Huang & CFO Kress hard to hoop on yearly.
Copilot:
Yes, Jensen Huang and CFO Colette Kress both affirmed Nvidia’s commitment to a yearly product cadence during the Q3 FY2026 earnings call on November 19, 2025. This aligns with Huang’s post-call Bloomberg interview where he reiterated that Nvidia will continue launching new architectures annually.
What Was Said During the Earnings Call
- Colette Kress stated: “By executing our annual product cadence and extending our performance leadership through full stack design, we believe Nvidia will be the superior choice…”
- This confirms that Nvidia is maintaining its yearly rhythm of major architecture releases, which has included:
- Ampere (2020)
- Hopper (2022)
- Blackwell (2024–2025 ramp)
- Vera Rubin (expected 2026)
- Jensen Huang reinforced this indirectly by discussing the “virtuous cycle of AI” and the need to continually advance compute performance to meet demand across training, inference, and agentic AI.
What He Said in the Bloomberg Interview
- Huang explicitly stated: “We’re going to continue our yearly cadence. That’s our rhythm now.”
- He emphasized that Nvidia’s full-stack innovation—from silicon to software—is designed to support this pace, and that customers are planning infrastructure around it.
Strategic Implications
- Yearly cadence = predictable upgrade cycle for hyperscalers, sovereign AI clusters, and enterprise buyers.
- It also pressures competitors (AMD, Intel, startups) to match Nvidia’s pace across architecture, packaging, and software stack.
- The cadence is tightly coupled to TSMC’s node roadmap, with Blackwell on 4NP and Vera Rubin likely targeting 2nm (N2) in 2026.
Let me know if you want to map out Nvidia’s historical cadence vs. TSMC’s node transitions or benchmark Vera Rubin’s likely specs. |
