Nvidia CEO: THIS will be ‘REALLY BIG’

Here's a comprehensive summary of the provided YouTube video transcript:

Key Concepts:

• NVIDIA's Q3 Earnings Performance
• Data Center Chip Sales Dominance
• NVIDIA's Architecture Versatility
• Blackwell and Rubin Chip Rollouts
• Supply Chain Management and Manufacturing
• Intel Partnership
• China Market Restrictions
• Customer Concentration and "Round Tripping" Concerns
• AI Safety and Regulation

NVIDIA's Third Quarter Earnings and Market Position

NVIDIA's third-quarter earnings significantly exceeded expectations, demonstrating the company's robust performance despite market volatility. While the broader market experienced a thousand-point swing, NVIDIA's stock, though down 2.6% from its highs, reflects strong investor confidence. The company's revenue reached $51.2 billion, with a staggering $57 billion in total revenue. The vast majority of this revenue, $51.2 billion, was generated from data center chip sales, highlighting NVIDIA's premier position as the provider of the "brains" powering the AI world. This success is achieved even without access to China's estimated $50 billion chip market.

NVIDIA's Architecture and Diversified Business

Jensen Huang, CEO of NVIDIA, emphasized the diversified nature of NVIDIA's architecture, stating that AI is not a singular concept but a system of models. NVIDIA's architecture supports a wide range of applications beyond chatbots, including physics, biology, chemistry, robotics, and self-driving cars. Furthermore, it accommodates numerous language model architectures. A recent partnership with Anthropic ensures NVIDIA's architecture is the sole platform capable of running every premier AI model, every open-source model, and every modality of model. This versatility allows NVIDIA to cater to a broad customer base and facilitate their cloud capabilities.

Next-Generation Chip Rollouts: Blackwell and Rubin

NVIDIA is on track to ship its next-generation chips, Blackwell and Rubin. Huang announced at GTC that starting this year and extending through the end of next year, NVIDIA has visibility of $500 billion worth of Blackwell and Rubin orders. Rubin chips are slated for shipment in the third quarter of next year, with silicon already functioning well and engineers actively bringing up the system. Volume production for Rubin is expected in Q3 of next year. In addition to the $500 billion visibility, NVIDIA has secured significant new wins in the last 30 days not included in that figure, such as opportunities with Anthropic, Elon Musk's ventures (XAI), and AI deployments in Saudi Arabia. This includes a partnership with Humane and AWS building 150,000-GPU AI infrastructure in Saudi Arabia. These are considered "net new" orders, indicating continued strong demand.

Supply Chain and Manufacturing Strategies

Learning from past demand-outweighing-supply scenarios, NVIDIA has significantly improved its supply chain planning for the Rubin launch. Huang stated, "We're not going to be caught flat-footed too many times." The company now possesses the largest supply chain globally, working with TSMC for manufacturing in Taiwan and, for the first time, in the United States. NVIDIA also partners with every memory vendor, securing three large sources for complex memory. System makers worldwide are collaborating with NVIDIA, ensuring readiness for the Rubin launch, which is anticipated to be a "really big year."

Partnership with Intel

Regarding a minor investment in Intel, Huang clarified NVIDIA's strategy. The most impactful contribution NVIDIA can make to Intel is by creating opportunities for the Intel architecture. This involves helping Intel build unique products that connect NVIDIA's ecosystem to Intel's, and also becoming a significant customer of Intel's foundry business. NVIDIA views this as a unique capability and an honor to invest in a major American company, aiming to strengthen American semiconductor companies.

China Market Restrictions and Future Outlook

NVIDIA is currently not selling chips to China through legal channels, despite historical sales of hundreds of thousands of chips to companies like ByteDance and Tencent. President Trump's administration continues to curb advanced chip exports to China, and President Xi Jinping has reportedly called for restrictions on NVIDIA's chips. Huang noted that NVIDIA's GPUs are large and require complex networking and software, making them difficult to smuggle. He stated, "I think -- but separately, our situation currently is that it is clear that we really need America to go back into the Chinese market to be able to compete there." He believes re-entry into the Chinese market would benefit both American companies and China's ecosystem. Currently, NVIDIA forecasts China sales to be zero for the next two quarters. However, if restrictions are lifted, China, a $50 billion market this year projected to reach $150-200 billion by the end of the decade, represents a significant revenue opportunity that would fuel further investment.

Customer Concentration and "Round Tripping" Concerns

Concerns have been raised about NVIDIA's increasingly concentrated sales, with filings indicating a shift from six to four major customers. Huang addressed worries about "round tripping" deals, specifically mentioning the perception that NVIDIA might be providing funding to companies like OpenAI in exchange for chip purchases. He clarified, "First of all, we haven't given a penny yet." The projected revenues do not include such investments, which would represent a "tiny percentage" of overall revenues. Companies like OpenAI, Anthropic, and XAI are raising substantial funding through equity and have seen their funding rounds oversubscribed, indicating they do not require NVIDIA's investment for their core operations. NVIDIA's investments are intended to deepen partnerships and support growth, thereby expanding their ecosystem.

AI Safety and Regulation

Huang expressed enthusiasm for an integrated AI world but also highlighted the critical need for AI safety. He stated, "The things that we have to start with is we have to make AI function safely." This includes ensuring AI does not "hallucinate" and is grounded in truth. A significant breakthrough has been AI's ability to research and find truth when uncertain, enabling it to reason and think, thus functioning safely. Regarding regulation, Huang agreed to support it, stating, "Oh, yeah, of course." He believes AI, as an application use case, should be subject to the same regulations as healthcare, consumer, and transportation products. However, he cautioned against overly regulating the fundamental technology itself, emphasizing that the technology needs to advance quickly. Regulation should focus on the "end use of, of AI."