How I Predicted It: The Real Reason Behind Samsung Electronics’ Collapse

Recently, it has become widely known both inside and outside the market that Samsung Electronics’ semiconductor division is in a full-blown crisis, from its foundry business to its memory business, which had long been considered technologically superior. As a result, the share price has been hitting new 52‑week lows day after day, and foreign investors are rapidly cutting their holdings of Samsung Electronics stock.

A few days ago, a Dong-A Ilbo article ran an unfiltered anonymous interview with an engineer who had worked at Samsung Electronics for 20 years, and figures from various parts of society are voicing concerns about the company. Because most of these views fail to reflect the industry reality, I am writing this piece to offer a broad, fact-based整理.

Depending on how you tally revenue, Samsung Electronics’ foundry business is estimated to rank second or third in the world. TSMC is number one by a wide margin, and the number two and three spots seesaw between Samsung and Intel depending on whether you count Samsung’s internal transactions as foundry revenue.

You might ask, “Isn’t being number two or three in the world pretty good?” But if you look at the structure of the foundry industry, you see that it is anything but. Foundry services are divided into leading‑edge nodes (sub‑7nm advanced processes) and legacy nodes (16nm and above, mature processes). Once you exclude the companies that have dropped out of the race because they cannot shoulder the capex required for leading‑edge nodes, there are only three players left: TSMC, Samsung, and Intel. It’s like being ranked third in a school that only has three students.

The data show that over the past two years TSMC’s market share has inched up, while Samsung has merely held its ground. The issue is that in April 2019, Samsung Electronics Chairman Lee Jae-yong unveiled the “System Semiconductor Vision 2030,” announcing a plan to invest 133 trillion won by 2030 in design and foundry, among other areas, to become the global number one in system semiconductors.

At the time, Samsung drew up a plan to overtake TSMC on the back of its formidable cash generation and massive capex, and that plan turned into concrete projects. The new foundry fab in Pyeongtaek and the new fab in Taylor, Texas are the results. The problem is that the fabs have been built or are under construction, but there are no customers. Judging from an October 18 article titled “Samsung’s Taylor Foundry in the U.S. Sees ASML Equipment Delivery Delayed”, it is highly likely that there is virtually no customer volume lined up for production at the Taylor foundry.

As mentioned earlier, a single piece of semiconductor equipment for leading‑edge foundry nodes costs hundreds of billions of won, and building a new fab requires at least several trillion to tens of trillions of won. Samsung Electronics is arguably the only manufacturer in the world that can commit to that level of investment without first securing customers. It was a decision made possible because the company is controlled by an owner family.

Around the same time, in late 2021 TSMC asked major customers such as Apple, Qualcomm, and Nvidia for large prepayments in exchange for locking in future production capacity, and booked roughly 5 trillion won of related liabilities in its earnings release—about a quarter of its quarterly revenue at the time. Prepayments are cash received before the corresponding revenue is recognized. A common pain point for manufacturers is securing working capital for the production process—cash tied up in raw materials, inventory, and labor before sales are realized. TSMC solved this by collecting prepayments.

The stark contrast between Samsung Electronics, which went ahead with massive investments before securing customers, and TSMC, which demanded hefty prepayments before even delivering products, ultimately stems from a difference in technological competitiveness.

The following is an excerpt from remarks by Democratic Party lawmaker Jung Jin-wook at a National Assembly audit session, quoted in a News Tomato article titled “The Dilemma of Samsung’s Business Divisions Exposed at the National Audit.”

The argument goes like this: because Samsung Electronics has both a semiconductor design division (System LSI) and a foundry (contract manufacturing) division under one roof, customers are reluctant to share their advanced chip designs with Samsung, and that is why Samsung struggles to secure clients.

As I briefly mentioned in “Why SK Hynix Is the Answer in Memory Semiconductors”, I believe it would make sense for Samsung Electronics to split into three divisions—mobile and appliances, memory, and foundry—but I set that aside because Chairman Lee Jae-yong is unlikely to see it that way. That was a point about organizational theory and why a spin‑off would be desirable from that perspective, not a commentary on conflicts of interest within Samsung.

As lawmaker Jung Jin-wook’s remarks and the News Tomato piece show, there is indeed a conflict of interest between Samsung’s business divisions, and resolving it would be desirable in the long run. But the fundamental reason Apple and Qualcomm—by far the biggest buyers in the mobile AP (application processor) market—do not entrust their volumes to Samsung is its lagging technology.

When Apple first launched the iPhone, the AP it used was a chipset designed by Samsung Electronics’ System LSI division and manufactured by its foundry division (which at the time was not yet a standalone division). This close partnership between Apple and Samsung Electronics continued through the iPhone 3G and 3GS. Starting with the A4 chipset used in the iPhone 4, Apple switched to its own in-house designs, but it still outsourced production to Samsung’s foundry. The customer that actively backed Samsung Electronics’ foundry business in its early days and helped it grow was Apple. Qualcomm also has a history of using Samsung’s foundry to manufacture its flagship APs.

