The Morning Download: Nvidia’s GTC Event Marks Pivotal Year for AI

Good morning. Nvidia’s annual GTC event in San Jose caps a transformative 12 months in the development of AI.

The company said it expects the gathering to draw more than 30,000 in-person attendees from over 190 countries. Participants will include developers, researchers, business leaders and AI-native companies, Nvidia said. I expect the themes to evolve from last year, when Isabelle Bousquette and I covered the event. We are excited to be back this year. Follow the WSJ Leadership Institute’s Morning Download for insights and highlights of WSJ coverage.

Nvidia co-founder and CEO Jensen Huang is scheduled to deliver a keynote address this morning reflecting among other things the evolution of AI agents. They are more sophisticated and pervasive than last year, with additional innovation and higher levels of adoption within sight.

Here's perspective from Isabelle: “Last year, AI had a lot still to prove. Agents and reasoning models were nascent. Companies didn’t fully understand what they were or how to use them. At the last GTC, a dozen people we interviewed gave us a dozen different definitions of what they thought an agent was – and none were using agents at scale in production.

“Today their maturity curve is much higher, and the explosive growth, and improvement, of AI coding tools has arguably given agentic tech its killer app. Still, the technology continues to speed far ahead of business adoption.”

Last year, Nvidia was also facing pressure to prove that the massive global infrastructure buildout was necessary, on the heels of the release of DeepSeek, a model released by a Chinese startup that said it had built sophisticated AI models that required fewer of Nvidia’s chips. But today, the certainty that we do need more compute feels stronger than ever. As Isabelle notes, “Mark Zuckerberg is talking about a data center the size of Manhattan. Others are wondering if we’ll run out of space on Earth for data centers and have to outsource to the moon.”

At the same time, this year, Nvidia faces more competition in chips, especially when it comes to putting AI to use solving problems for users. Its GPUs have been utterly dominant for training AI models, but the growth of such applied “inference” has created an opening for rivals including Google and Amazon. In December, Nvidia struck a licensing deal with Groq for its inference technology and hired the company’s founder, Jonathan Ross. More on that front is expected this week at GTC.

The frenzy over AI stocks has died down, with Nvidia shares feeling the pressure. The company’s business performance is nonetheless strong. GTC will be an opportunity for Huang to reassert the company’s leadership as it enters a new and ever-more competitive stage in AI.

GTC is often a bellwether for where AI is headed more broadly.  Tech stocks ended February down 3% for the year to date. The bubble has deflated but it hasn’t burst in the way some have predicted. The direction of public valuations notwithstanding, the fundamentals of AI adoption remain strong, driven by steady technological innovation in compute, memory, networking, AI models, agentic frameworks and tooling as well as security.

We will be covering GTC for insight into where that innovation is headed and what business leaders need to know so they can guide their companies through a critical period of change.

Target's tech chief has a big role to play in the company's turnaround. We briefly interrupt our Nvidia coverage to spotlight today's CIO Journal story on how tech leadership looks to reshape the retail giant. Target in recent years has struggled to boost sales amid shopper complaints of messy stores, high prices and less exciting products.

Isabelle spoke with Prat Vemana, Target's chief information and product officer, who is overseeing a tech team of a few thousand people. Solving the challenges around merchandising and demand forecasting are now their top priorities, he said.

Vemana said he is building out Target Trend Brain, an AI tool that helps designers cut through mountains of data from runway shows, social media, and market reports to identify trending colors, patterns, and cuts. Designers can also use natural language prompts to rapidly prototype initial concepts. Work that once took weeks now takes hours.

Other initiatives include upgrading demand-forecasting algorithms, particularly for seasonal and trendy items and integration with AI shopping agents.

Retail moves fast and Vemana's leadership philosophy is about moving at the same speed: "If there's a paradigm shift happening, you want to be there in that moment," he said. "That's why you see us playing in multiple dimensions."

What else to expect from Nvidia's GTC.  As the AI industry shifts from training large models to monetizing them, demand for efficient, cost-effective inference chips is surging, the Journal's Robbie Whelan writes.

“Nvidia is in a weird moment,” Paul Kedrosky, a venture investor and research fellow at the Massachusetts Institute of Technology’s Initiative on the Digital Economy, told the Journal. “For a long time, Jensen was saying, ‘We don’t need to have dedicated, stand-alone inference chips, you can just throw a Blackwell at it.’ But that ship has sailed, and there’s a host of new competitors.”

Rivals are gaining ground. The Journal reported earlier that AWS is planning to deploy Cerebras's chips in its data centers, combining them with its own Trainium processors for AI inference workloads. Cerebras already has a $10 billion deal with OpenAI. 

But Nvidia is not exactly sitting still. This week at GTC, Nvidia plans to roll out a server that combines a modified version of its new Rubin GPU with a Groq processor specifically tailored to inference computing. Nvidia is also planning to unveil new computing solutions that involve multiple CPUs that are unattached to GPUs. And Intel is teasing a major partnership announcement with Nvidia as part of the event. 

More Meta cuts. The company is planning layoffs that could hit 20% or more of staffers in a move driven by surging AI infrastructure costs and recognition of efficiency gains from AI assistants. Reuters reports that no date is set, but senior leaders are already planning cuts.

Musk announces second xAI reorg in a month. xAI's founding team is rapidly departing as Elon Musk rebuilds the company "from the foundations up," a month after merging it with SpaceX, the Journal reports. Musk acknowledged xAI is "behind in coding" versus OpenAI and Anthropic, and is hiring aggressively.

Musk's use of the drug ketamine will be barred from OpenAI's April 28 fraud trial, where a jury will decide if the AI company's founders deceived Musk when he donated $38 million by abandoning its nonprofit mission, Bloomberg reports.

OpenAI’s X-rated talk proposal freaks out its own advisers. Since OpenAI Chief Executive Sam Altman last year floated the idea of enabling erotic conversation in ChatGPT chatbot, opposition within the company has been fierce, WSJ reports. One member of its own well-being advisory council called it a potential "sexy suicide coach." Staff have also flagged risks. A core technical problem: its age-verification system was at one point misclassifying minors as adults about 12% of the time. The launch has been delayed indefinitely.

AI boom explodes San Francisco housing market. Rents citywide were up 14% year over year in February, the fastest growth in the country, according to Apartment List. Condo prices grew 12% year over year as of February, according to the real-estate brokerage Compass. And single-family home prices are up 23% with the median price at $1.96 million. 

Trump Administration set to receive $10 billion fee for brokering TikTok deal. Historians have called the fee nearly unprecedented for a government helping arrange a transaction, WSJ reports. Investors in the deal include Oracle, private-equity firm Silver Lake and Abu Dhabi investor MGX. Trump took a nearly 10% stake in semiconductor company Intel and has agreed to take a chunk of chip sales to China from Nvidia in exchange for granting export licenses.

We want to know what you feel about AI. No, not what you think about AI, but what you really feel about it. Create an emoji that expresses your AI sentiment and send it to us. We’ll publish the best ones in future newsletters.