Good morning. Chip giant Nvidia released three new open models on Thursday, which it said are aimed at helping companies build a variety of AI agents.

The models, which are part of its Nemotron family, represent Nvidia’s latest efforts to deliver innovation in open-weight or open-source models—and serve as a boon for the American open-model effort.

Also on Thursday, Meta Platforms said it was releasing a new open-weight AI model and collaborating with Apple to run it locally on the company’s high-end Mac computers. Taken together, the companies’ models help deliver more American-made AI options—though it’s too soon to tell if they will entice developers away from cheaper open models from China.

Kari Briski, Nvidia’s vice president of generative AI software for enterprise, told me earlier this week that the company is committed to regularly releasing newer, better open models.

“This is a message to developers and our partners and enterprises that they can depend on us,” Briski said.

Briski said Nvidia has seen greater interest in its Nemotron models because they’re U.S.-made, and because their underlying data can be inspected. Public-sector organizations and financial institutions, in particular, are wary of using Chinese-made models, she said. And, in general, companies have become more comfortable with customizing, or fine-tuning, open models.

One new model, dubbed Nemotron 3.5 Super VL, is designed to help businesses build so-called always-on AI agents, and is faster for long-running agentic tasks, Nvidia said. A unique aspect of the model, according to Briski, is that it can handle images and video in addition to text. 

The model has been customized by a number of companies, including Atlassian, CrowdStrike, Dassault Systemes, Salesforce and Synopsys, using their own data. Part of the reason these companies are post-training Nvidia’s model is because they can get better results from a customized model than a top-of-the-line frontier model, Briski said.

Is your company using open models, and are they U.S.-made? I would love to know. Send your feedback to me at belle.lin@wsj.com (if you’re reading this in your inbox, you can just hit reply).

• Facing growing public opposition, hundreds of proposed moratoriums, and dozens of scrapped projects, tech giants like Google and Amazon are disclosing data center plans earlier and offering community investments to win over skeptical locals, WSJ reports. Some activists still aren’t buying it. “We’re feeding them information on how to sell their data center to us,” said one resident in New Mexico’s Lea County, questioning Google’s efforts to elicit community feedback on a data center project.

• This was the week rising borrowing costs came for debt-reliant AI upstarts. CoreWeave, the poster child of the AI infrastructure boom and of the heavy borrowing and spending that come with it, saw its stock and bonds plunge as higher-for-longer interest rates and AI profitability fears pushed some of its debt toward distressed levels. The selloff was steep enough to prompt Nvidia CEO Jensen Huang to offer a public vote of confidence, helping spark a rebound that lifted CoreWeave shares 6% on Thursday. A Treasury rally gave the move an extra boost, pulling benchmark rates down from the two-decade highs they hit earlier this month.

• Is AI too big to insure? As AI incidents and lawsuits pile up, the biggest AI companies face potential liabilities in the tens or hundreds of billions of dollars. Specialized AI insurance exists, but it is available only in small amounts with a policy from Munich Re topping out at $50 million. Notes the WSJ: “The biggest AI developers have deep enough pockets to cover many liabilities. But a major incident that causes widespread economic or human harm could swamp a company’s financial buffers.”

• Agents like Meta’s Muse and OpenAI’s Dots are helping millions of people decide on purchases and make bids on online auctions. That could be a win for consumers and the tech companies behind the agents, but it poses challenges for many e-commerce players caught in the middle, WSJ’s Asa Fitch writes. For many e-commerce companies on the receiving end of agent traffic, the shift could be as much a problem as an opportunity, and indeed create new work for them. While those companies have become adept at gaming search-engine traffic and social-media conversations to bring in customers, how to adapt to a flood of agent-driven traffic is less obvious.

The WSJ Tech Council brings together CIOs, CTOs and CISOs advancing innovation and shaping the future. Join this trusted community where tech executives connect with peers to explore emerging trends and gain the perspective they need to stay ahead of disruption.

Request Information

Follow Isabelle Bousquette on LinkedIn, Instagram, X, and TikTok for more behind the scenes on her tech and AI coverage, and lately, her contributions to the WSJ Leadership Institute’s new Executive Resilience series, where she’s profiling America’s top execs about their fitness and wellness habits.

Follow Belle Lin on LinkedIn and X for her latest reporting on enterprise technology and AI.

Steven Rosenbush is chief of the enterprise technology bureau at the WSJ Leadership Institute. He also has a column. You can follow him on LinkedIn.

Tom Loftus is the editor of The Morning Download. He suggests following Isabelle, Belle and Steve on their various social channels. But if you insist, here’s his LinkedIn.