NVIDIA secures memory empire in South Korea, but questions linger over AI’s true cost – NaturalNews.com
NVIDIA secures memory empire in South Korea, but questions linger over AI’s true cost
The global artificial intelligence race just accelerated in a quiet corner of Seoul, where Nvidia CEO Jensen Huang and South Korea’s most powerful corporate chieftains gathered around plates of grilled pork belly and glasses of soju to ink a series of deals that will reshape how AI gets built, who controls it, and what it will mean for ordinary people who fear they are being left behind. The agreements, announced Monday, are not merely business transactions. They represent a tectonic shift in the balance of power between centralized corporate AI giants and the decentralized, open-source movement that true believers like myself have championed for years. At stake is nothing less than the future of human knowledge, the integrity of information, and whether AI becomes a tool of liberation or a weapon of corporate control.Key points:
To understand why these deals matter, you must first understand the hidden choke point in the AI industry. Every large language model, every AI-generated video, every real-time chatbot response depends on one thing: memory bandwidth. The chips that process AI calculations, the graphics processing units or GPUs made by Nvidia, are only as good as the memory chips that feed them data. Without high-bandwidth memory, the most powerful GPU in the world sits idle, waiting for information. This is why SK Hynix, a company based in Icheon, South Korea, has become arguably more important than any software company in the AI supply chain.
SK Hynix signed a multi-year technology partnership with Nvidia that will commit the memory maker to developing advanced types of memory for global AI data centres. Huang made it clear during his visit that SK Hynix’s current plans to double its memory wafer capacity by 2030 will not be enough. “SK Hynix has been Nvidia’s largest memory partner. SK Hynix will continue to be Nvidia’s largest memory partner,” Huang stated after a meeting with SK Group Chairman Chey Tae-won. The deal extends for more than two years with an option to keep extending, and Huang added that Nvidia already procures billions of dollars from SK Hynix each year, with that figure set to grow substantially.
This partnership reinforces a critical shift that the mainstream financial press has barely acknowledged. Memory chips are evolving from a commodity product, something you buy off the shelf like rice or steel, into a customer-specific business where each chip is designed and optimized for a particular client’s AI architecture. That means the companies that control memory supply also control the pace of AI development. And right now, South Korea holds the keys.
From data centres to humanoid robots: the real scope of Nvidia’s ambitionsThe deals announced in Seoul extend far beyond chips and servers. Nvidia is partnering with LG Group on electronics, mechanical systems, and AI for humanoid robots. Huang met with LG Chairman Koo Kwang-mo, and the two discussed future data centre architecture including cooling systems, power delivery, and the entire design and construction of data centres. This is not theoretical research. This is the blueprint for the physical infrastructure that will house the AI systems of the next decade.
Hyundai Motor Group’s Executive Chair Euisun Chung met with Huang later that same day, and the two agreed to deepen their partnership across autonomous mobility, robotics, and AI-powered manufacturing. Huang highlighted opportunities to accelerate industrial robotics development, stating that Nvidia and Hyundai would work together to bring AI to “all forms of mobility.” He referred to Hyundai’s planned AI data centre in Saemangeum as an “AI Valley” akin to California’s Silicon Valley. “I am very happy to build Nvidia in Saemangeum,” Huang said.
Doosan Group, which develops robots and makes materials used in Nvidia’s most powerful Blackwell chips, expects its energy solutions to be used in Nvidia’s data centre platforms. The company will also use Nvidia’s physical AI technology. This is the kind of deep integration that most news outlets gloss over. Nvidia is not just selling chips. It is embedding its technology into the foundational layers of South Korea’s entire industrial economy, from memory fabrication to automobile manufacturing to energy production.
Samsung Electronics, SK Hynix’s rival and the world’s largest memory chip maker by revenue, also held discussions with Huang. Samsung co-CEO Jun Young-hyun said the two companies discussed cooperation on next-generation foundry chips, autonomous driving chips, high-bandwidth memory HBM5 chips, and even Groq’s AI LP30 chips scheduled to ship in the second half of this year. The fact that Samsung is negotiating with Nvidia while SK Hynix has already signed a deal reveals the intense competition at the heart of this industry. Whoever wins the memory war will effectively control the speed of AI progress for the next five years.
Sources include: