Morgan Stanley recently organized a non-financing roadshow for Nvidia, with Jensen Huang also stepping in personally to address questions about recent product delays and whether future AI business growth is sustainable—some of this information is definitely worth everyone’s attention.

1️⃣ Growth hasn’t topped out; it’s actually accelerating further.
In the Morgan Stanley notes, it said Nvidia’s quarterly revenue is already approaching the $100 billion scale, but the message from management is that the growth rate will keep rising, without entering a plateau.
2️⃣ Rubin Ultra racks have not been pushed to 2028; they will still ship next year, which Huang personally denied on the roadshow.
3️⃣ Nvidia’s future business growth sources remain highly diversified.
Large model companies contribute 20%, while the leading frontier models still depend heavily on Nvidia. Hyperscalers are still the biggest revenue source, contributing 50% of revenue. As for the remaining revenue, it comes from AI Cloud, sovereign AI, and enterprise AI, which could become a new source of faster-growing demand.
4️⃣ ASIC competition has not clearly weakened Nvidia; instead, some customers are increasing their GPU share.
One frontier-model customer that originally relied mainly on ASICs and had a low Nvidia share now has its Nvidia usage share raised to nearly 50%, directly addressing market concerns that big tech’s in-house ASICs might replace Nvidia GPUs.
5️⃣ CPU and networking revenue are expanding Nvidia’s TAM.
Nvidia’s CPU business target this year is $20 billion, and Vera CPU isn’t only used for GPU servers—it’s moving into a broader AI server market.
6️⃣ Memory supply shortages may continue for many years, benefiting HBM, DRAM, and advanced packaging.
7️⃣ The real bottleneck right now isn’t demand, but the ability of the semiconductor supply chain to deliver on commitments.
Overall, this Nvidia roadshow not only confirmed that its core GPU business continues to grow, but also served as an endorsement for the entire AI semiconductor industry chain—reinforcing that industry demand hasn’t cooled off.
I personally would rather believe Nvidia and Jensen Huang’s view: as the leader of the AI industry chain, Nvidia itself is a benchmark for the broader AI mega-trend, and the credibility of the above points is still quite high—AI demand can hardly be said to have topped out.