Goldman Sachs Communacopia + Technology Conference
文章语言：
简
繁
EN
Share
Minutes
原文
会议摘要
Rapid progress in AI infrastructure spending by 2030 highlighted, with focus on generative computing over retrieval-based, co-design with more transistors post-Moore's Law, Nvidia's full-stack AI platform for diverse models, regional cloud market growth, addressing supply constraints, investments in AI startups, and Nvidia compute as an investable asset, alongside advancements in physical AI for self-driving and robotics.
会议速览
AI Infrastructure Spending Surging Towards 2030 Projections: Technical Advancements and Market DevelopmentsAcknowledging past predictions, the discussion highlights rapid progress in AI infrastructure spending, attributing this surge to significant technical advancements and evolving market dynamics, underscoring the validity of earlier forecasts.
Reflecting on What People Miss and Appreciate in LifeA discussion unfolds on the aspects of life that people often overlook or cherish, with participants sharing personal insights and perspectives on the subject.
From Power Distribution to AI: The Evolution of Computing and Information AccessTraces the advancement from power distribution enabling ubiquitous access to information through the internet, to the current focus on AI, aiming to answer any question and know everything, highlighting two compounding challenges in this transition.
Transitioning from Retrieval to Generative Computing for Universal Knowledge AccessThe dialogue explores the shift from retrieval-based computing, where information is pre-recorded and retrieved, to generative computing, which continuously generates answers based on queries and context. It highlights the end of Moore's law and the need for personalized, context-aware recommendation engines to achieve a world where anyone can ask anything and know everything.
Co-Design & Scalability: Nvidia's Response to Post-Moore's Law AI ChallengesThe dialogue discusses the shift towards co-design and scalability in the era post-Moore's Law, highlighting Nvidia's innovations like NVLink to connect multiple chips, essential for handling larger AI models and overcoming transistor limitations.
Semiconductor Industry's Growth Due to AI Demand and End of Moore's LawThe semiconductor industry is projected to expand significantly due to increasing demand for AI technologies and the end of Moore's Law. As AI models require more computational power, the industry is poised for growth. Coding has proven to be a successful application, and other tasks like recipe creation (referred to as 'cooking') could further drive adoption and ROI.
The Ubiquity of Coding and AI's Role in Cybersecurity and Business ProcessesDiscusses how coding underpins all business processes and AI's potential in enhancing cybersecurity through red and blue teaming, highlighting partnerships for AI model development.
NVIDIA's Unparalleled Growth: From GPU Pioneer to High-Value AI InnovatorThe dialogue underscores NVIDIA's evolution from a GPU company to a leading AI innovator, highlighting its cutting-edge architectures and exponential growth in AI factory capabilities, achieving month-over-month increases and setting new benchmarks in the industry.
Expanding Capabilities and Market Share in Frontier Model TechnologiesA strategic focus on increasing CapEx to grow and capture more market share, highlighting the company's unique position in running both closed and open frontier models, including Gemini, Anthropic, and OpenAI, showcasing rapid expansion and investment in the evolving tech landscape.
Navigating AI Market Growth: Upstream Supply Chain Challenges and Downstream AdvantagesDiscusses the rapid expansion of the AI market, highlighting challenges in the upstream supply chain and Nvidia's strategic advantages in the downstream market, including its role in securing land, power, and shell for Neo Clouds, and the versatility of its general-purpose computing solutions.
NVIDIA's Dominance in AI Market: Full Stack Solutions and Global EcosystemNVIDIA leverages its full stack AI platform and global ecosystem to dominate the AI market, collaborating with major enterprises and securing significant land power resources, aiming for 70% annual growth.
NVIDIA's AI Factory in Australia: A Collaborative Effort for 2 GW Renewable Energy by 2027Announcement of a 2 GW renewable energy initiative by 2027 in Australia highlights NVIDIA's AI Factory collaboration with local data centers, aiming to harness excess energy and advance AI technology.
