Written by: Techub News Compilation

Introduction

Recently, NVIDIA founder and CEO Jensen Huang appeared on Bloomberg's interview program alongside Siemens CEO Roland Busch. Against the backdrop of the AI wave sweeping global industries, the two giants announced a significant collaboration aimed at deeply integrating AI with physical manufacturing. This dialogue not only revealed how both parties are reshaping the manufacturing industry through the combination of software and hardware, but Huang also provided direct and profound insights on current hot topics such as energy bottlenecks, supply chain challenges, the competitive technology landscape, and the controversial "billionaire tax."

Summary

• NVIDIA and Siemens announced a comprehensive collaboration to integrate AI technology into industrial operating systems, accelerating the entire process from chip design to factory automation.
• Jensen Huang emphasized that energy is the core bottleneck of the AI industrial revolution, and NVIDIA is addressing this by improving chip energy efficiency (such as a tenfold increase in the Rubin architecture compared to previous generations), while calling for increased global energy investment.
• The AI Factory is key to achieving fully automated, high-yield manufacturing, with its core based on using AI to address the software limitations of robot system programming complexity.
• Regarding the "billionaire tax," Huang stated that he is "perfectly fine" with it, believing that the key to attracting talent lies in the environment itself, rather than the tax burden.
• Huang confirmed the feasibility of constructing data centers in space, arguing that its greatest value lies in producing "intelligence" (Tokens) that can be easily transmitted back to Earth.

Deep Integration of AI and Industry: From Digital Twins to Autonomous Factories

The core of the interview was the latest strategic partnership between NVIDIA and Siemens. Jensen Huang described it as "a big deal." The collaboration covers multiple aspects: accelerating Siemens' electronic design automation (EDA) and simulation software; integrating NVIDIA's physical AI and giant AI models into Siemens' Teamcenter software and factory automation operating systems.

The aim of this deep integration is to form a closed loop. NVIDIA uses the accelerated Siemens software to design its next-generation chips (such as Vera Rubin) and simulate the thermodynamic characteristics of its factories. Meanwhile, NVIDIA's automation and agent systems will empower Siemens' industrial operating systems, ultimately being implemented in partner factories like Foxconn. Huang emphasized that the goal is "to put technology to use as quickly as possible" to create real-world economic impacts.

Roland Busch of Siemens added that current technology is in place, with customers starting from product digital twins to gradually building digital twins of the manufacturing process and optimizing before actual production. AI is already operational in workshops, but the challenge lies in scaling. Scaling requires customers to possess a significant amount of skills, and implementation is not easy. The collaboration aims to lower the barriers to deployment and use.

Using NVIDIA's upcoming Vera Rubin platform as an example, Huang illustrated the necessity of such cooperation. This system integrates six different chips, with a single GPU consuming up to 240,000 kilowatts, but its energy efficiency and cost-effectiveness are ten times that of the previous-generation Blackwell architecture. Building such a complex system took "115,000 engineer-years." By accelerating EDA and simulation tools and ultimately designing the entire Vera Rubin system in Siemens' digital twin, it will become possible to create more complex systems and achieve scaling with greater efficiency.

Energy Bottlenecks and Efficiency Revolution

When asked whether energy has become a bottleneck for AI development, Huang answered affirmatively, believing it should be the norm. "Energy should always be a bottleneck in any industry," he asserted, "every industrial revolution has been constrained by energy, and this time is no exception." He further pointed out that in the United States, without former President Trump’s "pro-energy growth agenda," it would have been difficult for the industry to grow. The rise of new industries cannot be separated from energy; thus, the U.S., Europe, and even globally, need to invest more in energy.

In the face of energy constraints, NVIDIA's response is to maximize energy efficiency. From Hopper to Blackwell, energy efficiency has increased tenfold; from Blackwell to Rubin, energy efficiency will increase another tenfold. Huang explained that this directly affects customer revenue: factories of any size are limited by electricity, and at a given power, customers want to maximize computing power (Tokens) generated per watt. Therefore, each improvement in energy efficiency directly enhances customer capability and income potential.

