During his highly anticipated speech at Computex in Taipei, Nvidia co-founder and CEO Jensen Huang reaffirmed his vision to bring generative AI to every data center. In his first public appearance in nearly four years, Huang made a series of significant announcements, including chip release dates, the introduction of the DGX GH200 supercomputer, and key partnerships with major companies. Let’s delve into the highlights from his two-hour-long keynote.

Full Production of GeForce RTX 4080 Ti GPU for Gamers: Huang unveiled that Nvidia’s eagerly awaited GeForce RTX 4080 Ti GPU for gamers has entered full production and is being manufactured in large quantities in partnership with Taiwanese companies. This high-performance graphics card promises to deliver an exceptional gaming experience.

Nvidia Avatar Cloud Engine (ACE) for Games: To enhance character interactions in games, Huang introduced the Nvidia Avatar Cloud Engine (ACE) for Games—a customizable AI model foundry service. ACE provides game developers with pre-trained AI models, enabling more dynamic and lifelike non-player character (NPC) behavior through AI-powered language interactions.

Cuda Computing Model Reaches New Milestones: Huang highlighted the widespread adoption of Nvidia’s Cuda computing model, which now serves over four million developers and powers over 3,000 applications. The Cuda platform has witnessed over 40 million downloads, with a significant surge of 25 million downloads in the past year alone, reflecting its growing popularity.

HGX H100 GPU Server in Full Volume Production: The GPU server HGX H100, featuring the world’s first transformer engine, has commenced full volume production. Huang emphasized that companies all over Taiwan are manufacturing this groundbreaking server, designed to deliver high-resilience performance for data center applications.

Nvidia’s Strategic Acquisition of Mellanox: Huang lauded Nvidia’s strategic decision to acquire supercomputer chipmaker Mellanox for $6.9 billion in 2019. He expressed his belief that this acquisition has been one of Nvidia’s most pivotal moves, strengthening its position in the market.

Next Generation Hopper GPUs and Grace Hopper Superchip: Huang announced that production of the next generation of Hopper GPUs will commence in August 2024, exactly two years after the initial generation began production. Additionally, he revealed that the GH200 Grace Hopper superchip, boasting impressive specifications such as 4 PetaFIOPS TE, 72 Arm CPUs, 96GB HBM3, and 576 GPU memory, is now in full production. Huang emphasized that the Grace Hopper is the world’s first accelerated computing processor with a significant memory capacity, designed to excel in high-resilience data center applications.

Partnerships, including SoftBank, Google Cloud, Meta, and Microsoft: Nvidia forged a partnership with SoftBank to introduce the Grace Hopper superchip into SoftBank’s distributed data centers in Japan. This collaboration aims to host generative AI and wireless applications on a multi-tenant common server platform, reducing costs and energy consumption. Furthermore, Nvidia’s MGX reference architecture, which facilitates the development of over 100 server variations for AI, accelerated computing, and omniverse applications, is being utilized in partnerships with companies in Taiwan. Notable partners include ASRock Rack, Asus, Gigabyte, Pegatron, QCT, and Supermicro.

Spectrum-X Accelerated Networking Platform: Huang introduced the Spectrum-X accelerated networking platform, which significantly boosts the speed of Ethernet-based clouds. This platform includes the Spectrum 4 switch with 128 ports operating at 400GB per second, facilitating adaptive routing, performance isolation, and in-fabric computing. Additionally, the Bluefield 3 Smart Nic enhances congestion control when connected to the Spectrum 4 switch.

NWPP Partnership and Nvidia Isaac ARM: In a significant collaboration, Nvidia partnered with WPP, the largest advertising agency in the world, to develop a content engine based on Nvidia Omniverse. This engine will enable the production of high-quality photos and video content for advertising purposes, leveraging Nvidia’s advanced technologies.

Furthermore, Huang announced the availability of the Nvidia Isaac ARM robot platform, catering to robotics enthusiasts and developers. This comprehensive stack, ranging from chips to sensors, is built around the Nova Orin chip and represents a significant advancement in robotics.

Nvidia’s Remarkable Stock Performance: Huang acknowledged the soaring success of Nvidia’s stock over the past year, largely attributed to the company’s dominance in AI computing. With a current market valuation of approximately $960 billion, Nvidia stands among the most valuable companies globally, surpassed only by Apple, Microsoft, Saudi Aramco, Alphabet, and Amazon.

Uncertainties in the Chinese Market: As Nvidia’s cutting-edge silicon and advancements in generative AI gain attention, Chinese AI firms closely monitor the situation. However, the threat of potential U.S. chip bans poses a concern for their progress in generative AI, which relies on substantial computing power and data. Last year, the U.S. government imposed restrictions on Nvidia’s A100 and H100 graphic processing units (GPUs) in China, impacting training large language models like OpenAI’s GPT-4. Huang noted that the H100 chip, based on the Nvidia Hopper GPU architecture, has garnered substantial demand due to its remarkable AI training and inference capabilities. As China seeks self-reliance in crucial tech sectors, the ban serves as a reminder of the nation’s drive for technological autonomy.

Jensen Huang’s Computex keynote highlighted Nvidia’s unwavering commitment to advancing AI across data centers. With groundbreaking releases, strategic partnerships, and significant technological advancements, Nvidia continues to shape the future of computing. As the company expands its reach into various sectors, the industry eagerly anticipates the transformative impact of Nvidia’s innovations.