Transitioning Nvidia Compute to Investable Asset: A Shift in AI and Technology FinanceDiscusses the transformation of Nvidia compute into an investable asset, emphasizing the shift towards high CapEx and OpEx in AI companies, and the untapped opportunity for long-term value in computing systems.
NVIDIA's Role as a Foundational AI Platform and its Economic ImpactDiscusses NVIDIA's significance as a core AI infrastructure provider, comparing it to TSMC, highlighting its capacity and technology as key assets for AI development, and emphasizing its role in facilitating AI company growth and secure resource acquisition.
Investing with Confidence: Leveraging AI and Pipeline for Secure Returns and Industry GrowthSpeakers discuss the non-circular nature of investments, emphasizing informed decision-making, secure returns, and strategic industry support through AI platforms and pipeline development, leading to significant contracted value and demand visibility.
NVIDIA's Vision for Regional AI and the Rise of Physical AI in Autonomous VehiclesDiscusses NVIDIA's strategy for regional AI hubs and the potential of physical AI, focusing on self-driving cars as the first killer app. Highlights advancements in reasoning cars and predicts significant progress in autonomous vehicle technology by 2030.
Revolutionizing Logistics and Manufacturing with Autonomous Systems and Smart RobotsNvidia is spearheading partnerships with major players like Mercedes, Uber, and Amazon to integrate advanced autonomous systems in logistics and warehouse operations. Simultaneously, efforts are directed towards developing smart robots for manipulation systems, aiming to expand robotic capabilities beyond large car manufacturers to mid-size manufacturing companies, thus revolutionizing the industrial supply chain.
Physical AI's Role in 6G and Telecommunications: A Partnership with Nokia and NvidiaPhysical AI, exemplified by 6G technology and the AI Ran platform developed with Nokia and Nvidia, is set to revolutionize telecommunications. This distributed edge data center technology, which uses radio waves to sense the environment, is poised for significant success. Nvidia's contribution to physical AI includes the Dojo supercomputer, crucial for training models like those for self-driving cars, showcasing the company's pivotal role in this evolving field.
要点回答
Q：What does the speaker indicate about the advancements or market developments related to AI?
A：The speaker indicates that rapid progress is being made toward the significant AI infrastructure spending prediction, suggesting that advancements and market developments are validating the prediction.
Q：How does the speaker describe the evolution of computing from finding information to knowing everything?
A：The speaker describes the evolution of computing as moving from finding information (the ability to search and locate data) to knowing everything (the ability to understand and provide answers to any inquiry). This shift is marked by the development of AI and the transition from retrieval-based computing to generative computing.
Q：Why does the speaker believe that traditional recommendation engines are insufficient for the new computing needs?
A：The speaker believes traditional recommendation engines are insufficient because they are based on user preferences and can't provide answers to new, diverse queries that require understanding of context and prompt. The new computing needs involve generating answers that take into account the surrounding environment and individual context.
Q：What challenges does the end of Moore's law pose for computing, and how is Nvidia addressing these challenges?
A：The end of Moore's law poses the challenge of needing more transistors in chips, which leads to requirements for co-design, increased chip production, and better scale-up technology. Nvidia is addressing these challenges through co-optimization and the invention of nvlink to fuse multiple chips and scale up technology, anticipating the need for larger models and a more complex semiconductor industry.
Q：Why will the semiconductor industry's demand and size increase according to the speaker?
A：The semiconductor industry's demand will increase due to the accelerating use of AI, which requires more computational power and larger models. This growth will make the industry significantly larger as the demand for semiconductors accelerates and there is no deflationary technology like Moore's law to control it.
Q：What are the two fundamental ideas that the speaker believes will drive growth in the semiconductor industry?
A：The two fundamental ideas driving growth in the semiconductor industry are the introduction of a new layer of computing with new applications and the end of Moore's Law.
Q：Why is it challenging to measure the ROI of technology in the semiconductor industry?
