In yet another big announcement at the Computex Conference in Taipei, NVIDIA CEO Jensen Huang unveiled more of the company’s plans for the future of AI computing. The spotlight shone on the Rubin AI chip platform, set to launch in 2026, and the Blackwell Ultra chip, slated for 2025.
The Rubin Platform
As the successor to the highly anticipated Blackwell architecture, which is expected to ship later in 2024, the Rubin Platform represents a leap forward in NVIDIA’s AI computing capabilities. Huang emphasized the need for accelerated computing to tackle the ever-increasing demands of data processing, stating, “We are seeing computation inflation.” NVIDIA’s technology promises to deliver an impressive 98% cost savings and a 97% reduction in energy consumption, positioning the company as a frontrunner in the AI chip market.
While specific details about the Rubin Platform were scarce, Huang revealed that it will feature new GPUs and a central processor named Vera. The platform will also incorporate HBM4, the next generation of high-bandwidth memory, which has become a critical bottleneck in AI accelerator production due to soaring demand. Leading supplier SK Hynix Inc. is largely sold out of HBM4 through 2025, underscoring the fierce competition for this essential component.
NVIDIA and AMD Leading the Charge
NVIDIA’s shift to an annual release schedule for its AI chips highlights the intensifying competition in the AI chip market. As NVIDIA strives to maintain its leadership position, other industry giants are also making significant strides. During the opening keynote at Computex 2024, AMD Chair and CEO Lisa Su showcased the growing momentum of the AMD Instinct accelerator family, unveiling a multi-year roadmap that introduces an annual cadence of leadership AI performance and memory capabilities.
AMD’s roadmap begins with the AMD Instinct MI325X accelerator, set to be available in Q4 2024, boasting industry-leading memory capacity and bandwidth. The company also previewed the 5th Gen AMD EPYC processors, codenamed “Turin,” which will utilize the “Zen 5” core and are expected to be available in the second half of 2024. Looking ahead, AMD plans to release the AMD Instinct MI400 series in 2026, based on the AMD CDNA “Next” architecture, promising enhanced performance and efficiency for AI training and inference.
Implications, Potential Impact, and Challenges
The introduction of NVIDIA’s Rubin Platform and the company’s commitment to annual updates for its AI accelerators have far-reaching implications for the AI industry. This accelerated pace of innovation and development will enable more efficient and cost-effective AI solutions, driving advancements across various sectors.
While the Rubin Platform holds immense promise, there are challenges and considerations that must be addressed. The high demand for HBM4 memory and the supply constraints posed by leading supplier SK Hynix Inc. being largely sold out through 2025 could potentially impact the production and availability of the Rubin Platform.
Moreover, NVIDIA must strike a delicate balance between performance, efficiency, and cost to ensure that the Rubin Platform remains accessible and viable for a wide range of customers. Compatibility and seamless integration with existing systems will also be crucial to facilitate adoption and minimize disruption for users.
As the Rubin Platform sets the stage for accelerated AI innovation and development, businesses and researchers alike must stay informed and prepared to leverage these advancements. By leveraging NVIDIA’s Rubin Platform, organizations across various industries can drive efficiencies and gain a competitive edge in their industries.
