德国总理对话黄仁勋：AI时代中国的优势，德国补得上吗？

德国总理对话黄仁勋：AI时代中国的优势，德国补得上吗？ — 王志安

王志安

AI-assisted summary of 德国总理对话黄仁勋：AI时代中国的优势，德国补得上吗？ — original source. Check the original for full context.

Source published:

Key ideas

  • NVIDIA invested nearly $1 billion in Germany over the past year and is finalizing deals with four additional companies, anchoring its presence with over 1,000 local employees. (15:00)
  • Chancellor Olaf Scholz announced that a demonstration Mercedes-Benz S-Class powered by NVIDIA architecture is scheduled to achieve autonomous driving on German roads by year-end. (3:48)
  • Jensen Huang identified China as the only global power possessing elite depth in both heavy manufacturing and computer science, contrasting it with the U.S. offshoring of physical craft. (8:22)
  • Huang characterized domestic industrial data as a sovereign natural resource, warning Germany not to export its core engineering knowledge abroad only to import back processed intelligence. (6:03)
  • The German Chancellor confirmed securing a regulatory carve-out in Brussels for the machinery sector, shielding physical industry from generic EU AI rules to foster innovation. (17:18)

Detailed notes

## Background This video was published on YouTube by the channel 王志安 on October 25, 2026. The description provides chapter timestamps detailing an in-depth dialogue between the German Chancellor and NVIDIA CEO Jensen Huang on AI integration in German industry, digital twins, global manufacturing competition, startup investments, and AI safety regulation.

## TL;DR In an in-depth exchange, NVIDIA CEO Jensen Huang and the German Chancellor discuss the transformative potential of artificial intelligence across Germany's industrial and manufacturing sectors. Huang emphasizes that the frontier of 'Physical AI' and digital twins creates a level playing field, where Germany's electromechanical craftsmanship and deep engineering heritage give it a distinct advantage alongside China and the United States. Both leaders stress the urgent imperative for Germany's Mittelstand to harness domestic sovereign data and utilize AI to preserve institutional knowledge amid demographic aging. The conversation also highlights NVIDIA's expanding direct investments into German startups and calls for balanced, sector-specific application regulation rather than sweeping EU restrictions.

## Atmosphere & style The dialogue is warm, collegial, and spirited, opening with playful banter about swapping jobs and wearing leather jackets before diving into high-stakes economic strategy. Huang brings the dynamic, visionary cadence of a Silicon Valley founder offering candid counsel, while the Chancellor projects pragmatic enthusiasm and political validation for domestic industry. The exchange carries an underlying current of shared urgency, balancing warnings about demographic decline with genuine optimism regarding Germany's engineering prowess.

## Bulletin - [15:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=900s) NVIDIA invested nearly $1 billion in Germany over the past year and is finalizing deals with four additional companies, anchoring its presence with over 1,000 local employees. - [03:48](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=228s) Chancellor Olaf Scholz announced that a demonstration Mercedes-Benz S-Class powered by NVIDIA architecture is scheduled to achieve autonomous driving on German roads by year-end. - [08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) Jensen Huang identified China as the only global power possessing elite depth in both heavy manufacturing and computer science, contrasting it with the U.S. offshoring of physical craft. - [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) Huang characterized domestic industrial data as a sovereign natural resource, warning Germany not to export its core engineering knowledge abroad only to import back processed intelligence. - [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) The German Chancellor confirmed securing a regulatory carve-out in Brussels for the machinery sector, shielding physical industry from generic EU AI rules to foster innovation.

## Key takeaways - [00:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=0s) German Chancellor Olaf Scholz opened the discussion by joking that if he swapped jobs with Jensen Huang, he would wear a leather jacket to Berlin and invest heavily—a move Huang affirmed he is actively executing. - [00:58](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=58s) NVIDIA's evolution from a PC gaming graphics firm to an enterprise workstation and automotive leader was catalyzed by its 2000 partnership with Germany's ELSA, followed by landmark alliances with Audi and Mercedes-Benz. - [02:19](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=139s) Huang outlined a dual-factory paradigm for future industry, where companies operate high-fidelity digital twins to simulate designs and build AI brains before physically assembling autonomous products. - [03:48](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=228s) Chancellor Scholz made a direct plea to Germany's Mittelstand to unlock their proprietary operational data for AI integration, framing industrial data utilization as crucial for economic resilience. - [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) National industrial records and operational histories should be treated as sovereign natural assets, with Huang cautioning against exporting domestic engineering values and importing back processed intelligence. - [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) Domain expertise in heavy machinery and electromechanical systems remains irreplaceable by general-purpose AI, positioning 'Physical AI' as an amplifier rather than a substitute for skilled engineers. - [08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) China represents a unique global competitor because its industrial transformation occurred alongside the rise of computing, granting it world-class depth in both physical manufacturing and software engineering. - [08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) The generative AI transition resets the software development paradigm, allowing computers to write their own code and reaffirming that digital mastery cannot supplant electromechanical manufacturing. - [11:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=678s) Faced with demographic decline, German manufacturers must urgently deploy AI as an institutional memory bank to capture, codify, and preserve the tacit craftsmanship of aging master workers. - [12:38](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=758s) Chancellor Scholz asserted that the United States made a fundamental strategic mistake by exporting its manufacturing base, stressing that Germany's economic strength depends on keeping physical production domestic. - [15:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=900s) NVIDIA deployed nearly $1 billion in German tech ventures over the preceding year, backing startups across robotics, aerospace, and foundation models such as Agile Robots and Black Forest Labs. - [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) Huang articulated a three-pillar framework for AI governance comprising technological invention, industrial engineering discipline, and application-specific rather than foundational-level regulation. - [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) Germany successfully negotiated a separate regulatory pathway within the European Union for its mechanical and machinery sectors, protecting traditional manufacturers from blanket AI constraints. - [22:27](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1347s) Roughly 50% of German startups founded over the past two years are centered on artificial intelligence, bolstered by government efforts to unlock pension and venture capital reserves.

## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=0s) Swapping Jobs and the Berlin Leather Jacket

The conversation opens on a lighthearted note when the moderator asks the German Chancellor what he would do first if he could trade jobs with Jensen Huang for just a single day. The Chancellor immediately leans into Huang’s signature persona, joking that he would throw on a black leather jacket, board a flight to Berlin, meet with the German Chancellor, and announce that he is ready to invest. Huang immediately replies with a smile: “Well, that's exactly what I'm doing.” The Chancellor expresses his gratitude for NVIDIA's commitment, noting that no one has invested more in Germany and calling it a tremendous signal of confidence in the country's economic future.

[00:58](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=58s) NVIDIA's Deep Roots in German Industry

Jensen Huang reflects on the long, intertwined history between NVIDIA and Germany, emphasizing that German partnerships fundamentally transformed the trajectory of his company. Back in the year 2000, NVIDIA was strictly a consumer-facing graphics company focused exclusively on PC gaming. A pivotal partnership with the German graphics hardware company ELSA opened the door to professional workstations, which subsequently connected NVIDIA to Germany's automotive industry across every phase of vehicle creation—from initial styling and design to simulation, engineering, and manufacturing.

Huang traces the automotive milestones that followed, noting how German automakers continually pushed NVIDIA deeper into the automotive stack. In 2005, Audi became the first automaker to bring NVIDIA's computing architecture directly inside the vehicle. Later, in 2020, Mercedes-Benz Chairman Ola Källenius and Huang formalized a landmark partnership to build next-generation software-defined, autonomous vehicles. Huang underscores that the bond between NVIDIA and German industry is not a recent development, but a deep relationship built over decades.

- **2000**: Partnership with German company ELSA shifts NVIDIA from PC gaming into workstations and automotive styling, design, and simulation. - **2005**: Audi integrates NVIDIA hardware inside the production automobile. - **2020**: Mercedes-Benz and NVIDIA establish an expansive partnership to develop autonomous vehicle platforms.

[02:19](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=139s) Factories of the Future and Digital Twins

When asked to paint a picture of an engineering office or industrial plant of the future, Huang envisions an industrial landscape centered entirely around dual operations: the physical factory and its digital counterpart. In this coming era, every manufacturing enterprise will operate two parallel facilities. The digital facility—built as a high-fidelity digital twin of the enterprise—will mirror reality with such fidelity that the boundary between the virtual and physical realms dissolves. Most German engineers and industrial workers will spend the bulk of their workdays inside this digital twin, designing virtual cars endowed with digital brains before anything is physically stamped or assembled.

