Nvidia addresses investor concerns in California meetings
Nvidia used a closed-door non-deal roadshow in California this week to answer investor questions on product timing, ASIC competition, and whether its growth can hold up. Jensen Huang, Chief Financial Officer Colette Kress, and Investor Relations head Toshiya Har attended the sessions, which were organized by Morgan Stanley, according to PANews columnist qinbafrank.
The company’s message was direct: growth remains strong, the roadmap is intact, competition from ASICs is manageable, and the customer base is broadening.
Revenue outlook remains the main focus
Nvidia said quarterly revenue has already reached the $100 billion level and is still accelerating. In the roadshow discussion, that was presented as a sign that demand is running ahead of earlier expectations.
The company also said the Vera Rubin product cycle should be a major driver over the next 12 months, supporting additional orders and revenue. PANews said Morgan Stanley and other institutions project roughly $393 billion in revenue for fiscal 2027 and about $598.8 billion for fiscal 2028.
Nvidia also set a target of $20 billion for its CPU business in the current fiscal year.
Rubin Ultra still set for 2027
On the product roadmap, Nvidia used the meetings to reject several market rumors. The company said Rubin Ultra is still scheduled to ship next year, in 2027, and has not been pushed back to 2028.
Nvidia also said the current Kyber rack design is a transitional solution. A later version is expected to improve support for larger-scale vertical expansion within a rack. The company framed that as an optimization step rather than a delay or a sign of execution problems.
It added that core technologies, including 800V power delivery and inter-rack optical vertical expansion through optical links, are moving ahead as planned. PANews said Huang personally addressed the issue during the meetings.
Nvidia says GPUs remain central even as ASIC spending rises
Nvidia acknowledged that some cloud providers and leading AI labs are increasing investment in custom ASICs. Even so, the company said its position in key AI workloads remains strong.
One example discussed in the roadshow involved a major customer that had previously relied mainly on ASICs for frontier model development but has now lifted Nvidia GPU usage to nearly 50%. The input did not identify the customer by name.
Nvidia argued that its platform still offers the lowest token cost, while its stack across GPUs, networking, software, and CPUs gives it a strong ecosystem advantage. The company also said AI labs account for about 20% of AI computing demand at present, and that this segment is growing quickly.
Customer mix is shifting beyond hyperscalers
Nvidia said its revenue base is becoming less concentrated. Instead of relying as heavily on a small number of hyperscalers, the company is seeing more demand from AI labs, enterprises, neo-cloud providers, and sovereign AI projects.
According to the roadshow summary, newer growth engines such as enterprise, sovereign AI, and industrial customers could expand faster than traditional hyperscalers, helping reduce concentration risk.
HBM constraints may last for years
On supply, Nvidia said shortages in HBM and other memory products are likely to continue for years. The company’s view is that AI token demand is growing faster than storage and memory capacity can expand, leaving supply as a major industry bottleneck.
Huang said memory tightness is a long-term issue, and Nvidia is responding by diversifying its supply chain and planning ahead.
Free cash flow supports shareholder returns
Nvidia also said strong free cash flow should support capital returns, with more than 50% of future cash flow potentially allocated to shareholder returns such as buybacks.
Morgan Stanley said in its summary that market concerns around product delays, competition, and growth durability have been overstated, and described Nvidia as "firing on all cylinders." The bank also raised its price target to $288 while keeping a bullish stance.