Roland Busch provided additional insights from an industrial perspective. He noted that energy demand generally tracks GDP growth, but due to technological improvements in efficiency, there is some decoupling. However, data centers have created additional demand for high-quality energy, causing bottlenecks throughout the supply chain — from power generation (whether renewable energy or gas turbines), high-voltage transformers to medium-voltage and switch technology. Siemens' related business is therefore experiencing rapid growth, but there might still be supply shortages in certain regions. The key lies in whether there are good policies and planning to keep up with demand.

Supply Chain, Chinese Market, and Software Capabilities

Regarding the current severe memory bottleneck issue, Huang acknowledged its seriousness but stated that NVIDIA is fortunate to collaborate with all three HBM suppliers, who are all major customers and suppliers. The long-term relationships and careful planning give him confidence that "the situation will improve."

On the Chinese market, Huang was asked about the Chinese government's attitude towards introducing NVIDIA's latest AI chips (like H20). He responded that he has not communicated directly with the Chinese government, which typically expresses its intentions through companies. If Chinese companies are allowed to purchase and build solutions based on NVIDIA products, demand will be very strong, and currently, they are indeed seeing robust demand. This indirectly reflects communication from the Chinese side.

The interview also explored Siemens' software capabilities. Busch pointed out that Siemens has invested nearly $30 billion in building software capabilities, enabling it to create the most comprehensive physics-based digital twins. However, there is still space for improvement in areas such as operational software and factory operation software. For instance, in the life sciences sector, Siemens is shortening drug development cycles and reducing costs by acquiring and building data backbones (such as the Teamcenter community and Luma for Life Science). Siemens' unique advantages lie in its understanding of software as well as how to operate and automate physical facilities such as factories, buildings, and trains.

Future Vision: New Computing Platforms, Space Data Centers, and Autonomous Driving

Huang revealed details of the collaboration with Groq, indicating that it is not just a technology licensing agreement; NVIDIA has also recruited about 400 talented engineers from Groq. Groq's architecture focuses on low-latency token generation, excelling in inference, while NVIDIA specializes in training and inference. Their collaboration could pave the way for a "new realm" or "new platform" capable of solving future use cases, but he did not disclose specific details.

When asked about the possibility of discussing space data centers with Elon Musk, Huang did not confirm directly but believed it is a feasible technological platform. He pointed out that there is plenty of energy and cooling conditions in space, but system design would be drastically different from AI factories on Earth, although the chips themselves would be the same. Roland Busch analyzed from a business logic perspective: manufacturing any physical products that need to be returned to Earth (such as energy or hardware) in space is quite difficult, but producing "intelligence" (Tokens) can be easily transmitted back to Earth. Therefore, if something were to be produced from space, he would start with "intelligence."

Regarding autonomous driving, Huang was asked about his thoughts on Musk's response to his keynote speech. He first praised Tesla for having "the world's most advanced electric vehicle inventory and operation" and was "quite certain" that Tesla is using end-to-end technology. Regarding the differences in their technological paths, Huang pointed out that the NVIDIA DRIVE platform is also vision-based (with additional radar and LiDAR), so the approaches are quite similar. He believes Musk's approach represents state-of-the-art technology in autonomous driving robotics, and he would not criticize it but rather encourage Tesla to continue its good work.

Response to Social Issues: "Billionaire Tax" and Talent

As one of the most important employers and industry leaders in Silicon Valley, Huang was asked about the potential impact of California's proposed "billionaire tax" on the talent pool and industry. His answer was surprisingly nonchalant: "I'm perfectly fine with it." He stated that the choice to work in Silicon Valley is because of the talent pool there, and NVIDIA has offices worldwide to attract talent from various locations. Regarding California's tax policies, "Whatever it is, it is what it is. I've never thought about this issue." He emphasized that the key to attracting talent lies in the environment itself, rather than the tax burden.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。