Huang describes an ecosystem where artificial intelligence robots operate inside robotic factories, manufacturing physical cars that are themselves robots. In this software-defined, AI-driven paradigm, the nature of manufacturing shifts from bending metal to generating intelligence. Huang summarizes this structural shift: “So every company will build two things. They will build brains, and the brains that go into the products that they ship.”

```mermaid flowchart LR accTitle: The Dual-Factory Model accDescr: Diagram illustrating how the digital twin and physical factory interact to build intelligent products. subgraph Digital["Digital Twin (Virtual World)"] D1["Digital Engineering & Simulation"] --> D2["Digital Brain Development"] D2 --> D3["Simulated Robotic Assembly"] end subgraph Physical["Physical Factory (Real World)"] P1["Robotic Manufacturing Plant"] --> P2["Physical Autonomous Product"] end Digital -->|"Programs & Validates"| Physical ```

[03:48](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=228s) Bringing AI to the German Mittelstand and Automotive Sector

The discussion shifts to how the German government intends to translate cutting-edge AI into everyday industrial operations across Germany, from multinational conglomerates to small and medium-sized family businesses. The Chancellor highlights that concrete deployment is already underway, referencing a demonstration model of the new Mercedes-Benz S-Class that was presented to him, which is slated to achieve autonomous driving on German roads by the end of the year as one of the safest and most autonomous production vehicles in the world.

The Chancellor emphasizes that Germany’s ambition extends beyond the European Union; the federal government is fully committed to positioning the country as one of the most advanced hubs in the world for AI integration in physical industry. He identifies Germany's Mittelstand—the vast network of specialized small and medium-sized enterprises—as the nation's foremost competitive asset. Speaking directly to the Mittelstand business leaders in attendance, the Chancellor delivers a passionate plea: “Please. Open your doors for AI to make use of your data.” He stresses that capitalizing on the industrial data already housed within these companies is decisive for Germany's economic resilience, expressing strong optimism that uniting AI pioneers like NVIDIA with the Mittelstand and automotive suppliers will secure Germany's global leadership.

[06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) Sovereign Data, Domain Knowledge, and Physical AI

Huang strongly endorses the Chancellor’s perspective, framing domestic data as a vital natural resource equivalent to oil or minerals. He argues that a nation's accumulated data embodies its culture, societal values, historical experience, and hard-earned engineering knowledge. While Germany should actively import top-tier global AI tools and infrastructure to harvest this information, it must never surrender its sovereignty by sending its core knowledge abroad only to buy back processed intelligence. Huang states this principle plainly: “And you're absolutely right that you should import the technology to harvest that data and turn it into your national intelligence. But you should not export your culture, export your values, export your knowledge, export your history, and then import German intelligence back.”

Huang then identifies "Physical AI"—artificial intelligence capable of perceiving, reasoning about, and interacting with the physical world—as the next major frontier of technology. He asserts that general-purpose models will never supplant deep domain knowledge in heavy industry, robotics, and electromechanical engineering:

- **Data as a Sovereign Resource**: Domestic industrial records, operational histories, and institutional workflows belong exclusively to the home nation and should be turned into proprietary national intelligence. - **Physical AI vs. Pure Digital AI**: AI is moving into robotics, factory automation, and heavy machinery, domains where software must navigate the laws of physics. - **Inviolability of Domain Expertise**: As Huang puts it, “Artificial general intelligence will never replace domain expertise.” General models cannot intuit the precise physics and design tolerances of specialized electromechanical machinery; therefore, AI serves as an amplifying instrument rather than a replacement for skilled engineers.

[08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) Global Industrial Revolutions: China's Dual Strength vs. The AI Reset

Addressing Germany's standing in the broader geopolitical race, Huang contextualizes the AI revolution by comparing it to previous technological shifts. The electrification wave established three clear global industrial giants: Germany, the United States, and Japan. However, during the computing and digital revolution of the past fifty years, the United States largely offshored its heavy manufacturing, precision construction, and physical craft, resulting in China emerging alongside Germany and Japan as modern manufacturing titans.

Huang points out that China's unique competitive advantage stems from having undergone its industrial transformation precisely when the computer industry matured. Consequently, China is currently the only major power that possesses world-class depth in heavy manufacturing while simultaneously maintaining world-class depth in computer science and software engineering.

```mermaid flowchart TD accTitle: Evolution of Global Manufacturing Powerhouses accDescr: Flowchart tracing global leaders through electrification, computing, and the current AI reset. subgraph Electrification["First Major Wave: Electrification"] E["Industrial Leaders:<br/>Germany • United States • Japan"] end subgraph Computing["Last 50 Years: Computing Revolution"] C1["United States:<br/>Focus on Software & Digital<br/>Offshored Heavy Craft"] C2["Three Manufacturing Giants:<br/>Germany • Japan • China"] end subgraph Reset["Current AI Revolution: Physical AI Reset"] R["The Software Paradigm Shifts:<br/>Computers Program Themselves<br/>'Electrons will never replace atoms'"] end Electrification --> Computing Computing --> Reset ```

Despite China's head start in combining computing with manufacturing, Huang argues that the generative AI boom creates a clean slate. The prior fifty years required humans to write explicit lines of code, an arena dominated by American software firms. In the new era, computers write their own code based on demonstrated physical requirements, leveling the playing field. Huang stresses that digital mastery alone cannot replicate physical machinery: “All the digital technology in the world will not replace Electromechanical Physical Technology. Electrons will never replace atoms.”

[11:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=678s) Preserving German Craftsmanship and Addressing Aging Labor

Because physical products and electromechanical engineering cannot be replaced by software, Huang urges Germany to seize the current moment with absolute urgency. He warns against complacency, noting: “This industrial revolution is so mighty, is so powerful, you don't have another chance.” The biggest immediate threat confronting the German Mittelstand is not technological obsolescence, but demographic decline. As seasoned master craftspeople and engineers age out of the workforce, decades of tacit knowledge risk vanishing entirely.

Huang urges German business leaders to deploy artificial intelligence as an institutional memory bank. By using AI to capture, teach, and codify the specialized domain skills of veteran workers, Mittelstand companies can preserve their craft domestically and scale it across future generations. Referencing the Chancellor's own terminology, Huang calls on German industry to “re-engineer Germany” around this union of physical craftsmanship and machine intelligence.

[12:38](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=758s) Bridging Innovation to Implementation and Anchoring Manufacturing

The discussion turns to the critical task of bridging the gap between high-level AI research and practical implementation across factory floors and supply chains. The Chancellor affirms that deep collaboration with technology leaders like NVIDIA is an integral pillar of Germany's strategy. Hearing an American technology leader validate Germany's industrial strategy gives him immense optimism, reinforcing his long-standing political stance that Germany must remain an industrial, manufacturing-driven nation.

The Chancellor contrasts Germany's strategy with that of the United States, arguing that Washington made a fundamental strategic error over recent decades by letting its heavy manufacturing base erode. While the American tech sector is extraordinarily successful, its domestic physical manufacturing footprint has become precariously narrow. The Chancellor concludes that retaining the position of the world's third-largest economy hinges on keeping this broad industrial core intact, asserting: “And I believe and I share your view that the Americans made a big mistake to export their manufacturing.” Huang nods in agreement, opening his response by playfully acknowledging his own vantage point as the leader of a major American corporation.

[15:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=900s) Investing in Germany and the Startup Ecosystem

Jensen Huang traces NVIDIA’s deep entanglement with the German industrial landscape back through a chain of interlocking corporate relationships. He explains that NVIDIA’s initial collaboration with Audi naturally paved the way for its partnership with Mercedes-Benz, which in turn spurred broader investments in robotics and led directly to backing Agile Robots. Beyond robotics, Huang notes that NVIDIA has backed an array of German ventures, including Avelios, Reflex Aerospace, and generative AI pioneer Black Forest Labs. To Huang, these bets are entirely rational: Germany possesses a rare concentration of deep science, engineering, and mathematics that forms the bedrock not only of classical manufacturing, but of modern artificial intelligence. He applauds the government’s explicit commitment to cultivating startup culture, drawing a direct parallel to how the United States continually reinvigorates itself through entrepreneurship.

Putting capital behind that conviction, Huang highlights the scale of NVIDIA’s financial footprint across the country, noting that the company poured nearly a billion dollars into Germany over the past year alone and is closing investments in four additional companies that very week. Joking that NVIDIA has effectively turned into one of the country's most active venture capital firms, Huang playfully taunts the broader financial community: “And so for all of the other venture capitalists, you better hurry, because otherwise I'm going to take all the good deals and I'm going to be rich.” Beyond external investments, NVIDIA employs roughly one thousand people locally, with its Berlin office serving as the birthplace for several foundational corporate inventions, reinforcing Huang’s thesis that German technical and scientific talent is genuinely world-class.

- **Venture portfolio in Germany:** Strategic investments span robotics, health, aerospace, and foundation models, including Agile Robots, Avelios, Reflex Aerospace, and Black Forest Labs. - **Capital deployed:** NVIDIA invested nearly $1 billion in Germany over the preceding year, alongside four new company investments finalized that week. - **Local presence:** Over 1,000 NVIDIA employees are based in Germany, with key foundational architectural inventions originating from the Berlin engineering office.

