What did Nvidia say during the closed-door roadshow? What rumors were addressed?

What did Nvidia say in the closed-door roadshow? What rumors were addressed? This week, Morgan Stanley organized a non-deal roadshow for Nvidia in California, attended by Jensen Huang, Nvidia's CFO Colette Kress, and IR head Toshiya Har, who held intensive meetings with investors. This closed-door meeting mainly aimed to clarify and respond to market concerns about product progress, ASIC competition, and growth sustainability (FUD). Here’s a summary of the key content and points from the closed-door meeting:

1. Revenue Growth and Overall Outlook (Most Concerned)

1) Quarterly revenue has reached $100 billion, with growth still accelerating. This is seen as a strong signal, indicating demand far exceeds prior expectations.

2) The Vera Rubin product cycle is strong, which will bring significant orders and revenue increases in the next 12 months, serving as the main driver of short-term growth.

3) Institutions like Morgan Stanley provide optimistic forecasts: approximately $393 billion in revenue for fiscal year 2027, and approximately $598.8 billion for fiscal year 2028.

4) CPU business goal: strive to reach $20 billion this fiscal year.

2. Product Roadmap and New Product Progress (Key Clarifications)

1) Rubin Ultra: Clearly stated it will ship next year (2027), completely denying market rumors of a delay to 2028. The roadmap remains unchanged.

2) Kyber Rack: The current design is a transitional solution, which will later be replaced by a better solution to support larger vertical scaling of a single rack (larger computing clusters). This is an optimization adjustment, not a delay or issue; the roadmap maintains flexibility to reduce the risk of capacity scaling issues.

3) Core technology is progressing as planned:

◦ 800V power supply system.

◦ Inter-rack optical vertical expansion.

• Overall next-generation product cycle (including the Rubin series) has no significant changes; Jensen Huang has personally clarified this, showing the company's confidence in the schedule.

3. ASIC Competition and Market Share

1) Even with some cloud manufacturers and cutting-edge AI laboratories heavily investing in custom ASICs (application-specific chips), Nvidia still maintains a strong dominant position.

2) Key example: A major customer, previously reliant on ASICs for cutting-edge model development, has significantly increased its use of Nvidia GPUs to nearly 50% (industry speculation may refer to labs like Anthropic). This is seen as major validation— even the ASIC camp needs more Nvidia GPUs.

3) Reasons: Nvidia has the lowest token usage cost, and its fully stack ecosystem (GPU + network + software + CPU, etc.) creates a strong moat; cutting-edge AI workloads still highly depend on its platform.

4) AI laboratories currently account for about 20% of AI computing power demand, with significant growth in demand.

4. Customer Diversification and New Growth Engines

1) The customer structure has shifted from high dependence on a few leading hyperscale cloud manufacturers to diversification: AI laboratories, enterprise customers, emerging cloud service providers (neocloud), and sovereign AI projects (data security, industry autonomy, and geopolitical factors driving increased importance).

2) New growth areas (enterprise, sovereign AI, and industrial sector customers) may grow faster than traditional hyperscalers, reducing concentration risk.

5. Supply Chain and Supply Constraints

1) Supply shortages of memory (such as HBM) are expected to last for several years. The demand for AI tokens is growing faster than the expansion of storage/memory capacity, leading to industry-wide shortages (affecting data center components, GPUs, and AI operations).

Jensen Huang explicitly stated that the memory shortage is a long-term phenomenon.

2) Overall supply remains a major bottleneck, but Nvidia is responding through supply chain diversification and planning.

6. Other Management Perspectives and Capital Returns

1) Strong free cash flow (FCF) supports shareholder returns: more than 50% of future cash flow may be used for stock repurchases and other shareholder returns.

2) The moat of the comprehensive AI infrastructure platform (hardware + network + software) continues to strengthen.

Morgan Stanley Summary:

Concerns about product delays, competition, and growth sustainability have been exaggerated, and Nvidia is "firing on all cylinders."

Morgan Stanley maintains a strongly optimistic stance and raises the price target to $288 (corresponding to significant upside potential).

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。