Nvidia chief executive Jensen Huang said he expects at least $1tn in chip revenue across roughly the next two years, driven by new AI tools.
Huang told the gathered throngs at the GTC 2026 developers conference, “ I see through 2027 at least $1tn” in revenue. He added that he was “certain” demand would prove higher.
Nvidia shares briefly rose more than two per cent after Huang’s remarks at the $4.5tn semiconductor giant’s GTC event in Silicon Valley. The mood did not stick, and the stock slipped back, finishing lower than before the presentation.
His big numbers have not thrilled the cocaine nose jobs of Wall Street lately, with investors fretting about returns on vast AI infrastructure spend. Throw in the risk of Middle East conflict to supply chains and shortages of the memory chips Nvidia needs, and the nerves keep jangling.
The $1tn projection sits well above the cocaine nose jobs of Wall Street expectations and signals Nvidia is betting the AI boom has plenty of runway. Being the biggest supplier of advanced AI chips has turned it into the world’s most valuable company, which is a handy problem to have.
Deepwater Asset Management, managing partner Gene Munster, said Nvidia’s shares had hit a “wall of worry” about longer-term growth.
“Bottom line from Jensen’s keynote: Demand is measurably stronger than even the highest expectations, and investors are still having a hard time getting comfortable with that,” he said.
Huang had previously pitched $500bn in AI-driven revenue through the end of 2026, citing “high confidence demand and purchase orders” for Blackwell and Rubin kits. The new $1tn target is far higher than consensus estimates for Nvidia’s total revenue.
Capital IQ figures put analysts’ combined forecasts for Nvidia’s 2027 and 2028 fiscal years, ending January 2028, at about $835bn. That gap is where the hype lives.
Huang said demand for compute is being pushed by tools such as Anthropic’s Claude Code and the rise of “inference”, meaning running models and applications.
During his two-hour keynote, Huang rattled through everything from robotaxi partnerships to a chip meant for orbital data centres. He nodded at an idea Elon Musk has pitched around linking SpaceX with xAI.
He unveiled a new chip in the line-up, the Groq 3 “language processing unit”, designed to speed up AI systems’ responses to user queries. Nvidia is clearly hunting new architectures instead of relying on a single GPU answer to every problem.
“We are in volume production now,” before adding, “We’ll ship Groq 3 in the second half [of 2026], probably in the Q3 timeframe,” Huang said.
The chip will be made by Samsung, a departure from Nvidia’s usual reliance on Taiwan Semiconductor Manufacturing Company for its AI processors. It is the first product from Nvidia’s licensing deal with chip start-up Groq, which included hiring founder Jonathan Ross.