---

[17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) AI Safety, Engineering, and Balanced Regulation

The Chancellor turns the floor to regulatory strategy, noting the fierce ongoing policy debates in Brussels versus the United States over how to govern emerging technologies without stifling them. When asked how to calibrate rules to mitigate damage while avoiding overregulation, Huang insists that safety is not an impediment to capability but its prerequisite: “It is not a choice. You don't choose between safety and capability. They are one.” To illustrate, Huang lays out a three-pillar safety framework. The first pillar is technological invention itself; regulators cannot enforce standards if engineers have not built the underlying safety mechanisms, citing his joint work with Mercedes-Benz Chairman Ola Källenius to pioneer functionally and actively safe autonomous driving systems. Pointing to German engineering rigor, Huang stresses, “Safety is not hopes and dreams. Safety is technology and engineering.”

The second pillar centers on rigorous corporate engineering methodology and established industrial best practices, which fast-growing young startups must adopt rather than repeating avoidable errors. The third pillar is smart, application-level governmental oversight. Huang draws an analogy to the early web: while regulating the abstract infrastructure of the internet is almost impossible, governing specific applications is straightforward. Rather than paralyzing base models with sweeping restrictions, governments should empower existing regulatory authorities to incorporate AI within specific sector products, ensuring the domestic market thrives while operating responsibly. The Chancellor strongly affirms this approach, sharing that he successfully fought in Brussels to exempt the German mechanical and machinery industry from blanket European Union AI constraints, establishing specialized sector-specific rules instead so traditional industrial champions retain room to innovate.

```mermaid flowchart TD accTitle: Jensen Huang's Three Pillars of AI Safety accDescr: A structural breakdown showing the three foundational requirements for effective AI safety and regulation. subgraph Pillars["Framework for AI Safety"] P1["1. Technology Invention<br/>Engineers build active and functional safety into models"] P2["2. Industrial Best Practices<br/>Deep methodology and engineering discipline for startups"] P3["3. Application-Level Regulation<br/>Sector bodies govern specific deployments rather than base tech"] end P1 --> BalancedSafety["Safe, Responsible, and Thriving AI Ecosystem"] P2 --> BalancedSafety P3 --> BalancedSafety ```

- **Safety and performance unity:** High-stakes automated systems (such as autonomous vehicles) cannot exist commercially without provable functional safety. - **Application over infrastructure:** Following the model of the internet, regulatory intervention is most effective when targeted directly at concrete end-use applications rather than abstract foundational tech. - **EU regulatory carve-out:** The Chancellor secured a separate regulatory path within the EU for Germany’s machinery and manufacturing sector, shielding it from generic AI limitations to “give this industry a chance.”

---

[22:27](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1347s) Looking Ahead: Hopes for Next Year and Germany’s AI Future

Looking toward future collaboration, the moderator asks the Chancellor what benchmarks he hopes to see when Huang visits Germany again in a year. The Chancellor expresses a desire for regular annual check-ins, joking that his primary goal is to see Huang sitting in his office genuinely thrilled with Germany as an investment destination. Huang enthusiastically agrees, quipping, “Nothing would give me more joy than to see all of my investments in Germany turn into gold.”

Huang commends the Chancellor's political backing for structural reforms that unlock pension fund reserves and venture risk capital for domestic tech companies. He notes that the entrepreneurial momentum is already palpable on the ground, pointing out that approximately half of all new startups founded in Germany over the past two years are centered on artificial intelligence. The session closes on a humorous note from the moderator, who observes that several thriving German startup founders were once interns at NVIDIA, suggesting that when Huang returns next year, he should bring a pen to collect autographs from his former trainees.

- **Venture policy reform:** Government initiatives aim to mobilize domestic institutional capital, particularly pension assets and risk funds, into German tech startups. - **AI startup surge:** Approximately 50% of all newly founded German startups over the last two years are artificial intelligence enterprises. - **Talent pipeline:** Several current German AI founders trace their roots directly back to internship and engineering tenures at NVIDIA.

## Notable quotes > [08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) All the digital technology in the world will not replace Electromechanical Physical Technology. Electrons will never replace atoms. — Jensen Huang

> [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) Artificial general intelligence will never replace domain expertise. — Jensen Huang

> [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) Safety is not hopes and dreams. Safety is technology and engineering. — Jensen Huang

> [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) It is not a choice. You don't choose between safety and capability. They are one. — Jensen Huang

## Mentioned Jensen Huang, German Chancellor, Ola Källenius, NVIDIA, ELSA, Audi, Mercedes-Benz, Mercedes-Benz S-Class, Agile Robots, Avelios, Reflex Aerospace, Black Forest Labs

## Video description 章节 00:00 片头 00:06 交换一天工作 00:58 英伟达与德国的合作 02:19 未来工厂与数字孪生 03:48 让AI走进德国工业 06:03 数据、领域知识与物理AI 08:22 全球制造业与新一轮AI机会 13:25 德国的制造业基础 14:59 投资德国与创业生态 17:18 AI安全与监管 22:27 一年后的期待 24:06 来源与片尾

[Source](https://www.youtube.com/watch?v=dIi3X6VJ3v8)

中文简报

德国总理对话黄仁勋：AI时代中国的优势，德国补得上吗？ 在一次深入交流中，英伟达首席执行官黄仁勋与德国总理探讨了人工智能在德国工业和制造业领域的变革潜力。黄仁勋强调，“实体AI”与数字孪生前沿创造了一个公平的竞争环境，德国的机电工艺与深厚的工程传承使其与中国和美国并驾齐驱，具备独特优势。两位领导人均强调，在人口老龄化背景下，德国中小型企业迫切需要利用本土主权数据，并通过AI保存机构知识。对话还重点提及英伟达对德国初创企业不断扩大的直接投资，并呼吁推行平衡且针对具体行业的应用监管，而非欧盟一刀切式的限制。 英伟达过去一年在德国投资近10亿美元，目前正敲定与另外四家公司的交易，并通过1,000多名本土员工巩固其在当地的业务布局。 (15:00) 德国总理奥拉夫·朔尔茨宣布，一辆搭载英伟达架构的梅赛德斯-奔驰S级演示车计划于年底前在德国道路上实现自动驾驶。 (3:48) 黄仁勋指出，中国是全球唯一在重工业制造和计算机科学两大领域均拥有顶尖深度的强国，这与美国将实体制造外包的做法形成鲜明对比。 (8:22) 黄仁勋将本土工业数据定性为主权自然资源，警告德国不要将核心工程知识出口到国外，再倒退进口经过加工的智能。 (6:03) 德国总理证实已在布鲁塞尔为机械行业争取到监管豁免，使实体工业免受欧盟通用AI规则的限制，以促进创新。 (17:18)

Reading summary

