NVIDIA’s Blackwell AI Chips Face Delays Due to Design Issues, Now Expected in 2025 | Nvidia hardware | Cpu hardware examples | Gpu vs cpu laptop | Turtles AI
NVIDIA’s Blackwell AI Chips Face Delays Due to Design Issues, Now Expected in 2025
Design challenges push NVIDIA’s next-gen AI chip release to 2025, impacting supply chain dynamics
Giosky
Highlights:
- Delay in production of NVIDIA’s Blackwell chips, now expected in 2025.
- Design issues related to chip interconnection technique.
- Increased demand leads NVIDIA to boost TSMC orders by 25%.
- Sample distribution to partners for testing has commenced.
NVIDIA Faces Delays in Blackwell AI Chip Production Due to Design Issues
NVIDIA is encountering significant delays in the production of its next-generation Blackwell AI chips. Originally scheduled for release in Q4 2024, the chips may now be postponed to 2025. The delay is attributed to design challenges related to the chip-to-chip interconnection technique, an issue being addressed in collaboration with their manufacturing partner, TSMC, which is producing the chips using a 4nm process.
The Blackwell chips are touted as the most advanced NVIDIA has ever produced, boasting 208 billion transistors and a 10 TB/s chip-to-chip link. Anticipating strong market demand, NVIDIA has increased its order volume with TSMC by 25%, particularly to meet the needs of major tech companies like Amazon, Microsoft, Google, and Meta.
The GB200 NVL36 and NVL72 server units, incorporating these new chips, are expected to cost $1.8 million and $3 million, respectively. The delay could impact the adoption of these technologies by major cloud service providers, who rely on these advancements to power their AI infrastructures.
Despite the delays, NVIDIA has already begun distributing samples of the new chips to partners for testing, highlighting the strategic importance of the Blackwell platform. Full-scale production is now anticipated by 2025, by which time the Blackwell architecture is expected to become the dominant platform for high-performance AI chips.
This situation underscores the challenges faced by the semiconductor industry as it pushes towards increasingly powerful and complex solutions. The growing demand for AI chips is putting significant pressure on all stages of the production chain, from design to manufacturing and testing. This may lead to a reevaluation of development strategies and capacity management within the industry.