A：It is challenging to measure the Return on Investment (ROI) in the semiconductor industry because, as of the moment, it is hard to analyze the Profit and Loss (P&L) and identify clear margins.
Q：How is 'coding' crucial to the semiconductor industry and what does the speaker mean by 'coding'?
A：'Coding' is crucial to the semiconductor industry because it underpins all processes, from creating recipes to managing business operations. The speaker uses 'coding' in a broad sense to refer to the process of systematically defining and documenting methods and best practices for doing tasks.
Q：Why is cybersecurity expected to be a significant use case for AI?
A：Cybersecurity is expected to be a significant use case for AI because coding, and by extension, finding bugs and securing systems, are tasks that can be effectively automated. There is a clear right answer to what constitutes a secure system, and AI can learn from and apply these principles.
Q：Who are the partners mentioned in the speaker's plan for advancing AI in cybersecurity?
A：The speaker mentions a partnership with Cisco and Palantir, along with Nvidia, in advancing AI for cybersecurity. These collaborations aim to offer AI factory platforms that can help build proprietary AI models and cybersecurity red team and blue team models.
Q：Why does the speaker consider Nvidia to be a high growth value stock?
A：The speaker considers Nvidia to be a high growth value stock due to its continuous growth, with month-to-month increases of 27% in the Grace Blackwell NV link sales. The company's architecture, not just chips, contributes to the expansion of the AI factory and increases in revenue.
Q：What are the strategies contributing to the company's growth?
A：The strategies contributing to the company's growth include increasing their share of the world's CapEx, growing the number of model companies, running every model which allows the company to capture more, and benefiting from frontier closed-open models which are growing rapidly.
Q：What is the significance of running every model for the company?
A：The significance of running every model is that it enables the company to capture a larger share of the market as more model companies are emerging and new ones are starting up with innovative ideas.
Q：How is the company performing with new models and acquisitions?
A：The company is performing well with new models and acquisitions, successfully running Gemini and anticipating the rollout of Gradio, with Meta and Anthropic being new additions to their portfolio.
Q：Why is the company uniquely positioned with respect to open models?
A：The company is uniquely positioned with respect to open models because it is the only company that benefits from the growth of frontier closed-open models, running everything which provides a distinct advantage in this rapidly expanding market.
Q：How is the demand for AI models among cloud service providers and enterprises?
A：The demand for AI models among cloud service providers and enterprises is very high. Cloud service providers are building AI factories, and enterprises, including names like Jane Street, Hudson River, and major pharmaceutical companies, are adopting AI in their operations.
Q：What challenges exist in the AI market supply chain?
A：Challenges in the AI market supply chain include difficulties in packaging,DRAMs,LPDDR,connectors, voltage regulators, and wafers. The fast growth of the industry has placed significant strain on the supply chain.
Q：How does the company leverage general purpose computing to its advantage?
A：The company leverages general purpose computing to its advantage by providing a versatile and fungible platform that is highly valuable in the face of capital constraints. General purpose computing offers durability, versatility, and investability which is crucial given these constraints.
Q：What new capability does the Nvidia Compute provide?
A：The Nvidia Compute provides a new capability of being an asset-backed computer, which can be used to secure loans. This is a powerful feature exclusive to Nvidia, as no one else can provide the same level of security for their computing stack.
Q：What is the role of Neo Clouds in the company's market strategy?
A：Neo Clouds play a crucial role in the company's market strategy by securing land, power, and shell for data centers. This strategy allows the company to expand globally and be diversified in its approach to market various AI models and address a significant portion of the world's data center CapEx.
Q：What are the company's expectations for revenue growth and market expansion?
A：The company expects high confidence in delivering 70% revenue growth next year and unconstrained demand growth of over 100%. It plans to capitalize on its advantages in the market, including a diverse range of channels to reach various marketplaces.
Q：What is the speaker's company doing to track and secure land power and shell globally?