In a joint dialogue, the German Chancellor and NVIDIA CEO Jensen Huang argue that the global shift toward "Physical AI" provides Germany a vital opportunity to leverage its electromechanical and manufacturing base, counter demographic labor shortages through institutional AI memory, and protect competitiveness through sovereign data and application-level regulation. Physical AI and Sovereign Data (6:03) Huang identifies "Physical AI"—systems capable of perceiving, reasoning about, and interacting with the physical world—as the next major technology frontier, asserting that artificial general intelligence will never replace deep electromechanical domain expertise because "electrons will never replace atoms." Comparing this wave to past industrial revolutions, Huang observes that while China currently possesses the unique advantage of pairing world-class heavy manufacturing with advanced computer science, the generative AI reset creates a clean slate by enabling computers to write their own software from physical requirements rather than relying on human coding moats. To capitalize on this shift, Huang argues that domestic industrial data must be treated as a sovereign natural resource equivalent to oil or minerals. He stresses that while Germany should actively import advanced foreign AI infrastructure to process this information, it must avoid exporting its cultural, engineering, and historical knowledge only to repurchase it as foreign-processed intelligence. Dual Factories and Mittelstand Demographics (2:19) Huang outlines an industrial manufacturing model centered on dual operations: high-fidelity digital twins operating alongside physical robotic facilities. In this paradigm, engineers spend the majority of their time in virtual simulations designing digital brains and testing robotic assembly before any physical production occurs, shifting industrial work from bending metal to generating intelligence. Both speakers view Germany's Mittelstand—its network of specialized small and medium-sized enterprises—as the nation's core competitive asset, prompting the Chancellor to urge business leaders to open their proprietary operational data to AI systems. Huang warns that the Mittelstand's most immediate threat is demographic decline rather than technological obsolescence; as veteran craftspeople retire, companies must deploy AI as an institutional memory repository to capture, codify, and preserve tacit engineering techniques for future generations. Industrial Anchoring Versus Manufacturing Offshoring (12:38) The Chancellor contrasts Germany's long-term economic model with that of the United States, asserting that Washington made a fundamental strategic error over recent decades by offshoring its heavy manufacturing base and precariously narrowing its domestic physical footprint. He maintains that retaining Germany's position as the world's third-largest economy depends on keeping this broad industrial core intact rather than relying exclusively on purely digital technology. Huang agrees with this industrial emphasis, pointing to a multi-decade partnership between NVIDIA and German industry that evolved from graphics hardware maker ELSA in 2000 to Audi's in-vehicle computing integration in 2005 and Mercedes-Benz's autonomous platform partnership in 2020. The Chancellor notes that practical implementations are already emerging, highlighting a demonstration Mercedes-Benz S-Class slated to achieve autonomous driving on German roads by the end of the year. Capital Inflow and Application-Level Regulation (17:18) Reinforcing his confidence in German engineering and mathematics, Huang highlights that NVIDIA invested nearly $1 billion in Germany over the preceding year—backing ventures including Agile Robots, Avelios, Reflex Aerospace, and Black Forest Labs—while employing roughly 1,000 local staff. He notes that approximately half of all German startups founded over the past two years focus on AI, supported by government initiatives designed to unlock domestic pension reserves and risk capital for technology firms. On governance, Huang outlines a three-pillar safety framework comprising technological invention, corporate best practices, and targeted oversight. He advocates regulating concrete, sector-specific end applications rather than paralyzing foundational models with blanket restrictions. The Chancellor affirms this regulatory approach, noting that he secured a separate carve-out in Brussels for Germany's machinery and manufacturing sector to shield traditional industrial champions from generic European Union AI constraints. While both leaders argue that Germany is uniquely positioned to bridge physical engineering with machine intelligence, the practical execution of this strategy hinges on whether traditional Mittelstand firms will readily share proprietary operational data and whether public capital reforms can mobilize enough domestic investment to compete with global tech ecosystems.

精炼阅读摘要

在对话中，英伟达首席执行官黄仁勋与德国总理共同探讨了实体制造业与人工智能的深度融合路径。核心共识指出，虽然中国凭借重工业制造与计算机软件的双重深度在数字化时代建立起强大优势，但以“物理AI”为特征的下一代工业革命为德国实体经济提供了重置竞争格局的历史机遇；德国必须立足其机械制造底蕴与中小型企业的数据主权，构筑软硬件协同的工业护城河。 物理AI与双工厂：制造本质转向生成智能 (2:19) 黄仁勋在描绘未来工业图景时提出，制造业正迈入物理工厂与高保真“数字孪生（Digital Twin）”并行运营的时代。未来工业企业将在虚拟孪生系统中完成数字汽车设计、数字大脑开发及机器人产线模拟，随后再导入实体工厂完成生产。黄仁勋强调，工业制造的核心逻辑已由过去的“折弯金属”根本性转向“制造智能大脑，并将大脑装载入物理产品中”，未来产线将由AI机器人来制造身为机器人的自主化终端。 回顾历史合作，黄仁勋指出英伟达与德国工业的绑定跨越了数十年：2000年通过与德国图形硬件厂商ELSA的合作，英伟达从消费级PC游戏芯片切入专业工作站与汽车造型设计；2005年奥迪首次将英伟达计算架构搭载入量产汽车；2020年梅赛德斯-奔驰与英伟达展开全方位战略合作开发软件定义自动驾驶平台。针对产业落地，德国总理现场证实，配备高阶自动驾驶技术的全新梅赛德斯-奔驰S级示范车型正在加紧验证，计划在年内于德国道路上正式开跑。 数据主权与专有知识：破解老龄化技能断层 (6:03) 针对德国中小型企业（Mittelstand）的转型，德国总理向本土企业家发出明确呼吁，敦促其彻底开放沉淀多年的工业数据资产以拥抱AI。黄仁勋高度支持该主张，并将国家积累的工业与运营数据比作本国的自然资源。他强调，数据承载着德国的工程文化、历史经验与产业智慧；德国应当积极引入全球顶尖的算力技术提炼本土数据，但绝不能将原始专有数据流失海外、再付费买回被二次加工的智能产物，必须牢牢守住“数据主权”。 黄仁勋进一步指出，通用人工智能（AGI）不可能替代机械与自动化领域的垂直专业知识（Domain Expertise），因为软件模型无法凭空推导复杂的物理定律与精细机电公差。面对德国制造业老龄化加剧、资深技师退休导致隐性工艺断层的严峻危机，他建议中小型企业将AI部署为企业内部的“机构记忆库”，通过数字化模型固化、传授与扩展老一代工程师的精湛手艺，以机器智能锁住德国的核心工匠技艺。 地缘工业演进：中国双重优势与物理AI重置 (8:22) 在梳理全球工业变迁的历史脉络时，黄仁勋分析认为，早期的电气化革命确立了美国、德国与日本三大制造巨头；然而在过去半个世纪的计算机革命中，美国将大量重型精密制造与实体工艺外包，促成了中国崛起并与德、日并列为制造业强国。黄仁勋指出，中国的核心竞争力在于其工业化进程与全球计算机时代完全同步，因而成为当今唯一兼具世界级实体制造纵深与世界级计算机工程软件实力的超级大国。 尽管中国在“软硬兼备”上抢占先机，但黄仁勋认为生成式AI与物理AI的爆发正在拉开一场全新的工业洗牌。以往五十年软件行业依赖人工编写复杂代码，由美国软件企业独占鳌头；如今计算机能够依据物理反馈自我生成代码，而纯数字技术无论多么先进，其“电子”也永远无法凭空取代“原子”与实体机电硬件。德国总理认同该观点，直言美国此前大规模剥离外包重工业是重大战略失误，并明确指出德国若要稳固全球第三大经济体地位，就必须坚决捍卫完整的工业实体制造根基。 资本与监管平衡：应用级治理与实体产业赋能 (15:00) 在资本投入层面，黄仁勋透露英伟达在过去一年中向德国投入了近10亿美元，并在本土雇用了逾千名员工，其柏林办事处更是诸多底层技术创新的发源地；其投资版图已覆盖Agile Robots（机器人）、Avelios（医疗健康）、Reflex Aerospace（航天航空）以及Black Forest Labs（生成式基础模型）等德国高科技初创企业。黄仁勋积极评价德国政府盘活养老金与风险基金以激活创业生态的政策举措，并援引数据称近两年来德国约半数新创公司均围绕人工智能展开。 针对欧美在监管尺度上的激烈拉锯，黄仁勋提出了包含“底层技术发明”、“行业工程最佳实践”与“应用级精准监管”的AI安全三支柱框架。他强调安全并非阻碍能力的成本，而是商业落地的核心前提，监管应参照互联网治理逻辑聚焦终端产品部署而非一刀切限制基础架构。对此，德国总理透露，他在布鲁塞尔积极斡旋并成功为德国机械与机电行业争取到了专门通道，避免其受到欧盟宽泛AI法案的过度束缚，确保传统支柱产业保留充足的创新与试错空间。 该对话表明，德国在AI时代的战略支点并非盲目复制硅谷式的纯软件互联网模式，而是通过技术互补将深厚的机电工程底蕴转化为不可替代的“物理AI”护城河。不过，尽管德国正通过监管松绑与长线资本改革加速孵化AI生态，但如何在劳动力老龄化危机倒逼之下，切实促使作风保守的德国中小型企业跨越数据安全顾虑、高效完成垂直领域的算法提炼，仍是该国工业模式能否顺利突围的关键考验。

Detailed notes — zh