Companies like Apple and Qualcomm, which need to produce hundreds of millions of units, inevitably want to diversify their supply chains. It lets them lower the unit cost and hedge against potential supply risks. If a fire breaks out at the TSMC plant that makes iPhone chipsets and production is halted for two weeks, there is essentially nothing Apple can do right now, and that would ultimately translate into lower revenue. For this reason, Apple actually dual-sourced production of the A9 chipset used in the iPhone 6S in 2015 between TSMC’s 16nm process and Samsung Electronics’ 14nm foundry process. This led to the so‑called “Chipgate” controversy, where it was alleged that iPhone 6S models equipped with A9 chipsets manufactured by Samsung had a significantly higher battery drain under heavy load than identical models using chipsets manufactured by TSMC. Apple responded with a statement saying, “High‑stress workloads are not representative of real‑world usage, and our internal testing shows a 2–3% difference between manufacturers at most,” but from that point to this day, it has stuck exclusively with TSMC.

When I put up the post above, I had no idea the yield on advanced nodes would turn out to be this disastrous and carelessly said it wasn’t that important. What I really wanted to highlight at the time was Samsung Electronics’ utterly shambolic PDK (process design kit). Design engineers at fabless semiconductor companies like Qualcomm use the PDKs provided by foundries as the basis for design, simulation, and verification, then hand over the finalized layouts to the foundry, which manufactures the chips accordingly. The problem was that the performance gap between the simulated results and the actual silicon was far too large—and always in the wrong direction.

I know “corporate culture” is such a broad notion that plenty of HR experts lump every problem under the sun into that bucket and declare, “It’s a culture issue.” That’s not the kind of vague, hand‑wavy argument I’m trying to make here.

Instead, let’s look at an article written in April 2022 by the head of SemiAnalysis, a semiconductor research firm, on how Samsung Electronics’ toxic corporate culture has been wrecking its semiconductor business:

At the time, the Exynos processor designed and manufactured by Samsung for its new Galaxy models delivered performance far below that of Qualcomm’s chips from the same generation, sparking controversy. The mobile division that decided to use Exynos blamed the LSI (design) division, and LSI in turn blamed the foundry division that fabricated the chips—a classic blame game where everyone points the finger at the next guy.

Two and a half years after that analysis was published, the blame game between divisions is still ongoing, and the next‑generation Exynos processor currently in development is likewise expected to underperform Qualcomm’s rival offering. An even bigger problem is that yields on the new process node are so low that the chip has reportedly been dropped entirely from next year’s Galaxy S25 lineup.

Around the start of the 2020s, even Samsung’s DRAM business—its core competitive stronghold—began to fall behind rivals in both performance and cost structure. Until the late 2010s, it was widely viewed as being roughly 1.5 years ahead of SK Hynix and Micron, to the point where it earned the nickname “Mass‑Production Electronics”: competitors would announce new product development, and Samsung would already be ramping to high‑volume manufacturing. Now, however, it finds itself chasing Micron, which sits in third place by market share.

At some point, Samsung Electronics developed a culture that prioritized headline‑grabbing achievements over solid engineering, and that mindset has led to a fundamental erosion of its technological capabilities.

A few days ago, Mark Zuckerberg, founder and CEO of Meta, said on the Acquired podcast that “If the leaders don’t have a deep understanding of technology, then that company is not a tech company.”

Samsung Electronics is effectively controlled by Chairman Lee Jae‑yong, who studied Asian history at Seoul National University and later attended Harvard Business School. The group’s number‑two executive, Vice Chairman Jung Hyun‑ho, who was recently cited as the person who ordered that reports be written “so that even an elementary school student can understand them,” studied business administration at Yonsei University and also attended Harvard Business School.

There are successful companies that, like Apple, are essentially technology firms but are led by CEOs who are not technologists by training, such as Tim Cook. Arguing that Samsung Electronics will soar again only if Chairman Lee Jae‑yong and Vice Chairman Jung Hyun‑ho completely step back from management—and that otherwise the company is doomed to sink—would be an overreach. However, it does seem clear that the KPIs set for senior executives are fundamentally misaligned, and as a result the company is moving in the wrong direction. I was taught that the most important quality in a leader is the ability to put people more capable than yourself in the right roles and delegate your authority to them.

Today, the scope of knowledge a company needs to innovate and survive has far surpassed the amount of information any individual can store or process. That is why “collaboration” and “frictionless sharing of knowledge” have become more important than ever.

What about Samsung’s corporate culture? Its HR and compensation systems are still stuck in the “merit-and-punishment” principle devised by the late Chairman Lee Kun-hee, while the company itself has grown into a dinosaur. Numerous business divisions within the same company are all looking in different directions, each striving to deliver its own results. Even if you pour in the finest ingredients, the dish will fail if they do not come together in harmony. You do not need to watch “Black-and-White Chef” to know that.

There can be no single, consistent answer. Have Samsung Electronics employees really been working with one heart toward a shared goal up to now?

Only the strong will of a single person can truly transform Samsung Electronics’ culture.