Summary of Key Points
This news report highlights a long-term technical partnership between NVIDIA and SK Hynix to jointly develop high-end memory required for AI devices, aiming to address the storage shortage caused by the surge in AI demand. During Jeong-Hoon Huang's visit to South Korea, he emphasized that the storage shortage would persist for several years, and SK Hynix plans to double its production capacity within five years. However, despite the cooperation announcement, the stock prices of both SK Hynix and Samsung declined, reflecting market concerns about short-term supply-demand imbalances and intensified competition.
Detailed Analysis
1. What exactly is the cooperation about? Joint development of high-end memory for AI devices
The collaboration between NVIDIA and SK Hynix goes beyond mere memory transactions; it involves a deep integration:
- Joint product development: They will work together to design memory for AI supercomputers (Vera Rubin), AI CPUs (Vera), PCs with AI capabilities (RTX Spark drivers), and robotic computing platforms (Jetson Thor).
- Supply stability: This partnership aims to overcome the challenges of long development cycles and high investment associated with high-end memory, ensuring that NVIDIA has sufficient memory for building its AI infrastructure.
- New market expansion: SK Hynix will gain access to new markets in personal AI (e.g., AI assistants on smartphones/PCs) and physical AI (e.g., robots interacting with the real world).
- Additional collaborations: NVIDIA is also partnering with Korean companies like Naver (similar to Baidu) and Doosan Group to build AI data centers, while SK Telecom will use NVIDIA technology to develop a gigawatt-scale AI cloud by 2027, providing AI training and inference services for local businesses.
In short, the two companies are jointly producing high-speed, large-capacity memory needed for AI applications, while helping SK Hynix enter new markets and ensuring that NVIDIA has a steady supply of memory.
2. Why is NVIDIA eager to partner with SK Hynix? The explosion in AI demand
Jeong-Hoon Huang clearly stated during his visit that the storage shortage would continue for several years due to the following reasons:
- Surging AI demand: Training large models and running AI applications require large amounts of high-bandwidth memory (HBM), which is faster and has higher capacity than regular memory, but it is more difficult to produce.
- Supply chain challenges: Every link in the supply chain—from chip manufacturing to packaging and silicon photonics—is struggling to meet demand.
- Proactive supply chain management: As a leading AI company, NVIDIA needs to secure a stable supply of HBM to avoid disruptions in its operations. SK Hynix, being the market leader in HBM (with over 50% of the market share last year), is an ideal partner for this purpose.
3. What can SK Hynix gain from this partnership? New markets, stable orders, and technological advancement
For SK Hynix, this collaboration is highly beneficial:
- New market opportunities: It will enter the fields of personal AI and physical AI, breaking away from its traditional storage business focus.
- Stable orders: With long development cycles for high-end memory, early partnerships with NVIDIA reduce the risk of unsold products.
- Technological improvement: SK Hynix can learn advanced memory technologies from NVIDIA, enhancing the competitiveness of its products.
- Synergies with related companies: SK Telecom’s AI cloud will also use SK Hynix’s memory, creating a closed-loop ecosystem.
In summary, SK Hynix not only secures a long-term supply but also opens up new business opportunities.
4. Why did stock prices decline? Market concerns about short-term issues
Despite the positive aspects of the partnership, SK Hynix’ stock price fell by 3%, and Samsung’s fell by 8% on the same day:
- Short-term supply shortages: Even with the collaboration, increasing production capacity will take time (SK Hynix aims to double its capacity in five years), so costs and profits may remain high in the short term.
- Intensified competition: Samsung has joined NVIDIA’s list of HBM suppliers this year, potentially reducing SK Hynix’s market share.
- Cautious market expectations: Investors are skeptical about immediate benefits and may be cautious due to overall tech sector adjustments.
5. The future of the storage industry: Shortage expected until 2030, with increased production by manufacturers
According to Jeong-Hoon Huang and SK Hynix’s CEO:
- Persistent shortage: Supply bottlenecks are likely to persist until 2030, with memory prices remaining stable in the coming years.
- Manufacturers’ expansion: SK Hynix plans to double its production capacity, and Samsung and Micron will also expand.
- Deeper integration of AI and storage: AI giants will continue to collaborate closely with storage manufacturers, making storage a core component of the AI ecosystem.
In conclusion, the storage industry is facing strong demand and tight supply in the coming years, leading manufacturers to aggressively expand their production capacity. This news reflects how the rapid growth of the AI industry is driving improvements in the storage supply chain—NVIDIA securing memory, SK Hynix increasing production, and the entire industry preparing for the AI era. However, short-term supply-demand issues and increased competition have led to cautious market sentiment.
This news article illustrates how the demand for computing power driven by AI is forcing upgrades in the storage supply chain, with companies like NVIDIA and SK Hynix taking proactive steps to meet the demands of the AI revolution.