## Background 本内容由王志安于 2026 年 10 月 25 日发布在 YouTube 平台。视频聚焦德国总理与英伟达 CEO 黄仁勋的高峰对话，围绕未来工厂、数字孪生、物理 AI、全球制造业格局重塑以及欧洲 AI 监管与投资生态等核心议题展开深入探讨。 ## TL;DR 对话中，黄仁勋与德国总理深入剖析了工业 AI 的演进路径，强调通用大模型无法替代深厚的垂直工程底蕴，数字孪生与物理 AI 将成为未来工业制造的核心命脉。黄仁勋提出“电子无法取代原子”，指出全球技术正迎来全面重置，呼吁德国利用 AI 将老一辈工匠的工程经验转化为数字资产，破解老龄化困境。德国总理重申德国坚守制造业立国的战略定力，反思了将重工业外包的产业教训，并呼吁中小型企业全面拥抱数据赋能。双方还系统阐述了 AI 安全监管的“三大支柱”框架，并对英伟达在德近 10 亿美元的投资布局及本土活跃的 AI 创业生态表达了高度信心。 ## Atmosphere & style 整场对话开场轻松风趣，两位嘉宾在互换工种的幽默调侃与皮夹克笑谈中迅速拉近距离，随后层层递进切入宏大厚重的工业战略思考。现场洋溢着顶尖科技领袖与国家掌舵人之间高度同频的智力碰撞，既有对历史工业转移与美欧产业模式的坦诚反思，也充盈着时不我待的紧迫感与昂扬乐观的行动魄力。交谈节奏张弛有度，既有对硬核工程与物理规律的严谨敬畏，又在资本喊话与初创人才调侃中展现出真挚而富有人情味的现场张力。 ## Bulletin - [15:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=900s) 英伟达过去一年在德投资近10亿美元，并在柏林拥有约千人工程团队，投资版图涵盖梅赛德斯-奔驰、Agile Robots及Black Forest Labs等企业。 - [10:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=600s) 黄仁勋提出“电子永远无法取代原子”，指出AI可自我编程的时代将重置过去50年的软件红利，物理工业专长将成为下半场核心竞争力。 - [08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) 中国是当今全球唯一兼具深厚制造业纵深与强大计算机科学基础的国家，而德国则需利用物理AI固化其面临老龄化威胁的工匠技艺。 - [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) 针对欧洲AI监管难题，黄仁勋提出技术工具、工程规范与分类终端应用监管三大支柱，德国总理透露已在欧盟立法中为本土机械制造争取到豁免空间。 ## Key takeaways - [00:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=0s) 黄仁勋回顾英伟达自2000年与德国ELSA合作切入工作站起，已深度融入德国汽车工业的设计、仿真、制造及自动驾驶全周期。 - [02:19](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=139s) 未来企业将由物理工厂与数字孪生工厂共同构成，大部分员工将在虚拟环境中设计大脑并完成验证，再由自动化实体工厂交付。 - [03:48](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=228s) 德国中小型企业（Mittelstand）积累的海量工业与机电数据是国家走向未来的核心资产，亟需结合尖端算力完成智能化改造。 - [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) 数据与工业知识属于不可剥夺的国家资产，各经济体应避免出口本土知识与文化后再进口外包的AI智能。 - [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) “物理AI”（Physical AI）代表人工智能与实体世界交互的未来，通用智能永远无法替代特定细分领域的深度机电专门知识。 - [08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) 全球工业格局从电气化时代的美德日三强演变为信息时代的德日中三强，美国大规模外包制造业为其留下了工业基础狭窄的战略教训。 - [08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) 中国因经济腾飞恰逢计算机时代，形成了全球独一无二的制造业与计算机科学双重优势。 - [10:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=600s) AI自动编程正在重置纯软件的先发优势，德国必须抓住时间窗口利用AI将即将退休的老工匠经验与工艺代码化，阻断技术断代危机。 - [13:25](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=805s) 德国总理明确坚守制造业立国之本，认为制造业实体是德国参与全球AI竞争并区别于美国发展路径的最大底牌。 - [15:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=900s) 德国深厚的基础科学和高阶数学积淀，为英伟达发展物理AI和先进制造生态提供了不可替代的人才支撑。 - [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) 黄仁勋建议AI监管借鉴互联网经验，将监管重心放在垂直终端应用上，而非直接限制底层的通用算力和技术。 - [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) 安全是技术与工程问题而非空泛口号，企业必须在算法创新与严格的工程实践中实现功能安全，而非在安全与能力之间做单选。 - [22:27](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1347s) 德国政府推动养老基金进入风险投资领域，助力本土创新生态爆发，过去两年新成立的德国初创公司中半数以上为AI企业。 ## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=0s) 换岗一天的玩笑与英伟达的德国渊源 对话在一段轻松幽默的互动中开场。主持人向德国总理提问，如果能与英伟达创始人兼 CEO 黄仁勋（Jensen Huang）互换一天工作，他首先会做什么。总理打趣地拒绝了换岗的提议，黄仁勋则接过话茬笑着设想了自己的版本：“我会穿上我的皮夹克，坐飞机去柏林，去见德国总理，并告诉他我已经准备好投资了。”总理随即回应，这正是黄仁勋当下正在做的事情，并对英伟达在德国的巨额投资表达了真挚的谢意：“没有人比你在德国投资更多了。非常感谢，这对我们国家是一个绝佳的信号。” 黄仁勋借此深情回顾了英伟达与德国制造业数十年的紧密合作。在 2000 年，英伟达还只是一家面向消费级市场的技术公司，专注于 PC 游戏硬件；正是一家名为 ELSA 的德国公司与之建立伙伴关系，才将英伟达带入了工作站领域，并由此敲开了德国所有汽车制造商的大门。 ```mermaid flowchart LR accTitle: 英伟达与德国汽车工业合作演进 accDescr: 展示英伟达从2000年PC游戏显卡走向工作站、车载系统及自动驾驶的历程。 A["2000年: PC游戏技术"] --> B["携手德国ELSA公司<br/>切入工作站领域"] B --> C["2005年: 奥迪引入<br/>英伟达进入车内系统"] C --> D["2020年: 联手康林松<br/>共同打造自动驾驶汽车"] D --> E["覆盖汽车全生命周期<br/>造型/仿真/制造/整车"] ``` - 2000 年：与德国 ELSA 合作，英伟达成功从消费级游戏显卡扩展至专业工作站及汽车产业客户。 - 2005 年：奥迪（Audi）首次将英伟达技术带入汽车座舱内部。 - 2020 年：英伟达与梅赛德斯-奔驰董事会主席康林松（Ola Källenius）正式建立战略伙伴关系，联合开发下一代自动驾驶汽车。 - 当前深度整合：英伟达已渗透至德国车企的造型设计、工程仿真、工厂制造直至车载计算的完整生命周期。 ### [02:19](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=139s) 未来工厂与数字孪生：虚拟映射现实 主持人随后将话题转向工业未来的构想，请黄仁勋描绘未来工程办公室或工厂的真实形态。黄仁勋描绘了一幅“双重工厂”的图景：未来的每家企业都将同时拥有两座工厂——一座是物理世界的实体工厂，另一座则是完全对应的数字工厂（Digital Twin）。 黄仁勋指出，数字孪生将逼真到与现实无异的程度。未来大部分德国工业员工将主要在企业的数字孪生环境中工作：工程师在虚拟空间内设计数字汽车，赋予其数字大脑；软件与 AI 控制的机器人在数字工厂中模拟装配；完成验证后，实体工厂再通过高度自动化的软硬件系统制造出同样具备智能大脑的实体车辆。黄仁勋概括道：“每家公司都将制造两样东西：他们将制造大脑，以及装载这些大脑并最终交付的产品。” ### [03:48](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=228s) 德国工业拥抱 AI：中小型企业（Mittelstand）的独特机遇 主持人向德国总理提问，德国政府正采取哪些具体举措，确保 AI 能够深入从大型集团到中小企业的广泛工业日常之中。总理透露，梅赛德斯-奔驰刚刚向他展示了全新 S 级轿车的模型，并承诺该车型将在当年年底前在德国实现最安全、自动化程度最高的自动驾驶落地，这有力证明了前沿技术正直接发生在德国本土。 总理着重强调，德国在全球范围内最强大的优势在于其中小型企业群体——即著名的“中小型企业”（Mittelstand）。他当场向在座的德国中小企业主发出呼吁：“请向 AI 打开大门，充分利用你们的数据。”他指出，德国工业长年积累的数据是决定德国未来走向的关键财富，如果能将英伟达这样的顶级 AI 算力与德国庞大的制造业底蕴（涵盖机械制造、汽车及零部件供应链）相结合，德国完全有信心成为全球工业 AI 普及度最高的领军国家。 ### [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) 数据主权、领域知识与物理 AI 的不可替代性 黄仁勋高度赞同德国总理对于数据价值的判断，并从国家战略高度提出了“数据即自然资源”的论断。他强调，德国的数字化经验与工业数据承载着本土的文化、价值观、工艺知识与历史积淀，属于不可剥夺的国家资产。德国应当引进全球最顶尖的 AI 技术工具来开采和提炼这些数据，转化为国家智能（National Intelligence），但绝不能本末倒置。黄仁勋犀利地指出：“你绝不应该出口你的文化、出口你的价值观、出口你的知识、出口你的历史，然后再把德国智能进口回来。这毫无意义。” 谈及人工智能的发展脉络，黄仁勋明确提出了下一波浪潮——“物理 AI”（Physical AI）。这是 AI 真正与真实物理世界产生交互的阶段，涵盖了机器人学、工业制造、汽车与重工业。他向德国制造业注入强心剂，强调通用人工智能（AGI）永远无法抹杀行业专门知识： > “无论你的通用智能有多聪明，你永远不可能比那些在机电技术某一特定细分领域极度专业的人更聪明。它无法被取代。因此，AI 只能增强你的能力，而不会取代你的能力。” ### [08:22](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=502s) 全球工业竞争格局重塑：中美德日的历史位移 面对主持人和现场关于德国在全球 AI 竞赛中处于何种方位的提问，黄仁勋从宏观工业革命的历史演进切入进行了深度剖析。他指出，上一次重大工业革命是电气化时代，当时催生了三大地理中心：德国、美国和日本。而在过去 50 年间，以计算机和数字技术为主导的信息革命中，美国在很大程度上将重工业、制造业和核心装配工艺外包转移给了中国，使得当今全球制造业三强演变为德国、日本和中国。 ```mermaid flowchart TD accTitle: 全球工业强国演变与AI重置 accDescr: 分析电气化时代到计算机时代中美德日三足鼎立的迁移与AI带来的新一轮产业重置。 subgraph S1["电气化时代"] US1["美国"] DE1["德国"] JP1["日本"] end subgraph S2["计算机与外包时代（过去50年）"] US2["美国<br/>专注软件/编程<br/>外包重工业"] CN2["中国<br/>兼具计算机科学<br/>与深厚制造业"] DE2["德国<br/>坚守机电制造业"] JP2["日本<br/>坚守高精制造业"] end subgraph S3["物理AI时代（全面重置）"] AI["AI能够自我编程<br/>电子无法替代原子<br/>工艺专长+AI迎来重塑"] end S1 --> S2 S2 --> S3 ``` 黄仁勋客观分析了中国的独有优势：中国的经济转型与工业跃升恰逢计算机行业的黄金期，因此中国是当今世界唯一一个在“制造业纵深”与“计算机科学基础”两端都极其强大且驾轻就熟的国家。 ### [10:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=600s) 电子无法替代原子：以 AI 传承工匠底蕴与“重构德国” 虽然美国在过去半个世纪凭借纯软件编程取得了无与伦比的霸权，但黄仁勋指出，随着 AI 能够自我编程、由人类提供需求示范的新时代来临，过去 50 年积累的技术红利已经迎来全面“清零重置”（Complete Reset）。纯数字虚拟技术无论如何发展，都无法凭空制造出物理设备： > “世界上所有的数字技术都无法取代机电物理技术。电子永远无法取代原子。” 黄仁勋由此向德国发出紧迫警示：当前德国中小企业普遍面临人口老龄化和技术劳动力短缺的严峻挑战，老一辈工匠脑海中积累的高深手艺与机电专长正随着时间流逝而面临断代危机。德国必须抢抓这次不可逆的工业革命契机，利用 AI 将工人的隐性工艺知识和工程经验进行“代码化”与固化，把这些灵魂资产永远留在德国本土，以实现德国工业的彻底重构。 - 编程范式变革：从手写代码转向让 AI 领悟意图，数字领域的软件优势正在向拥有物理数据的工业领域大转移。 - 破解人口困境：利用 AI 学习并固化即将退休的资深工程师知识，阻断制造业技艺失传。 - 绝无仅有的机会窗口：这场工业革命势头磅礴，企业与国家必须抱定“不会再有第二次机会”的危机感迅速行动。 ### [13:25](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=805s) 坚守制造业战略定力与德美路径反思 听完黄仁勋的剖析，德国总理对德国的经济前景与工业模式表达了强烈的乐观。总理高度肯定英伟达作为全球科技巨擘所分享的洞察，并强调自己始终坚定主张德国绝不能放弃制造业立国之本： > “你可能知道，我一直坚持我们必须保持一个拥有制造业的国家，我们决不能放弃这一点。” 总理深刻反思了全球产业链变迁中的得失，直言他完全赞同黄仁勋的判断：美国当年将大量制造业大规模外包是一个重大战略失误，导致其如今本土工业根基过于狭窄单一。相比之下，德国始终保有完整的工业纵深和深厚的机电制造业基础，而这正是德国在 AI 时代与尖端科技结合、参与全球最高水平经济竞争的最大底牌。面对德国总理对美欧产业差异的直率论述，黄仁勋作为一家美国顶尖科技公司的掌门人，以会心的回应承接了关于未来跨国工业协作的讨论。 ### [15:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=900s) 投资德国：从汽车产业到AI创投生态 黄仁勋详细回顾了英伟达在德国的深度合作与投资版图。他指出，英伟达与德国企业的合作是双赢且水到渠成的：最早与奥迪（Audi）的合作延伸到了梅赛德斯-奔驰（Mercedes-Benz），而与奔驰在自动驾驶领域的协同又促成了对机器人产业的投资，进而推动了英伟达对敏捷机器人公司（Agile Robots）的注资。此外，英伟达还投资了 Avelios、Reflex Aerospace 以及去年备受瞩目的生成式 AI 初创团队 Black Forest Labs。黄仁勋赞叹德国拥有无与伦比的基础科学与深厚数学底蕴，而这正是现代制造业、前沿科学以及人工智能的基石。 谈及创业生态，黄仁勋格外赞赏总理支持与推动初创企业发展的政策主张。他谈到，美国经济之所以能够不断自我重塑与焕发生机，核心动力就在于源源不断的企业家群体。过去一年里，英伟达在德国的投资已接近 10 亿美元，且就在对话当周还在投资四家本土新公司。黄仁勋开玩笑地对台下其他风险投资机构喊话：英伟达在很多方面已是德国最大的 VC 之一，“其他风险投资人最好动作快点，否则所有好项目都会被我抢光，然后我就发大财了”。目前英伟达在柏林设有大型办公室，本土员工达千人规模，英伟达诸多核心发明正是出自这批优秀的德国工程师团队之手。 - **重点投资与合作脉络**：从传统车企奥迪、奔驰，延伸至机器人公司 Agile Robots，前沿创投覆盖 Avelios、Reflex Aerospace 以及文生图模型团队 Black Forest Labs。 - **资金与团队规模**：过去一年在德投资近 10 亿美元，对话当周新增投资 4 家企业；英伟达在柏林等地拥有约 1000 名员工。 - **底层优势**：德国在深层科学（Deep Science）与高阶数学领域的积淀全球领先，为物理 AI 和先进制造提供了不可替代的人才库。 --- ### [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) AI安全与监管的三大支柱：技术、最佳实践与应用监管 德国总理向黄仁勋抛出一个棘手难题：当前布鲁塞尔与全欧正在就 AI 监管展开激烈辩论，各方观点分歧巨大，缺乏监管不可行，过度严苛的规管同样会扼杀发展，在兼顾潜在技术危害的同时，欧洲究竟该如何找到最优解？