Jensen Huang, CEO of NVIDIA, highlighted the company's robotics and industrial digitization advances at COMPUTEX 2024 in Taipei. Huang described how manufacturers like Foxconn use NVIDIA technology, such as Omniverse, Isaac, and Metropolis, to create advanced robotic facilities. "Robotics are here. Physical AI is here. "This is not science fiction, and it is being used all over Taiwan," Huang explained.
AI Factories AND Accelerated Platforms
NVIDIA has stated that the accelerated platforms are now fully functioning. From AI PCs powered by NVIDIA RTX to corporations constructing AI factories with NVIDIA's full-stack computing platform, the future of computing is all about speed and efficiency. "The future of computing is accelerated," Huang said.
Sustainable Computing
NVIDIA is focusing on sustainable computing by integrating GPUs and CPUs to produce up to 100x faster performance with only a threefold increase in power usage. This results in 25 times greater performance per watt than CPUs alone. "Accelerated computing is sustainable computing," Huang explained.
New Semiconductor Roadmap
Huang revealed a new semiconductor release roadmap with a one-year timeframe. NVIDIA intends to launch the Rubin platform following the next Blackwell platform, which will have enhanced GPUs, an Arm-based CPU called Vera, and new networking technologies like as NVLink 6. "Our basic philosophy is straightforward: build the entire data centre scale, disaggregate and sell to your parts on a one-year rhythm, and push everything to technology limits," Huang said.
AI-enabled customer devices
NVIDIA's RTX AI PCs are designed to improve user experiences, including over 200 RTX AI laptops and over 500 AI-powered apps and games. The RTX AI Toolkit and new PC-based NIM inference microservices for the NVIDIA ACE digital human platform demonstrate NVIDIA's dedication to AI accessibility.
Enabling developers with NVIDIA NIM
NVIDIA announced NIM (NVIDIA Inference Microservices) to make it easier for the world's 28 million developers to build generative AI applications. NIM offers models as optimized containers that may be deployed across multiple systems. Partners like as Cadence, Cloudera, and Synopsys are integrating NIM to accelerate generative AI implementations.
Industry Partners and AI Factory Deals
Taiwanese suppliers, such as ASRock Rack, ASUS, and GIGABYTE, use NVIDIA GPUs and networking solutions to develop powerful AI systems. The NVIDIA MGX modular reference design platform is compatible with these systems, allowing for the best performance in AI applications. AMD and Intel have also developed CPU designs that support the MGX architecture.
Next Generation Networking
Huang highlighted plans to release Spectrum-X devices annually to address the demand for high-performance Ethernet networking for artificial intelligence. NVIDIA Spectrum-X, the first Ethernet fabric designed for AI, improves network performance by 1.6x over typical Ethernet fabrics. CoreWeave and GMO Internet Group were among the early adopters.