A：The speaker's company is tracking every single gigawatt of land power and shell around the world, working with partners like Neo Clouds, Oems, and AI native companies to understand and secure a lot of it.
Q：What is the potential growth rate mentioned by the speaker?
A：The speaker feels confident about a 70% year-over-year growth and believes in working to improve it.
Q：What was announced in Australia and why is it significant?
A：In Australia, the speaker's company has announced the standing up of 2 GW per 2027, which is significant because it represents an investment of 80 billion, highlighting their work with data center companies in the region and their role in providing a platform for AI and access to regional land power and shell.
Q：What is the role of Nvidia's AI factory platform?
A：Nvidia's AI factory platform brings together the needs of companies in regions like Australia, which have excess energy but require technology and an ecosystem of off-takers.
Q：What impact has the investment in AI companies had recently?
A：There has been a significant investment with $400 billion of VC funding going into AI natives in the last 6 months, indicating that every technology company will need to be a high CapEx company moving forward.
Q：How is Nvidia's computing system being positioned in the financial industry?
A：Nvidia is working with the financial industry to position its computing system as an investable asset, a transition that is expected to happen fairly quickly and recognizes the long-term value of Nvidia's computing systems.
Q：What is the economic productivity of Nvidia's technology?
A：Nvidia's technology is very productive, with a simple math example showing an annual revenue potential of about $10 billion from a single GW of data centers, compared to the current renting out revenues of about $50 billion.
Q：How does Nvidia compare to TSMC?
A：Nvidia is compared to TSMC in terms of providing a foundational platform for technology and startups, with Nvidia being seen as a platform for AI, similar to how TSMC is viewed by companies in the semiconductor industry.
Q：How does Nvidia support AI labs and Neo Clouds?
A：Nvidia provides guarantees and backstops to AI labs and Neo Clouds, assisting them in securing land, power, and shell for their operations, including through investments and asset-backed financing.
Q：Why does the speaker believe their financial strategy is not circular?
A：The speaker believes the financial strategy is not circular because the return on investment is significant, with a small amount of money put in and a much larger amount coming back, supported by lined up contracts and real demand.
Q：What is the significance of creating a network of Neo Clouds for Nvidia?
A：The network of Neo Clouds is significant as it will serve as distribution channels for Nvidia's architecture and provide a place for AI natives to establish themselves, supporting the regionalization of AI and the growth of Nvidia's global partner network.
Q：What is the first killer app for physical AI and which company has made groundbreaking work in this area?
A：The first killer app for physical AI is self-driving cars, and Nvidia has made groundbreaking work in this area with their development of the world's first reasoning and thinking car, called Alpha Mayo.
Q：Which companies are involved in partnerships related to physical AI, and what are some of the expected applications?
A：Companies involved in physical AI partnerships include Waymo, Tesla, Nvidia, Mercedes, Uber, and Amazon. Some expected applications include AMR and warehouse delivery vehicles inside logistics centers, grocery delivery vehicles, and other types of vehicles that can operate autonomously.
Q：What limitations do current manipulation systems have, and who are the companies involved in developing smarter robots for medium-sized manufacturing companies?
A：Current manipulation systems are pre-reprogrammed, limiting their usefulness to the largest car companies. To expand their functionality to mid and medium-sized manufacturing companies, smarter robots with reasoning systems are needed, which are probably a couple of years away from being developed.
Q：How will physical AI impact telecommunications and what is the partnership between Nvidia and Nokia about?
A：Physical AI will permeate into telecommunications, specifically with the development of 6G technology that uses a different spectrum of electromagnetics to sense the world. Nvidia and Nokia have formed a partnership to build a platform called AI-Ran, which is a distributed edge data center based on Nvidia's CudaMani I.
Q：What is the role of Nvidia in physical AI before deployment and what is the Dojo supercomputer used for?
A：Nvidia plays a role in physical AI before deployment by training the models. They have a collaboration with Tesla on the development of the Dojo supercomputer, which is equipped with numerous GPUs and is used for training self-driving cars.