黄仁勋直言“安全是重中之重”，并明确指出：“选择并不存在，你无法在安全与能力之间二选一，它们本为一体。”针对 AI 治理，黄仁勋系统提出了“三大支柱”框架，并建议欧洲参考互联网治理经验，重点监管具体领域的终端应用，而非封死基础技术本身。 ```mermaid flowchart TD accTitle: AI 安全与监管的三大支柱 accDescr: 展示黄仁勋提出的技术创新、企业最佳实践以及针对终端应用监管的三层框架。 subgraph Pillar1 ["第一支柱：底层技术"] T1["发明安全技术与算法"] T2["实现功能安全与主动安全验证"] end subgraph Pillar2 ["第二支柱：行业实践"] P1["建立深厚工程方法论"] P2["初创企业遵循最佳工程实践"] end subgraph Pillar3 ["第三支柱：分类监管"] R1["避免单纯规管基础算力/技术"] R2["针对垂直终端应用快速立法"] end Pillar1 --> Pillar2 Pillar2 --> Pillar3 Pillar3 --> OUT["在安全可控的前提下确保工业创新繁荣"] ``` 黄仁勋阐述道，首先必须通过硬核工程来发明保障安全的技术工具，正如他和奔驰 CEO 康林松（Ola Källenius）为打造最安全的自动驾驶系统所做的那样，因为“安全不是空想与口号，安全是技术与工程”。其次，企业自身必须践行严密的工程方法论与质量规范，避免年轻团队因野蛮生长犯下不必要的低级错误。最后才是政府与行业协会的法规介入。黄仁勋提醒，监管互联网很难，但监管互联网的具体应用却相对清晰，AI 亦是同理。 德国总理对此深表认同，并分享了欧洲立法的现实处境。总理强调，自己更倾向于在全球或至少在欧盟层面形成兼顾安全与产业活力的标准，目前欧洲与美国的监管哲学仍存差距，但他已成功在欧盟层面的立法博弈中为德国的机械制造与工业装备争取到了特例空间，使机械制造行业免受统一泛化 AI 框架的过死束缚。总理呼吁：“给德国一个机会，给这个产业一个机会。” > “安全不是空想与口号，安全是技术与工程。” —— 黄仁勋 > “给德国一个机会，给这个产业一个机会。” —— 德国总理 --- ### [22:27](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1347s) 一年后的展望与德国创投生态的未来 在对话尾声，主持人询问总理，如果一年后黄仁勋故地重游，他最希望看到怎样的局面。总理风趣地回应，他希望与黄仁勋保持常态化沟通，至少每年在总理办公室见一次，共同探讨合作进展，更希望看到黄仁勋坐在自己的办公室里，对德国这一投资热土由衷感到高兴与振奋。 黄仁勋随即幽默接话表示，看到自己在德国的所有投资全部“变成金子”，将是他最大的快乐。黄仁勋进一步盛赞德国政府在打通养老基金与风险资本通道上的务实改革，这将为本土创新生态注入充沛活水。尤其令他振奋的是，过去两年里德国成立的初创公司中，有一半以上都是 AI 企业，这展现出惊人的创新爆发力。 主持人调侃称，这些初创公司里有不少创始人当年曾是英伟达的实习生，如今已自立门户挑大梁，没准明年黄仁勋再来柏林时，就得带着钢笔追着找这些年轻创始人签名留念了。对话在全场对两位嘉宾的热烈掌声中圆满落幕。 > “看到我在德国的所有投资全部化为真金，没有什么能比这带给我更多快乐了。” —— 黄仁勋 ## Notable quotes > [10:00](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=600s) 世界上所有的数字技术都无法取代机电物理技术。电子永远无法取代原子。 — 黄仁勋 > [02:19](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=139s) 每家公司都将制造两样东西：他们将制造大脑，以及装载这些大脑并最终交付的产品。 — 黄仁勋 > [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) 无论你的通用智能有多聪明，你永远不可能比那些在机电技术某一特定细分领域极度专业的人更聪明。它无法被取代。因此，AI 只能增强你的能力，而不会取代你的能力。 — 黄仁勋 > [06:03](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=363s) 你绝不应该出口你的文化、出口你的价值观、出口你的知识、出口你的历史，然后再把德国智能进口回来。这毫无意义。 — 黄仁勋 > [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) 选择并不存在，你无法在安全与能力之间二选一，它们本为一体。 — 黄仁勋 > [17:18](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=1038s) 安全不是空想与口号，安全是技术与工程。 — 黄仁勋 > [13:25](https://www.youtube.com/watch?v=dIi3X6VJ3v8&t=805s) 我一直坚持我们必须保持一个拥有制造业的国家，我们决不能放弃这一点。 — 德国总理 ## Mentioned 黄仁勋, 德国总理, 英伟达, ELSA, 奥迪, 梅赛德斯-奔驰, 康林松, Agile Robots, Avelios, Reflex Aerospace, Black Forest Labs, 梅赛德斯-奔驰 S 级轿车 ## Video description 章节 00:00 片头 00:06 交换一天工作 00:58 英伟达与德国的合作 02:19 未来工厂与数字孪生 03:48 让AI走进德国工业 06:03 数据、领域知识与物理AI 08:22 全球制造业与新一轮AI机会 13:25 德国的制造业基础 14:59 投资德国与创业生态 17:18 AI安全与监管 22:27 一年后的期待 24:06 来源与片尾 [Source](https://www.youtube.com/watch?v=dIi3X6VJ3v8)

Source quotes

[10:00] 黄仁勋: 世界上所有的数字技术都无法取代机电物理技术。电子永远无法取代原子。 [2:19] 黄仁勋: 每家公司都将制造两样东西：他们将制造大脑，以及装载这些大脑并最终交付的产品。 [6:03] 黄仁勋: 无论你的通用智能有多聪明，你永远不可能比那些在机电技术某一特定细分领域极度专业的人更聪明。它无法被取代。因此，AI 只能增强你的能力，而不会取代你的能力。 [6:03] 黄仁勋: 你绝不应该出口你的文化、出口你的价值观、出口你的知识、出口你的历史，然后再把德国智能进口回来。这毫无意义。 [17:18] 黄仁勋: 选择并不存在，你无法在安全与能力之间二选一，它们本为一体。 [17:18] 黄仁勋: 安全不是空想与口号，安全是技术与工程。 [13:25] 德国总理: 我一直坚持我们必须保持一个拥有制造业的国家，我们决不能放弃这一点。

Source quotes

[8:22] Jensen Huang: All the digital technology in the world will not replace Electromechanical Physical Technology. Electrons will never replace atoms. [6:03] Jensen Huang: Artificial general intelligence will never replace domain expertise. [17:18] Jensen Huang: Safety is not hopes and dreams. Safety is technology and engineering. [17:18] Jensen Huang: It is not a choice. You don't choose between safety and capability. They are one.

Original transcript

[0:00] It turns out if you could swap jobs Okay. With Jensen for just a day. What would you do first, you think? Oh, that's a great idea, to swap the jobs. Would you like to go there? No. Well, I give you my answer. I rather... would put my leather jacket on I would take a flight to Berlin I would like to see the German Chancellor, and I would tell him I'm ready to invest. Well, that's exactly what I'm doing. I think we could arrange a meeting. Nobody has invested more in Germany. Thank you for that. We are really grateful. Thank you for doing this and this is a great signal for our country. Thank you. Thank you. You know, NVIDIA's story and Germany's story played a very large role. In fact, in the year 2000, NVIDIA was a consumer-oriented technology. We were only building for PC gaming. And it was a company in Germany, ELSA, that partnership led us into workstations, would led us into all of the car companies. And today, we work with the car companies from styling, design, engineering, simulation, manufacturing all the way out to the cars. It was in 2005. that Audi brought us into the car industry inside the car. And it was 2023. Was it 2023, 2020? Yeah, 2020. that Ola and I establish a partnership to build autonomous vehicles. And so Germany and NVIDIA has been very close for a long time. Jonathan, I think I already heard that you wouldn't like to swap jobs. So I won't even get there asking you what you would do as a chancellor. We let you be you for now. And I suspect you can already... picture things that we might not even imagine yet. So take us into an engineering office or a factory of the future. What do you see? What is this going to look like? Oh, that's interesting. In fact, The factory, every single company in the future, we'll have two factories. the factory that is real, and the factory that is digital. and the factory that is digital will be so real It might as be real. If it manifests itself when you create the digital twin of your company, of your car company, that digital twin might as well be real. And so in the future, I believe that most people here in Germany will mostly work inside a digital twin of their company. They will build digital cars. Those cars will have digital brains. the manufacturing The factories will have digital, of course, artificial intelligence robots. building inside a robotic factory, building cars that will be robotic. and everything will be completely software driven, everything will be completely AI driven. So every company will build two things. They will build brains. and the brains that go into the products that they ship. - Fascinating. - Yeah, it's gonna be cool. Chancellor, of course, the ambition is to bring AI into everyday use across German industry, from large manufacturers, to also smaller firms. Could you perhaps share a few concrete examples of how you and the German government are ensuring that this widespread deployment is actually taking place? Sure. First of all, I would like to say that I'm really impressed what Nvidia and German companies are already doing in this country. just handed out a model from the new S-Class of Mercedes to me. promising that this car will be autonomous. driving in Germany by the end of this year. Safest and the most autonomous. So this is happening in Germany. So we are... In the government, we are fully committed to be one of the most advanced places. in the world, not just in the European Union, but in the world, to work with AI to integrate AI into our industry, And one of the biggest, if not the biggest advantage we have in Germany is our so-called Mittelstand. And I'm very happy to see that