Nvidia Rises Over 2% as Jensen Huang Expects Chip Sales to Double by 2027

TradingKey - As of press time on September 17, Eastern Time, Nvidia (NVDA) opened up about 2.1% at $218.58. Jensen Huang's latest forecast regarding 2027 chip sales has drawn market attention.

[Source: TradingView]

During an AI executive conference held in Scotland, Jensen Huang stated that the company expects its chip sales in 2027 to double those of 2026. Public reports have not yet disclosed the baseline sales figures, specific product scope, or calculation methodology.

Nvidia currently primarily sells Blackwell and Blackwell Ultra platforms, while the next-generation Vera Rubin has entered full production. At the GTC conference in March this year, Jensen Huang projected that cumulative revenue from AI chips such as Blackwell and Rubin would reach at least $1 trillion by 2027.

Vera Rubin Runs Qwen3-VL With Up to 3.7 Times the Inference Throughput of GB300

Nvidia announced on September 16 its initial MLPerf Inference v6.1 preview submission results for Vera Rubin NVL72.

When running Qwen3-VL, the system's inference throughput reached up to 3.7 times that of the GB300 NVL72 across offline, server, and interactive scenarios; when running DeepSeek-R1, it reached up to 2.5 times that of the GB300 NVL72.

[Source: Nvidia]

The relevant results fall under the Preview category of the MLPerf Inference v6.1 Closed Division, with the comparison utilizing submission numbers 6.1-0106 and 6.1-0074.

Vera Rubin has entered full production, with production systems scheduled to begin shipping in the fall of 2026. Nvidia's latest financial report shows that CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud, and Nebius already have Vera Rubin racks operational.

Jensen Huang Opposes Industry-Wide Slowdown in AI R&D

At the Dreamforce conference in San Francisco on September 15, Jensen Huang stated that AI safety is paramount. Companies should not bring products to market if they cannot verify their functionality, capabilities, and safety, and should pause the development or release of such products once risks become difficult to control.

He opposed requiring the entire industry to uniformly slow down AI R&D, advocating instead that individual companies take responsibility for product safety and manage risks through engineering and product safety mechanisms.

Nvidia's efforts to expand chip shipments will remain subject to customer capital expenditures, wafer fabrication, HBM supply, advanced packaging capacity, and server delivery schedules. The mass-production pace, cloud platform deployment, and customer orders for Vera Rubin will influence its subsequent revenue contribution.