The Rise of AI Chips Skyrockets Samsung's Profits by 1,800%

What Happened?

Samsung Electronics, the world's largest memory chip maker, announced that its second-quarter 2024 operating profit surged 1,800% year-on-year to 10.4 trillion won (approximately $7.5 billion). This explosive growth is primarily due to strong demand for chips used in artificial intelligence, such as HBM (High Bandwidth Memory) and high-performance SSDs. According to the BBC, semiconductor demand continues to outpace supply, driving prices up. For perspective, in the same quarter of 2023, Samsung posted operating profit of just 668.5 billion won, its lowest in 14 years due to post-pandemic demand slump. The recovery has been fueled by the rise of generative AI, which requires enormous amounts of high-bandwidth memory to train and run models like GPT-4o or Gemini.

Why Is This Important?

This milestone not only confirms Samsung's dominance in the semiconductor market but also signals a shift in era: artificial intelligence is redefining the technology value chain. Unlike previous cycles driven by PCs or smartphones, AI is now the main catalyst. Companies like NVIDIA, AMD, and Intel also benefit, but Samsung, as a key memory supplier, captures a significant portion of the value. Historically, semiconductor upcycles have been tied to mass consumer products: the PC boom in the 90s, smartphones in the 2010s. However, the current cycle is unique because it is led by AI infrastructure, which consumes memory intensively. For example, a single AI server can require up to 8 times more memory than a traditional server. For investors, this indicates that demand for AI chips is not a bubble but a sustained trend backed by multi-billion-dollar investments from giants like Microsoft, Google, and Meta. For consumers, it could mean greater availability of devices with integrated AI, such as Samsung's upcoming Galaxy AI, but also higher prices in the short term due to rising component costs.

Consequences and Outlook

The AI chip boom has several implications: first, it accelerates investment in semiconductor manufacturing, with Samsung and TSMC expanding their plants. Samsung has already announced plans to invest $45 billion in its chip business by 2025, including new HBM production lines in Pyeongtaek. Second, it increases technological dependence on a few suppliers, creating geopolitical risks, especially between the US and China. The trade war has led to restrictions on exporting advanced chips to China, and Samsung, with factories in both countries, is caught in the middle. Third, it drives innovation in artificial intelligence, as more memory and bandwidth enable larger and faster models. However, there is also a risk of oversupply if demand stabilizes. The semiconductor industry is cyclical: the last boom-and-bust cycle (2021-2023) saw revenue drops of up to 50% for memory makers. Although the AI boom is real, corrections could occur if hyperscalers reduce investments or if alternative memory technology (like phase-change memory) matures. Additionally, competition is intensifying: SK Hynix is the leader in HBM3 with a 50% market share, while Micron has also entered strongly. Samsung, though leading in total volume, has lost ground in next-generation HBM, which could limit its future growth.

What Readers Should Know

• Direct cause: Unsatisfied demand for AI chips, especially HBM, has driven prices up. DRAM and NAND memory prices rose between 20% and 30% in the second quarter, according to TrendForce.
• Competition: SK Hynix and Micron also compete in HBM, but Samsung leads in total memory volume. However, SK Hynix was the first to supply HBM3E to NVIDIA, giving it a technical edge.
• Beyond Samsung: This growth reflects a global trend benefiting the entire semiconductor ecosystem. Companies like ASML, which makes lithography machines, have also seen orders surge.
• Risks: Potential overinvestment, trade tensions, and dependence on a single sector. If AI demand slows, the industry could face overcapacity.
• Opportunity: For startups and tech companies, now is the time to integrate AI into products leveraging available hardware. HBM memory enables larger and more efficient models, lowering cost per inference.

"Semiconductor demand continues to outpace supply, pushing prices up," notes the BBC in its analysis. This shortage is not temporary: HBM manufacturing requires advanced processes and long lead times, keeping supply tight at least until 2025.

In conclusion, Samsung's 1,800% profit increase is not an anomaly but a symptom of a profound transformation. Artificial intelligence is not only changing software but also the physical foundation that supports it. For industry professionals, this means the war for AI chips has only just begun. The key question is whether Samsung can maintain its lead against more agile competitors and whether demand will continue to grow at the same pace. What is clear is that AI has entered a hyperscale phase that will redefine the tech industry for the next decade.