Recorded talk

NVIDIA at COMPUTEX 2023 Highlights

enplatform captions

Summary

Computex 2023 presented a full-stack effort to enable large generative-AI models and industrial digitalization: NVIDIA announced the DGX GH200 (connecting 256 Grace Hopper Superchips into a massive “1X of flop GPU”), an MGX open modular server design, and the Spectrum X ethernet platform aimed at hyperscale AI. Complementing hardware, NVIDIA highlighted Omniverse and generative AI services for factory digitization, a WPP Omniverse ad content engine, and ACE for games to animate virtual characters. Evidence came from Jensen’s 90-minute keynote, partner demos, SoftBank’s planned Grace Hopper–powered data center, and a MediaTek in-vehicle AI partnership; the transcript does not provide performance benchmarks, cost, or deployment timelines. The announcements create an integrated path from chips to datacenter networking to application-level services, but operational details and metrics remain unspecified in the source.

Key technical points

DGX GH200 AI supercomputer: system seamlessly connects 256 Grace Hopper Superchips into a massive "1X of flop GPU" for building large language models and generative AI chatbots.
MGX: an "open modular server" design announced for accelerated computing (presented as a modular hardware approach to datacenter servers).
SoftBank is building a next-generation data center powered by NVIDIA Grace Hopper aimed at generative AI and 5G/6G communications workloads (example of a customer deployment).
Spectrum X: an Ethernet platform specifically designed for hyperscale "degenerative" AI (phrasing from the transcript) — positioned for large-scale AI networking needs.
Omniverse announcements: electronic manufacturers are adopting NVIDIA's generative AI services and Omniverse to digitize state-of-the-art factories (digital twin / industrial metaverse use case).
NVIDIA and WPP partnership to create a digital ad content engine inside Omniverse (application of Omniverse for media/content generation).
ACE for games: a generative-AI capability described to animate and add life to virtual gaming characters (game-focused application of generative models).
MediaTek collaboration: partnership announced to develop in-vehicle AI cabin solutions for next-generation software-defined vehicles spanning entry-level to premium segments (edge / automotive AI application).
Greg Estes framed Omniverse, generative AI, and accelerated computing as enabling a new era of 3D workflows—termed the industrial metaverse—targeted at enterprise 3D/industrial pipelines.
NVIDIA’s 90-minute keynote and partner ecosystem presence (nearly 3,500 in person plus tens of thousands watching live) were used to demo end-to-end platform connections from chips to applications.
Over 30 NVIDIA partners showcased technologies on the showroom floor at the Nangang Exhibition Center (partner ecosystem engagement and demos).
A dedicated AI developer Meetup allowed developers to meet Jensen and engage directly, indicating emphasis on developer relations and community feedback loops.

Engineering relevance

Engineers should note the announced system-level components that connect chip design, server form-factor, networking, and application platforms: DGX GH200 (256 Grace Hopper Superchips) for large-model training, MGX modular servers for datacenter deployments, and Spectrum X for hyperscale AI networking. Omniverse and generative AI services were positioned as the application and digital-twin layer for factory digitization and media workflows, while ACE and the MediaTek vehicle partnership show targeted application stacks for games and in-vehicle AI. The source provides concrete architecture announcements and partner deployment examples (e.g., SoftBank) but does not include operational benchmarks, cost, or rollout schedules, so engineering planning will require follow-up for performance/scale metrics and integration details.

Japan relevance

Several items in the transcript directly touch Japanese industry relevance: SoftBank’s announced next-gen data center using Grace Hopper targets Japan’s telecom and cloud infrastructure; Omniverse adoption by leading electronics makers maps to Japan’s large manufacturing and factory-digitalization base; the MediaTek in-vehicle AI cabin partnership and mention of software-defined vehicles relate to automotive OEMs and suppliers in Japan. The partner ecosystem presence at Computex and the developer meetup also signals regional engagement opportunities for Japanese engineering teams to trial NVIDIA stacks and collaborate with partners showcased at the Nangang Exhibition Center.

Full transcript

Transcript

platform captions

foreign [Music] we're back [Music] our first live event in almost four years [Music] hello from Taipei I'm Dion Harris head of accelerated Computing product Solutions at Nvidia computex Asia's most important Trade Tech show is back in full swing after a three-year hiatus and this week Nvidia took center stage at computex from Jensen delivering a memorable opening keynote to nearly three dozen Partners showcasing Technologies on the showroom floor and an AI developer Meetup and more Jensen kicked off things with a bang speaking to a packed House of more than 3 500 with tens of thousands watching live from around the world in a sweeping 90-minute talk he covered a series of dramatic announcements and demos together they showed how nvidia's accelerated Computing platform and partner ecosystem are powering a new era of generative Ai and Industrial digitalization here are some of the key announcements first he announced the Nvidia dgx gh200 AI supercomputer the system seamlessly connects 256 Grace Hopper Superchips into a massive 1X of flop GPU enabling developers to build large language models for generative AI chat Bots among other important use cases second he presented nvidia's mgx and open modular server design for Accelerated computing third he shared how SoftBank is building a next-gen data center powered by Nvidia Grace Hopper for generative Ai and 5 and 6G Communications fourth he showed how nvidia's Spectrum X ethernet platform which is specifically designed for hyperscale degenerative AI fifth he talked about announcements in the Omniverse how leading electronic makers are embracing nvidia's generative AI services and Omniverse to digitize their state-of-the-art factories and Nvidia is teaming up with wpp to create a digital ad content engine in Omniverse and finally Jensen described Ace for games which leverages generative AI to breathe life into virtual gaming characters hey Jen how are you later on opening day Jensen appeared as a special guest at mediatek's CEO Rick size press event they announced a partnership to develop in-vehicle AI cabin solutions for next-gen software-defined vehicles from entry level to Premium on day two Greg Estes head of corporate marketing and developer programs at Nvidia gave a talk on racing toward the industrial metaverse he described how a new era of 3D workflows is being enabled by Nvidia and its Partners using Omniverse generative Ai and accelerated computing at the non-gang Exhibition Center over 30 Nvidia Partners showcased our Cutting Edge Technologies separately we had a dedicated AI developer Meetup where they got a chance to meet and talk with Jensen in person it's been quite an exciting week for us all we're already looking forward to next year's edition of computex see you then