Introduction: On June 2, 2024, at COMPUTEX Taipei, Jen-Hsun Huang, the founder and CEO of NVIDIA, delivered a thrilling speech sharing how the era of artificial intelligence is boosting a global industrial revolution. Huang announced a series of groundbreaking technologies and products, including the upcoming Blackwell Ultra AI chip in 2025, as well as the new-generation AI platform Rubin and its overclocked version, Rubin Ultra. The release of these technologies heralds the advent of the robot era, where the future world will undergo transformative changes driven by physical AI-powered robots.

I. Accelerated Iteration Speed and Enhanced Computational Performance

Huang stated that future AI platform updates will follow an annual cadence, breaking the traditional Moore's Law. This rapid iteration speed will bring unprecedented computational performance enhancements. NVIDIA's accelerated computing technology has already achieved a 100-fold increase in speed, with power consumption increasing only three-fold and costs rising 1.5 times. This means future robots will possess more powerful 'brains,' capable of processing complex tasks faster and more efficiently.

II. Multi-dimensional Function Library Support

NVIDIA boasts over 350 domain-specific function libraries, covering areas ranging from deep learning, physical simulation, to gene sequencing. These libraries provide robots with a rich skill set, enabling them to function in various scenarios. For instance, in automotive manufacturing, robots can utilize these libraries for precise assembly and quality inspection. In healthcare, robots can assist doctors in surgeries and treatments.

III. Development of NIM Cloud-native Microservices and Digital Humans

With the popularity of cloud-native technology, NIM (NVIDIA Inference Manager) cloud-native microservices will become a key technology in the robot era. They enable robots to access computing resources and services anytime, anywhere, much like humans. Additionally, the evolution of digital humans ushers in a new chapter, with robots interacting with humans in a more natural and intelligent manner.

IV. Widespread Application of Physical AI and Robots

Huang predicts that the next generation of AI will need to understand the physical world, suggesting this can be achieved through mutual learning among videos, synthetic data, and AI. He believes the robot era has already begun, with all moving objects set to operate autonomously in the future, emphasizing the profound impact of AI and robotics on society.

V. Computation and Cost Savings

Through CUDA and GPU acceleration technologies, NVIDIA has significantly reduced computational costs, achieving substantial savings. This provides cost-effective solutions for AI and other high-performance computing applications.

VI. Ecosystem Construction

NVIDIA has built an extensive developer ecosystem by developing domain-specific libraries (like cuDNN) and acceleration libraries, supporting innovations across industries from healthcare to automotive, with over 5 million developers.

VII. Technological Breakthroughs and Applications

Showcased a series of cutting-edge applications, including the Coolitho computational lithography platform, gene sequencing library Pair of Bricks, combinatorial optimization library Co OPT, and quantum computer simulation system Coup Quantum. Also demonstrated how accelerated computing reduces marginal costs, fostering the exploration and application of new algorithms.

VIII. Network and Communication Optimization

Introduced high-speed network solutions for large-scale AI facilities, combining InfiniBand and Ethernet, and innovative network architectures designed to reduce latency and enhance data transmission efficiency. Highlighted the challenges and strategies for data communication in deep learning scenarios.

I won't delve too deeply into the rapid iteration of technological products, but focusing on the industrial revolution sparked by physical AI-driven robots, here are some of my thoughts:

1. Automotive: High-precision sensors and advanced visual recognition technology enable robots to automatically perform precise measurement of parts and assembly quality inspection. Additionally, the development of autonomous driving technology will bring revolutionary changes, enabling faster and more accurate recognition of the surrounding environment through multi-modal large model processing, combined with real-time roadside data, to ensure multi-dimensional driving safety, ultimately achieving autonomous driving, avoiding traffic accidents, and improving travel efficiency.

2. Manufacturing: Physical AI-driven robots enhance production efficiency, reduce costs, and minimize human error. Collaborative robots work alongside human workers, enabling flexible production line layouts and efficient material handling. Through real-time data analysis, robots can predict equipment failures and perform autonomous maintenance, reducing downtime and maintenance costs.

3. Healthcare: Improve diagnostic accuracy and treatment effectiveness. Surgical robots perform minimally invasive surgeries under doctor supervision, enhancing surgical precision and patient recovery speed. Rehabilitation robots offer personalized rehabilitation programs to help patients recover quickly.

4. Energy: Boost energy efficiency and reduce carbon emissions. Intelligent inspection robots monitor and maintain wind turbines and solar panels in real-time, ensuring stable clean energy supply. Home energy management robots adjust electricity plans based on user needs and weather conditions, minimizing energy waste.

5. Gaming & Entertainment: Provide a more immersive experience. Robotic characters in virtual reality games simulate human behavior more realistically, enhancing interactivity and fun. Smart speakers and chatbots assist users with music recommendations, news updates, and more.

Meanwhile, Huang provided several real-world examples illustrating the industrial transformation driven by the seamless integration of robots and digital twins: Foxconn and Delta are building digital twin facilities in their factories, achieving perfect real-world and digital integration, with Omniverse playing a crucial role. Pegatron and Wistron are also following this trend by establishing digital twin facilities in their robotic factories. Regarding the future integration of digital twins, robots, and AI, here are some bold predictions: 1. Urban Infrastructure Management: City managers can monitor the status of urban infrastructure like bridges, tunnels, water supply, and drainage systems in real-time. Physical AI-driven robots regularly inspect and maintain these facilities, ensuring their proper functioning and preventing potential safety hazards. 2. Traffic Management: Optimize traffic flow and reduce congestion. Physical AI-driven traffic signal control robots adjust signal timing based on real-time road conditions, enhancing road efficiency. Additionally, the development of autonomous vehicles presents new opportunities and challenges for urban traffic management. 3. Environmental Monitoring and Governance: Physical AI-driven monitoring robots deployed in environmentally sensitive areas provide real-time data on air quality, water quality, and other environmental indicators. This data supports environmental governance, such as adjusting industrial emission limits and optimizing green spaces. 4. Public Safety and Emergency Response: Assist police with patrol, surveillance, anti-terrorism, and emergency response. In disasters like fires or earthquakes, robots can quickly respond, providing rescue support and minimizing casualties and property damage.

The era of physical AI-driven robots has arrived, profoundly transforming our lifestyles and work patterns. In this disruptive industrial revolution, we eagerly anticipate more innovations and breakthroughs...