Yazılar

Nvidia to Supply Over 260,000 Blackwell AI Chips to South Korea

Nvidia, the U.S. semiconductor leader, announced it will deliver more than 260,000 of its latest Blackwell AI chips to South Korea’s government and major conglomerates, including Samsung Electronics, SK Group, Hyundai Motor Group, and Naver. The move solidifies South Korea’s position as an emerging artificial intelligence powerhouse in Asia.

The agreement follows a meeting between Nvidia CEO Jensen Huang, South Korean President Lee Jae Myung, and leading business figures on the sidelines of the Asia-Pacific Economic Cooperation summit in Gyeongju. While the deal’s value and delivery schedule were not disclosed, Huang highlighted that just as Korea has led in producing ships, cars, and semiconductors, it can now “export intelligence” to the world through AI.

South Korea’s government plans to use over 50,000 of the chips to enhance its AI infrastructure, while Samsung, SK Group, and Hyundai will each deploy up to 50,000 chips in smart factories and autonomous systems. Internet giant Naver will purchase 60,000 chips to boost its computing power for projects like the National AI Computing Center.

Amid ongoing U.S.-China trade tensions that have restricted Nvidia’s access to the Chinese AI chip market, the company is expanding its footprint in alternative markets such as South Korea. The partnership reinforces Nvidia’s role in driving global AI advancement, particularly in industries such as manufacturing, robotics, and autonomous driving.

NextSilicon Unveils New Processor to Compete with Intel, AMD, and Nvidia

Israeli startup NextSilicon has announced the development of a new central processing unit (CPU) aimed at challenging industry leaders Intel and AMD, while positioning itself as an alternative to Nvidia’s computing systems. The company’s flagship “Maverick-2” chip is already in production and is designed for high-precision scientific computing tasks such as nuclear weapons modeling — an area historically dominated by Nvidia.

NextSilicon, which has raised $300 million in funding, revealed that its upcoming CPU is based on RISC-V, an open-source instruction set architecture increasingly adopted by major chipmakers like Nvidia and Broadcom. The CPU is currently a test prototype but will eventually complement the company’s Maverick-2 chip to create a more complete high-performance computing ecosystem.

The startup claims its chips can perform certain computations faster and more efficiently than Nvidia’s offerings, without requiring software rewrites. U.S. Sandia National Laboratories has been testing systems powered by NextSilicon chips for three years. According to senior scientist James H. Laros III, the results have been “impressive,” showing strong potential for advancing computational performance while minimizing code changes.

Metagenomi Uses Amazon’s AI Chips to Power Next-Gen Gene Editing

Biotech company Metagenomi (MGX.O) has begun using Amazon Web Services’ custom AI chips to accelerate the discovery of new gene-editing technologies, marking one of the first major biotech applications of Amazon’s in-house silicon beyond large language models and chatbots.

The Emeryville, California-based firm, which is developing tools to deliver gene therapies directly into human cells, said AWS Inferentia chips have given it a major cost advantage over Nvidia’s AI hardware, cutting computational expenses by about half while maintaining comparable performance.

Metagenomi’s approach relies heavily on artificial intelligence to design and test enzymes capable of safely editing DNA. The company scans nature for rare proteins that might serve as effective delivery vehicles for genetic material and then uses AI to generate millions of variants in search of the most effective designs.

“We generated over a million different proteins from a rare class of enzymes used in gene editing,” said Chris Brown, Metagenomi’s head of discovery. “It was a clear cost advantage to use the Inferentia platform. Unless you cast a broad enough net early, you risk missing key breakthroughs entirely.”

Amazon’s Inferentia chips, first introduced in 2019 to enhance the AI capabilities of its Alexa virtual assistant, are now being used by other industries beyond software — with biotechnology emerging as a new frontier for AI-driven hardware.

By applying cloud-based AI to the complex problem of gene delivery and editing, Metagenomi hopes to make treatments for genetic disorders faster and more affordable, while demonstrating how custom AI infrastructure can accelerate scientific discovery.