Jensen Huang personally attended Morgan Stanley’s roadshow: quarterly revenue is nearing $100 billion, Nvidia’s growth rate is still accelerating, and he denied that the Rubin Ultra has been delayed.

Jensen Huang personally attended Morgan Stanley’s roadshow: quarterly revenue is nearing $100 billion, Nvidia’s growth rate is still accelerating, and he denied that the Rubin Ultra has been delayed.

Jensen Huang took the stage himself, delivering a key message at the Morgan Stanley investor roadshow: Growth is not only far from peaking, it is accelerating.

This week, Morgan Stanley hosted a non-deal Nvidia roadshow (NDR) in California. Nvidia CEO Jensen Huang, CFO Colette Kress, and Head of Investor Relations Toshiya Hari attended in person and met with several institutional investors.

The participation of the executive team is itself a signal—the company wants to directly address market concerns about product progress, ASIC competition, and growth sustainability.

Morgan Stanley analyst Joseph Moore subsequently released a report stating, the meeting was "very positive," Nvidia set the tone at "accelerating growth" for this stage, and indicated that even as quarterly revenue approaches $100 billion, growth will continue to accelerate. The bank believes that this growth narrative is attractive to both value and growth investors, reiterates Nvidia as its top pick in the semiconductor sector, and maintains an "Overweight" rating.

Rubin Delay Rumor? Nvidia Denies

Before the roadshow, there were rumors that Rubin Ultra might be delayed until 2028. Jensen Huang directly denied this at the meeting.

Analyst Joseph Moore said, Rubin Ultra will still be shipped next year. Some rack designs of the Rubin system have indeed been adjusted—original Kyber rack plans are being replaced by "better solutions," possibly supporting larger-scale computational domains—but this is an optimization at the system architecture level. 800V power and optical interconnects between racks are progressing as planned, and the product timeline is unchanged in substance.

An ASIC customer’s computing power share rises to 50%, Anthropic?

The most notable detail from the roadshow comes from changes in the AI lab customer base.

Joseph Moore describes in the report: "AI labs currently account for about 20% of Nvidia’s total demand. Notably, a more representative cutting-edge model was previously developed mainly on ASICs, with very little Nvidia involvement, but now it has risen to nearly 50%, while other cutting-edge models still run primarily on Nvidia’s platform."

The bank did not directly name the customer. But judging from features like "cutting-edge models" and "mainly using ASICs," the market generally believes this points to Anthropic—whose backer Amazon is the main driver behind the Trainium chip.

This change directly addresses a core market concern: Will the major push by cloud vendors to develop their own ASICs erode Nvidia’s market share?

Moore’s judgment is: Both things can happen at once. Hyperscale cloud vendors can continue to develop custom chips, while Nvidia maintains high market share. The reason is that customers ultimately compare not the price of a single chip, but the comprehensive cost per token. Moore cites industry research stating, "Nvidia’s solution still offers lower per-token costs in many scenarios," keeping it competitive in both training and inference loads.

Moore also points out that from 2024 to 2026, Nvidia’s overall share of AI computing actually increases.

Diversified Growth Sources: Three Growth Lines Open Simultaneously

Analysts, based on Nvidia’s new business categories, identified three growth lines:

First, AI labs (about 20% of total demand). Besides leading models continuing to use Nvidia’s platform in depth, previously ASIC-leaning customers are also increasing their GPU configurations.

Second, traditional hyperscale cloud vendors (about half of revenue). Microsoft, Meta, Amazon, Google remain the largest customer group, but their expansion is increasingly constrained by electricity, land, and the speed of data center construction. Nvidia’s income from this group is expanding from GPUs to CPUs and networking equipment.

Third, new AI clouds, sovereign AI, industrial, and enterprise customers. Analyst Moore believes these customers, constrained by space, power, and geopolitics, tend to purchase AI infrastructure solutions with higher system integration, and growth here may outpace traditional hyperscale cloud vendors.

Sovereign AI is particularly noteworthy. For data security and industry independence, countries are building local models and computing infrastructure, and these projects are relatively less affected by competition from self-developed ASICs.

CPU and Networking: Nvidia’s Addressable Market Continues to Expand

Nvidia reiterated at the roadshow that its CPU business goal for this fiscal year is about $20 billion.

Moore points out, nearly half may come from standalone CPU racks—that is, Vera CPUs are not just used as management nodes in GPU servers but are entering a broader server market. Vera chips are designed for single-threaded workloads, using larger chip area, fewer cores, and are memory-optimized for AI scenarios.

The network business is also expanding Nvidia’s income boundary. As AI cluster scale grows, data transmission between GPUs becomes a bottleneck. Nvidia is transforming from a single GPU supplier to an AI infrastructure platform covering GPUs, CPUs, networking interconnects, and system architectures.

Starting to Target Value Investors

Analyst Moore notes Nvidia is actively expanding its investor base, making value investors a focus in communication.

The reason is that Nvidia is already heavily held by growth-oriented funds, with some institutions nearing single-stock limits. Moore expects Nvidia may use more than 50% of its cash flow in the future for buybacks and shareholder returns, making it exhibit value stock cash flow traits while maintaining high growth.

Moore expects Nvidia’s revenue to grow 82% in fiscal year 2026 and 52.4% in 2027, with a target price of $288 corresponding to roughly 42% upside at current prices.

Growth Expectations Are Strong, But Valuation and Supply Remain Key Variables

Morgan Stanley maintains Nvidia’s “Overweight” rating, with a target price of $288. Nvidia’s closing price on July 9 was $202.78, corresponding to a current market cap of about $4.97 trillion.

The bank expects Nvidia’s revenue to grow 82% in fiscal 2026 and 52.4% in fiscal 2027. The core logic is that generative AI drives continued growth in cloud computing capital expenditures, Blackwell will remain a key solution for generative AI workloads, and subsequent Rubin products are expected to maintain the company’s performance leadership.

But Moore also points out, the risks Nvidia faces have not disappeared. If supply catches up to demand faster than expected, data center business growth could noticeably slow. Other risks include:

AI development costs dropping significantly;Competitors launching more competitive products;And customers accelerating deployment of self-developed custom hardware.

Based on the information from the current roadshow, Nvidia’s main challenge is not whether AI demand exists, but how to translate demand into deliverable system revenue under multiple constraints such as memory, networking, power, and data center space.

Risk Warning and DisclaimerThe market carries risks, and investment must be carried out prudently. This article does not constitute personal investment advice, nor does it take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions herein are suitable for their own circumstances. Investing based on this article is at your own risk.