Nvidia CEO Jensen Huang Announces Plans for AI Accelerator Upgrades and New Platforms
Nvidia CEO Jensen Huang Unveils Plans for Annual AI Accelerator Upgrades, Introduces Blackwell Ultra Chip for 2025 and Rubin Platform for 2026
In a recent announcement, Nvidia Corp. CEO Jensen Huang revealed the company’s strategy to upgrade its AI accelerators on an annual basis. The tech giant plans to release a Blackwell Ultra chip in 2025 and is currently developing a next-generation platform called Rubin for 2026.
During the Computex trade show in Taiwan, Nvidia also introduced new tools and software models, signaling its commitment to the advancement of artificial intelligence technology. Huang emphasized the importance of generative AI in driving a new industrial revolution and highlighted Nvidia’s role in bringing this technology to personal computers.
While Nvidia has established itself as a leader in AI data center systems, the company is now looking to expand its customer base beyond cloud-computing giants. Huang anticipates a broader adoption of AI across various industries, from shipbuilding to drug development, emphasizing the transformative power of accelerated computing.
Huang pointed out the growing need for accelerated computing solutions as data processing requirements continue to increase exponentially. He highlighted Nvidia’s technology as a cost-effective and energy-efficient solution, offering significant savings compared to traditional computing methods.
The upcoming Rubin AI platform is set to leverage HBM4, the next iteration of high-bandwidth memory crucial for AI accelerator production. Nvidia’s focus on innovation and performance enhancement reflects its roots in gaming cards for desktop PCs, a background that is now influencing the integration of AI functions in computer systems.
Collaborations with companies like Microsoft and Qualcomm are driving the adoption of AI-enhanced laptops, with Nvidia graphics cards playing a key role in enhancing performance and enabling new features in software applications. Nvidia is also providing tools and pretrained AI models to empower software developers in leveraging AI capabilities on PCs.
Additionally, Nvidia is introducing a new server design program, MGX, to expedite the development of server computers using Nvidia chips. The company’s previously announced products, such as Spectrum X for networking and Nvidia Inference Microservices, are now widely available and being adopted across industries.
Huang also highlighted the use of digital twins in Nvidia’s Ominverse virtual world, showcasing the potential for advanced simulations and modeling tasks. The adoption of these tools by companies like Foxconn underscores the impact of AI technology on optimizing operations and planning in manufacturing.
Overall, Nvidia’s commitment to annual AI accelerator upgrades and the introduction of cutting-edge technologies like the Blackwell Ultra chip and Rubin platform demonstrate the company’s leadership in driving innovation in the AI space.
–With assistance from Jane Lanhee Lee.
(Updates with Rubin announcement)
Most Read from Bloomberg Businessweek
©2024 Bloomberg L.P.
