NVIDIA Revamps AI Chip Lineup with B300 Series, Embracing TSMC’s Advanced Packaging

NVIDIA is shaking things up in the world of AI chips with the rebranding of its upcoming Blackwell Ultra series to the B300 series. This strategic move involves the introduction of the high-performance B300 and GB300 chips, both of which are set to leverage TSMC’s advanced CoWoS-L packaging technology.

According to a recent report from TrendForce, the B200 Ultra has been rechristened as the B300, while the GB200 Ultra has been renamed to the GB300. Similarly, the B200A Ultra and GB200A Ultra have been rebranded as the B300A and GB300A, respectively.

This renaming strategy reflects NVIDIA’s focus on refining its Blackwell chip segmentation to cater to the diverse needs of the AI market. The company aims to better address the high-performance demands of cloud service providers (CSPs) and the cost-performance requirements of server OEMs, offering greater flexibility in its chip offerings based on supply chain capabilities.

NVIDIA is poised to launch its newly rebranded B300 and GB300 chips in the second or third quarter of 2025. Meanwhile, the B200 and GB200 chips are currently being shipped in limited quantities, with production expected to ramp up throughout the fourth quarter of 2024 and into the first quarter of 2025.

One of the key features of the B300 and GB300 chips is their use of the new HBM3E 12-Hi stack memory. SK hynix has recently showcased this fast memory technology, which is anticipated to enter production between the fourth quarter of 2024 and the first quarter of 2025. TrendForce reports that suppliers will require approximately six months to refine their manufacturing processes and stabilize production yields for the new HBM3E memory.

The B300A specifically targets OEM clients, with mass production projected to begin in the second quarter of 2025, following peak shipments of the H200 chip.

With its innovative B300 series and commitment to advanced packaging technologies like TSMC’s CoWoS-L, NVIDIA is well-positioned to continue its leadership in the rapidly evolving AI chip landscape. This move underscores its dedication to providing high-performance, cost-effective solutions that meet the diverse needs of its customers in the cloud, enterprise, and server markets.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top