we are having here tonight... many smaller and medium sized companies from Germany. Please. Open your doors... for AI to make use of your data. [5:00] This is so decisive for the German future. This is a real secret and a real advantage for Germany to use, to make use of all the data which are already available in our companies. And if we could combine this AI with companies like yours and our companies in Germany, the big ones, but also the so-called Mittelstand, I'm really optimistic that we will be one of the leading, if not the leading, country in the world in terms of spread AI for all the companies of our manufacturers, machinery, automotive industry, suppliers of automotive industry. I think this is a great opportunity, and I'm decided to meet that. Yeah, you're absolutely right. So first of all, the data of Germany is part of your natural resources. It belongs to the nation. The data of your digital experience or otherwise known as data the data of your country is your culture your values your knowledge your experience It's your natural resource. It belongs to the nation. And you're absolutely right that you should import the technology to harvest that data and turn it into your national intelligence. But you should not export your culture, export your values, export your knowledge, export your history, and then import German intelligence back. Makes no sense. So that's number one. And so you're going to, of course, use the best AI technology available in the world and AI services in the world. But you have every capability to build your own AI. So that's number one. The second is the next wave of AI is physical AI. This is where AI interacts with the physical world. Robotics, manufacturing, cars, heavy industry, that will never be replaced by digital technology The work that you know that knowledge belongs to you The expertise of your domain, that belongs to you. Artificial general intelligence will never replace domain expertise. You could be as smart of a general intelligence as you want, you will never be as smart as somebody who is very very specialized in a particular field of electromechanical technology it cannot be replaced and so AI can only augment your capability it will not replace your capability so take advantage of the technology as a new instrument As for some of these advantages, Jensen, we have on the one hand decades of industrial data. We have very skilled labor, if you like. We have the engineering expertise. So this certainly, as you seem to point out, speaks for a lot of advantages for Germany. You have a global perspective However there are other countries that also have strengths that also like to take advantage of their strengths You talk to so many governments across the globe to so many companies across the world If you take this global perspective, how is Germany braced for this sort of next race in AI? As you know, the last major industrial revolution was electrification. before the one I'll describe. Electrification resulted in really three geographical leaders. Germany, United States, Japan. For the last 50 years, because of the current industrial revolution, which is computing and digital technology, The United States has largely exported its heavy manufacturing, its building, its craft, really to China. So now the world has three again. But it's Germany, Japan and China. The advantage that China has is that they came into the [10:00] Their industrial revolution, if you will, their economic transformation happened during precisely the time of the computer industry. They are quite computerized. China is probably the only nation today that is deep in manufacturing deep in computer science. and very comfortable in both. Well, the good news is there's a complete reset. The last 50 years was about software programming. United States, is obviously world class. But there's a new era now. The computer programs itself. You just have to show the computer what you want. Computer programming, All of the advantages of the last generation has now been accrued into artificial intelligence. There's a complete reset. All the digital technology in the world will not replace Electromechanical Physical Technology. Electrons will never replace atoms. Mm-hmm. And that's an advantage that you have. The work you know, the domain expertise you have, the craft that is here, It will remain here. But you have to take advantage of this new technology. This industrial revolution is so mighty, is so powerful, you don't have another chance. Because you always... Whether we have another chance or not, we should always assume we don't. This Transition is so important. You should combine your deep skills, your deep expertise, in the physical products in the electromechanical products. Mm. and combine it with artificial intelligence. I would not be surprised if currently... Most of the Mirashtats There are challenges. a reduction in labor because people are getting old. The craft of that is captured in the knowledge, the skills of the people. are still starting to get older and older every generation. Take advantage of artificial intelligence. Let them codify. Let them program, let them teach. this artificial intelligence and keep it here in Germany. And let it re-engineer the Chancellor's words Chancellor's words. Re-engineer Germany. This is your chance to do it. Well, I think it's become obvious prerequisites to scale industrial AI in Germany are here and they are very promising and the opportunities related to it are impressive it will be key of course chance that to bridge the gap between Innovation and implementation in factories and in value chains Is a cooperation with NVIDIA a part of the plan to do so? Yeah, sure. And this is already happening. And I would like to say that I'm extremely grateful about what we have just heard from this business leader of the world in terms of AI. So you are giving me and hopefully us a great optimism about our business model in Germany. And you might know that I'm always insisting that we have to remain. a country with manufacturing. And we should never give up that. And now we are seeing a great opportunity in combining our experiences with manufacturing and AI. So if we are successful in bringing these two aspects and parts of the business of the future. Together, I'm really very optimistic that Germany can be one of the most successful economies in the world. Because we are already, we are still, the third biggest economy in the world. So thank you for telling us what the opportunity is and comparing us with others. And I believe and I share your view that the Americans made a big mistake to export their manufacturing. And now they are faced with a very small, successful, but very small based industry and this is our advantage to have this broader access and broader approach to the future of our industry. And look, I'm a large American company. [15:00] And we have incredible partnerships here in Germany for good reason. It's good for us. It's good for us. Our partnership with Audi led to our partnership with Mercedes. Our partnership with Mercedes led to our business. investment in robotics, which led to our