AI hardware

Nvidia is expected to exceed earnings expectations again, driven by strong AI demand. Analysts project significant revenue and profit growth, fueled by its leading GPUs that power AI development. The company is also actively enabling AI data center expansion. Despite increasing competition and memory cost challenges, Nvidia’s new Blackwell and Vera Rubin AI systems are shipping, signaling continued expansion and investor confidence in its long-term AI strategy.

Nvidia is reportedly increasing prices on AI servers, including Vera Rubin and Grace Blackwell models, by over 15% for some clients. This move is driven by rising costs for critical memory components like HBM, essential for their AI accelerators. The price adjustment reflects escalating R&D and supply chain investments, as well as Nvidia’s dominant market position, though it may prompt some customers to explore alternatives.

South Korean investors are significantly increasing their investments in U.S. markets, particularly in semiconductor-related stocks and leveraged ETFs, to hedge against potential domestic market corrections. This trend persists even as global institutions show interest in Korean equities. Notably, a premium is observed for U.S.-listed depositary receipts of Korean companies, a phenomenon that puzzles market analysts. The shift may reflect a geographical diversification of investment themes rather than a fundamental change in outlook.

AMD is acquiring Taalas, an AI inference chip startup, to enhance its AI hardware portfolio. This move addresses the growing need for specialized silicon beyond general-purpose GPUs for efficient AI inference, a crucial component of the generative AI boom. The acquisition signifies AMD’s strategy to offer integrated AI systems, complementing its GPU offerings with dedicated inference accelerators for low-latency applications and a broader AI ecosystem.

SK Hynix achieved record quarterly profits, driven by surging demand for AI hardware. Despite narrowly missing analyst expectations, year-over-year revenue and operating profit saw remarkable growth, reflecting strong sales of High Bandwidth Memory (HBM) chips. The company’s HBM3 and upcoming HBM3E variants are crucial for AI accelerators. A significant multi-year deal with Nvidia underscores SK Hynix’s vital role in the AI revolution and its strong position for continued growth in the advanced computing sector.

AMD invests up to $5 billion in Anthropic for AI infrastructure, agreeing to supply tens of billions of dollars in AMD-powered AI systems, including Instinct MI450 accelerators. This strategic partnership, alongside similar deals with OpenAI and Meta, highlights AMD’s aggressive push to capture the AI market by integrating its technology and financial backing into large-scale AI deployments, positioning itself as a key hardware and investment partner for leading AI developers.

AMD is launching its “Helios” rack-scale AI system to challenge Nvidia’s dominance in the AI hardware market. Major clients, including Microsoft, Meta, OpenAI, and Oracle, have committed to deploying Helios, drawn by its integrated approach and focus on cost-efficiency per token. Analysts estimate Helios’s competitive pricing and believe AMD could capture a significant market share. This launch marks a critical step in AMD’s decade-long strategic resurgence, leveraging its integrated hardware and software capabilities, and acquisitions to offer a compelling alternative in the rapidly expanding AI compute landscape.

SK Hynix led an Asian tech stock rally, recovering from recent AI spending concerns. Semiconductor firms in South Korea, Japan, and Taiwan saw gains, mirroring a Wall Street rebound. Despite strong AI infrastructure demand, some warn of speculative excess and potential market volatility in the AI hardware sector.

This week’s Tech Download explores the growing reliance of teens on AI chatbots, drawing parallels to social media’s impact. Concerns are rising about AI dependency potentially displacing real-life relationships, mirroring past social media issues. Meanwhile, the tech sector sees significant job cuts at Microsoft, SpaceX joining the Nasdaq 100, and increased investment in AI hardware. Geopolitical factors are also influencing AI adoption, with Alibaba banning certain AI tools due to security concerns.

Nvidia’s next-gen Kyber rack-scale AI architecture, vital for its 2027 Rubin Ultra chips, faces delays until 2028 due to manufacturing challenges with its PCB midplane. This setback could create opportunities for rivals like AMD and Google, who are investing in custom AI chips. Despite these issues, current-gen Rubin systems are in full production and demand remains strong.