformation. The company is using its research-and-development (R&D) prowess to take some risks on next-generation logic chips, for example with its new advanced chip architecture. The company does not break out how much of its capital spending is going to memory chips and how much to logic. According to CLSA, a broker, there is an emphasis on logic chips, which are also more R&D intensive.

An expanding constellation

SE is also mulling a $17bn factory to manufacture cutting-edge logic chips in Texas, to appease America’s desire to bring more chipmaking back home from Asia (and, possibly, to partake in a delayed $52bn subsidy splurge on the semiconductor industry that Congress is considering). And the new customers it is courting, such as Nvidia and Tesla, have no overlap with its other businesses, notes Sanjeev Rana of CLSA.

Help could come from the fraught geopolitics of semiconductors. Although rising technonationalism over chip design and manufacture makes governments favour domestic production and local champions, it may nevertheless end up benefiting SE. As China ratchets up military pressure on Taiwan, which it considers part of its territory, fears are growing over TSMC’s future. According to another semiconductor executive, many firms that use TSMC are scrambling to reduce exposure to the Taiwanese company, just in case. As TSMC’s closest rival, Samsung could be a big beneficiary. SE has the biggest industrial complex of semiconductor fabs and engineers in the world, and some of the best chip technology, says Mark Newman, a former Samsung group executive who is the chief commercial officer of Nyobolt, a battery startup.

One way to turbocharge the transition would be to split SE into its constituent businesses, as investment bankers have long recommended. This would also eliminate the potential conflicts of interest that have hampered SE’s foundry division. A dual listing in America, meanwhile, could help with the KOSPI-related drag.

Neither a break-up nor another listing looks likely, however. Mr Lee seems reluctant to countenance the radical first option. One attempt at persuading SE into the second around 2016, as part of an activist campaign by Elliott Management, an American hedge fund which had taken a stake, failed. Aware of this, shareholders are therefore putting pressure on SE to at least do something about its unused cash. One idea would be to pay out 100% of free cashflow to them. Alternatively, SE could do a big acquisition. The company states that “the founding family is clearly aligned with all other shareholders in its objectives to create maximum value and see that value properly reflected in the market.”

To make a material difference to SE’s financial performance any deal would have to be big. Mr Lee’s predisposition and preferences make such a gamble improbable in software and services. That leaves chipmaking as the place where the firm’s cash could be spent. One potential takeover target is NXP Semiconductors, a Dutch firm that specialises in the fast-growing market for automotive chips. With a market value of $50bn it would be a heavy lift, but not an impossible one.

If Samsung Electronics is to become a logic-chip star to rival TSMC, Mr Lee had better get lifting. Last year he vowed not to hand management of SE to his children (though the Lees are likely to retain the biggest stake in the company through various family-controlled vehicles). The promise to be the last Lee to run the firm, combined with what insiders say are other improvements to corporate governance, clears the path to the top for its multitude of talented executives. They must be hoping that Mr Lee leaves them a legacy that is less complicated than his father’s.

For more expert analysis of the biggest stories in economics, business and markets, sign up to Money Talks, our weekly newsletter.

An early version of this article was published online on October 17th 2021

This article appeared in the Business section of the print edition under the headline “The third star”

Related posts