Latest / Boardroom Conversations with Fexingo: CEOs, Strategy, and Corporate Leadership Explained / How the Samsung Board Defended the Memory Chip Empire
Transcript
- Lucas: Samsung Electronics makes more memory chips than any company on earth. Roughly 40 percent of the world's DRAM and over 30 percent of NAND flash flows through its fabs. But in early 2023, the board stared at a problem that felt existential. Luna: The worst memory chip downturn in decades, right? Lucas: Exactly. Memory chip prices had collapsed by over 50 percent year-over-year. Samsung's semiconductor division posted its worst quarterly profit in 14 years — just 640 billion Korean won, down 97 percent from the year before. Luna: And the board had to decide whether to ride it out or make drastic cuts. Lucas: And they had to do it while their biggest rival, SK Hynix, was gaining ground in a critical new technology: high-bandwidth memory, or HBM, used in AI accelerators. So the board was balancing a short-term crisis with a long-term technology shift. Luna: Let's start with the short-term. What did the board actually do? Lucas: In April 2023, Samsung's board approved a historic production cut. They slashed memory chip output — something the company had avoided for decades. In previous downturns, Samsung would keep producing, flooding the market, and using its scale to force smaller competitors out. Luna: But this time was different. Lucas: Right. The board said: 'We're cutting output by a meaningful level, especially for NAND flash.' Within a quarter, they reduced NAND production by over 50 percent. It was a painful move — they idled fabs, furloughed workers. But it sent a signal to the market that Samsung would no longer chase volume at any cost. Luna: And did the market respond? Lucas: It did. Memory chip prices stabilized by the second half of 2023. By early 2024, prices were recovering, and Samsung's semiconductor division swung back to profitability. But the board wasn't just managing the downturn — it was also betting on a new technology curve. Luna: That's where HBM comes in. Walk me through that. Lucas: High-bandwidth memory is essentially stacking DRAM chips vertically to deliver massive data throughput — exactly what AI models like NVIDIA's GPUs need. SK Hynix had been first to market with HBM3, the previous generation, and had locked in NVIDIA as a customer. Samsung was late. Luna: So the board had to decide: play catch-up or leapfrog? Lucas: They chose leapfrog. In late 2023, Samsung's board approved an aggressive push into HBM3E — the next generation. They allocated a significant portion of their annual capital expenditure, which was about 54 trillion won that year, to hbm specific fabrication lines. Luna: That's roughly 40 billion dollars. Not a small bet. Lucas: Not at all. And it paid off. By mid-2024, Samsung had qualified its HBM3E products for NVIDIA's supply chain. By early 2025, it was ramping volume to compete head-on with SK Hynix. The board's willingness to invest counter-cyclically — cutting output in legacy memory but pouring money into next-gen — was a masterclass in capital allocation. Luna: But the board didn't just face market challenges. There was also the leadership crisis. Lucas: Right. In 2023, Jay Y. Lee, Samsung's de facto leader and vice chairman, was convicted of stock manipulation related to a 2015 merger. He faced a potential multi-year prison sentence. The board had to consider a leadership vacuum at the very top. Luna: Did they have a contingency plan? Lucas: They did. The board had already been strengthening the role of the CEO and the professional management team. During Lee's absence, the board empowered co-CEOs — Kyung Kye-hyun for semiconductors and Han Jong-hee for consumer electronics — to run day-to-day operations. They also established a 'crisis management committee' within the board to oversee strategic decisions. Luna: So they basically institutionalized decision-making so the company didn't rely on one person. Lucas: Exactly. And it's worth noting that Samsung's board has a reputation for being relatively independent compared to other Korean chaebols. In 2023, only two of its nine board members were insiders. The rest were outside directors with backgrounds in academia, law, and finance. That independence gave them the credibility to make tough calls without being seen as Lee's rubber stamp. Luna: But let's be honest — Samsung's governance still has critics. The Lee family retains enormous influence through a web of cross-shareholdings. Lucas: That's true. The board has pushed for governance reforms, including unwinding some of those circular ownership structures. In 2024, Samsung announced a plan to simplify its holding company structure, but progress has been slow. The board has to balance the family's interests with minority shareholder demands. It's a tension that hasn't been resolved. Luna: So what's the takeaway for other boards? Lucas: I think there are three things. First, counter-cyclical investment works when you have the balance sheet to withstand short-term pain. Samsung had 130 trillion won in cash reserves. Second, having an independent board with real decision-making authority is crucial during a leadership crisis. And third, betting on technology transitions — even when you're behind — can create a second act. Luna: It's interesting, because this episode really shows how a board's job isn't just oversight. It's also about strategic conviction. Lucas: And that conviction is tested most when the market is screaming 'sell' and the media is screaming 'scandal'. The Samsung board didn't panic. They cut where they needed to cut, invested where they needed to invest, and kept the company focused on the long game. Luna: You know, conversations like this one — where we dig into real boardroom decisions — that's the kind of thing that really helps people understand how business works. And if these episodes have moved your own thinking forward in some small way, the best way to keep them coming is listener support. Lucas: Absolutely. We keep this podcast ad-free and independent, and that's only possible because of listeners who chip in. If you find value in what we do, you can support us at buy me a coffee dot com slash fexingo. It's a small gesture that makes a big difference. Luna: And it really helps us keep the focus on the substance, not the sponsors. So thank you to anyone who's already done that. Lucas: So back to Samsung. The board's biggest test may still be ahead. The memory chip market is cyclical by nature, and the AI boom that fueled demand for HBM may eventually normalize. The board now has to decide how much capacity to build for a world where AI chips might not grow at 50 percent a year forever. Luna: And there's also the foundry business. Samsung is trying to challenge TSMC in contract chip manufacturing, but that requires billions more in investment with no guarantee of success. Lucas: Right. The foundry business lost market share in 2024, dropping to around 10 percent, while TSMC held over 60 percent. The board has to decide whether to keep pouring capital into a segment where they're a distant second, or pivot resources back to memory where they're the leader. That's a question for 2026 and beyond. Luna: But based on what we've seen so far, I'd bet they'll make the hard call. Lucas: I think so too. The Samsung board has shown it can be decisive, independent, and long-term oriented. That's a rare combination in any company, let alone a family-controlled Korean conglomerate. And it's why, despite the scandals and the downcycles, Samsung remains the memory chip king.