partnership with agile robots. We just made an investment in them. We made investment. Avelios made an investment in Murillo. I made an investment in... reflex aerospace. Last year we made investment in Black Forest Labs. Here in Germany, You have deep science and technology and deep science and math. which are the fundamentals for manufacturing, for the deep sciences, for artificial intelligence. And so there's so much-- There's so much technical expertise. And there's so much deep science and deep math. The one thing that you said also, is that you believed in a policy of advancing and encouraging AI startups. or startups in general. I think that's fantastic. One of the things that I really love about the United States, of course, is that we constantly reinvigorate, we constantly reinvent ourselves, and the way we do that is through all the entrepreneurs. And so one of the things that I really love to see We've invested tremendously in Germany. In the last year, we've invested almost a billion dollars This week we're investing in four new companies. And so In a lot of ways, we are one of the largest VCs in Germany, because I believe in Germany. And so for all of the other venture capitalists, you better hurry, because otherwise I'm going to take all the good deals and I'm going to be rich. And so... And so the important thing is The thing I want to recognize is that the science and technology capability of Germany is just second to none. We have a thousand employees here. We have a fantastic office here in Berlin. Some of the fundamental inventions of our companies came from our office here in Berlin and Germany. And so, This is, you want to invest here. Incredible people. May I ask you a question? Could you give us an advice? on regulation. We are having a very strong debate in Brussels on the European level, compared to the US. what is the best way to regulate this? industry or this new technology. I think we agree on No regulation is not the right answer. too tight regulation. is not the right answer either. But what is from your view the right answer in terms of keeping the potential damage in mind that this could happen and already happened. So what is your advice on regulation? What can we do better than others? Safety is paramount. Yeah. There is no future for this industry without safe products. Just as there's no future in autonomous vehicles if the Vehicles are not safe. And so... It is not a choice. You don't choose between safety and capability. They are one. They have to go together. And so, Three pillars. The first is... You can't regulate if there's no technology. So the technology industry must... Yeah. The technology industry... must invent the technology for safety. Just as Ola and I had to invent the technology And we have the safest autonomous vehicle in the car, in the world. We had to invent the technology that allows you to achieve the functionality you want to see, but also the guarantee of functional safety and active safety we want to deliver. And so, That requires invention of technology. That's the first pillar. But that's something that Germany is incredibly good at. Your science, your math, your technology is incredibly good. You have to invent the technology of safety. Safety is not hopes and dreams. Safety is technology and engineering. Second, The companies need to have best practices. Sometimes the young companies, they're still trying to grow up and They make maybe more mistakes than they should. We need to make sure that companies make deliver on the best practices, best methods, And even in safety, as many of the people in the audience, there are probably many in the automotive industry, You know that safety is deep methodology, deep best practices, deep engineering. And that methodology has to be adhere to. And then lastly, there's industrial and government regulations. If it turns out, that [20:00] Remember, AI is two things. Just as the Internet, regulating the Internet is hard, but regulating Internet applications is not so hard. We should be regulating the applications right away. And so if there are AI technologies that you see that's being applied to a new type of product that is not covered by the existing regulatory bodies, they have to evaluate as soon as possible and make sure that their regulation includes AI. And I think I've also heard you say that you want to find that balance between innovation and regulation. There's nothing to regulate if the industry doesn't thrive. You want the German AI industry to thrive as well as do it safely, responsibly. So I think the balance is really important to do. And so I think that the answer is we know there technology we can invent and deploy in order to have AI safety We know that there are best practices that we can adhere to as companies and industries And then in addition to that if necessary we make sure we regulate Just briefly, Chancellor, what would be your answer to the same question, maybe very briefly? No, thank you. This was extremely helpful because we are talking about that on the European level. We are talking that on the national level. And my approach is that we at least need the European level and better we have a global level of regulation and of safety for this fantastic new technology. So this is my personal view but we are still far from that because we are having a different approach between the Americans and the Europeans but we are working on that and most in this I do know that I was successful to get the AI regulation out of the normal regulation of AI in the European Union and to have a special regulation for the machinery industry in Germany and not being part of the general AI limitation and regulation So this was a quick achievement which we had Give Germany a chance Yeah. Give this industry a chance. And just finally, very, very briefly, Chancellor, if Jensen came back in a year's time, which we very much hope, what would you hope for him to see? I would like to see him regularly. let's say at least once a year and discussing progresses we made and seeing a guy sitting in my office and being extremely happy about the German investment place. Nothing would give me more joy than to see all of my investments in Germany turn into gold. Okay, so I have every belief And I very very happy to hear your personal support of the startup ecosystem I very happy to hear that there are regulations involved to make it possible for the pension funds and the risk capital to come into Germany And then I'm also super excited to hear about all of the AI startups and the startups that have happened in the last couple of years. Half of the startups in Germany being founded in the last two years is AI. Yeah, that's terrific. That's good to see. who knows maybe you are going to ask them for some autographs next year you might want to bring a pen when you come once they have their breakthroughs some of them I understand were your interns in the past and now they run their own companies it's been a pleasure to have you here thank you to both of you ladies and gentlemen please give an applause to Chancellor Merz and Jensen Hang thank you thank you Thank you for having me.