NVIDIA's annual developer conference will be an exuberant affair. But is that enough to calm an anxious market?

As developers, Wall Street analysts, investors and journalists converge on Silicon Valley next week for Nvidia Corp.'s annual developer conference, there's a different vibe in the air.

The boom in artificial-intelligence stocks has shifted vastly early in the new Trump administration. Jitters about tariffs, a weakening economy and potential data-center overbuilding have led to a downdraft in many names, especially Nvidia and other stocks in the chip sector.

Nvidia shares so far this year are down 11%, still reflecting fears about DeepSeek, a Chinese startup that said it created an AI model at a much lower cost than its U.S. rivals. That development in January shaved nearly $600 billion in market cap from the semiconductor giant in a single day, and shares have had a tough time finding meaningful recovery.

For its part, though, Nvidia will use its huge platform at the GTC conference to woo the faithful and to get the technology industry excited again about its growth prospects and its dominance in AI. GTC stands for GPU Technology Conference, a reference to Nvidia's core graphics-processor units.

Much like Apple Inc.'s late co-founder and chief executive Steve Jobs, whose fans would attend MacWorld religiously and hang on his every word, Nvidia CEO Jensen Huang has become a revered figure in Silicon Valley, even wearing signature attire - in his case, a black leather jacket.

The company will surely discuss many near-term products, and Huang's vision for AI and robotics. He'll likely talk about speeds and feeds of the company's current Blackwell platform and the Blackwell Ultra line to follow. But investors will also be looking for assurance that Nvidia's subsequent product line, code-named Rubin, will get off to a smoother start than Blackwell.

"It is clear that the Blackwell introduction has not gone as smoothly as the company might have hoped, and there remains added volatility given how laser-focused investors are on this supply chain," Bernstein Research analyst Stacy Rasgon wrote in a note to clients. "However, it does seem they have worked through things."

During Huang's keynote on Tuesday, Nvidia will likely focus on showing off some of the features of its most advanced Blackwell systems, and talk about how much more advanced Rubin - named after American astronomer Vera Rubin - will be. There will also likely be some discussion of the surrounding parts, such as its advances in high-speed optical networking and water-cooling options. Huang may also introduce some of Nvidia's biggest partners on the stage or drop their names, as he did last year, causing spikes in those stocks.

"It will be a victory lap of Blackwell and who is using it and who is shipping it," said Pat Moorhead, principal analyst at Moor Insights & Strategies. "We will see more technical details on Rubin architecturally, not just the chip but also the rack [that the chips comes in]."

He expects much of the keynote will be somewhat redundant compared with what Huang talked about in his keynote at the Consumer Electronics Show in January, with the potential for some updates. At CES, Huang outlined Nvidia's big push to create a robotics software platform, called Cosmos, which is meant to advance the development of physical AI systems like robots and autonomous vehicles. He also talked about agentic AI and creating custom AI agents that will make work more efficient.

Alongside all the deep technology and science discussion that Huang is known for, investors will be listening for any hints that the AI data-center building boom is slowing down. Some investors fear a tariff-fueled economic slowdown could clamp down on GPU spending.

But these clues are unlikely to come from anything in Huang's keynote or even during press and analyst Q & As that are also scheduled. Many will be keeping their ears open for chatter from Nvidia developers or customers all around the conference, and what they are sensing from customers. For its part, Nvidia said in its most recent earnings call that it cannot supply enough Blackwell systems to meet current demand.

"I think a bigger question for investors is along the lines of, 'What are some of the things that could slow down the AI train?'" said Moorhead.