The Information AM

Save 25% on an annual subscription to read the most important news about technology and business first. For even more, Save $250 on The Information Pro for unlimited access to our proprietary org charts, databases and surveys.

1.	Nvidia, OpenAI and SoftBank Finalize Deal For Massive Ohio Data Center Project By Phoebe Liu Source: The Information

Nvidia, OpenAI and SB Energy confirmed plans to team up on a massive data center in southern Ohio for which Nvidia will provide credit support covering up to $105 billion and OpenAI will commit to a 20-year lease.

The project, at the PORTS-Pike Technology Campus in Pike County, Ohio, will exclusively house Nvidia hardware, the companies said in a statement Monday. As part of the deal, Nvidia will invest $1.5 billion in SB Energy, the SoftBank-backed firm developing the data center and power plant, the statement said, confirming a report Saturday in The Information, which said Nvidia could invest another $1.5 billion in SB Energy at its upcoming IPO. The Information first reported on the three companies’ plans to collaborate on the Ohio project in June.

The PORTS-Pike project will entail 10 gigawatts of new energy but only “8 IT-GW of AI factory capacity,” the companies said, without explaining where the remaining 2 gigawatts would go. OpenAI committed to lease those 8 gigawatts, with Nvidia agreeing to provide unspecified credit support on land, power, and shell buildout for the first phase of 4.25 gigawatts, plus an option to take the remaining 3.75 gigawatts.

The Information also previously reported that Nvidia was in talks to provide credit support on around $100 billion for OpenAI for the first phase of the project. Nvidia said in a filing Monday that its aggregate payment obligation is capped at $105 billion for its initial commitment and subject to specified conditions. That support is in the form of residual-value guaranties, under which, if OpenAI were to default on its lease, Nvidia would cover the shortfall between the guaranteed minimum value of the lease and amounts recovered through a replacement lease or sale, the filing said.

Nvidia may choose to also support the second phase of the buildout, Nvidia CEO Jensen Huang said in a blog post Monday. A person involved in the deal said it’s not technically on the hook to do so and has a couple of years to decide.

The project—exclusively using Nvidia’s full stack of chips, networking equipment and data center architecture—could represent $150 billion to $200 billion in Nvidia revenue per generation of Nvidia hardware, Huang wrote. He added that OpenAI has committed to a total of roughly $600 billion of Nvidia compute through 2030 in PORTS-Pike and other projects. OpenAI is already one of the largest drivers of demand for Nvidia hardware, and $600 billion would likely represent a significant share of Nvidia’s revenue going forward.

Goldman Sachs and JP Morgan advised SB Energy, while Morgan Stanley advised Nvidia.

2.	Meta Social Media Addiction Trial to Begin By Rocket Drew Source: The Information

Opening arguments are set to begin Tuesday in a trial in Oakland, Calif. that will assess whether Meta Platforms designed its social media products in ways that have harmed young people.

In the lawsuit, attorneys generals for California, Kentucky, Colorado and New Jersey are suing Meta. The plaintiffs allege that Meta’s social media sites, including Instagram, are designed to maximize screen time, leading to addiction and other harms to users’ mental health. They are seeking up to $1.4 trillion from Meta.

A spokesperson for Meta said in a statement that the states’ “claims are unsubstantiated and their financial demands are vastly disproportionate.” The spokesperson added that “we stand by our record of creating strong protections for teens, and look forward to making our case in court.”

In March, a jury in Los Angeles found Meta and YouTube liable for $6 million in damages for defective product design that harmed a young user. In a separate case in New Mexico, a judge ordered Meta to pay $942 million for harming teens’ mental health this year.

This brief has been updated to include a direct statement from Meta.

3.	Andreessen Horowitz Focus of DOJ Antitrust Probe Over Databricks, Fivetran Board Seats By Julia Hornstein Source: Bloomberg

Andreessen Horowitz is the target of a Justice Department antitrust probe over its board seats at competing AI companies, Bloomberg reported.

The investigation, which has been going on for almost a year, started around the same time as the agency’s review of Fivetran’s acquisition of dbt Labs, a merger first reported by The Information. Partner Martin Casado served on the board of dbt Labs and Fivetran, while co-founder Ben Horowitz serves on the board of Databricks, which competes with Fivetran on selling services that help companies parse, evaluate, store and move data.

The probe is occurring as Andreessen Horowitz has continued to deepen its ties to the second Trump administration. Co-founders Marc Andreessen and Horowitz each donated millions of dollars to groups supporting Donald Trump’s re-election, though Horowitz also donated $2.5 million to a super PAC supporting Kamala Harris later that year. In June, Andreessen was appointed to a Defense Department policy board. Andreessen was also appointed to co-lead a Federal Reserve task force on productivity and jobs in July.

Fivetran declined to comment. Andreessen Horowitz and Databricks didn’t immediately respond to requests for comment.

4.	Alibaba’s Small, On-Device Model Gains Traction By Juro Osawa Source: The Information

Alibaba Group said its new small-size AI model that customers can run on their own personal computers has been downloaded more than 1 million times in just a few days since its release, making it one of the company’s fastest-growing models.

Alibaba’s new Qwen3.8-27B, a 27 billion-parameter open-source model that understands text, image and video prompts, can be downloaded for free, and users can host it on their own machines instead of paying for cloud-based access to the model through application programming interfaces. By installing special software, users can run the model even on their laptops, Alibaba said.

There is strong demand for such models from individuals and startups seeking low-cost options or enterprise customers who prefer to run models locally for privacy concerns.

Early reviews suggest that the performance of Qwen3.8-27B, despite being a locally run model, is comparable to some proprietary models that are available through cloud services. On the Artificial Analysis Intelligence Index, the Alibaba model has the same score as OpenAI’s GPT-5.6 Luna, the cheapest and fastest version of the flagship GPT-5.6. “This is the first time a local model has scored frontier level capability,” U.S. coding agent startup Cline wrote in an X post about Qwen3.8-27B.

While trying to capture the market for small models that can run on consumer hardware, Alibaba is also competing with other Chinese AI firms in a fierce battle to develop much larger, more powerful open-source models that challenge the most advanced U.S. models. Alibaba recently released its biggest-ever model, Qwen3.8-Max, which has 2.4 trillion parameters, in direct competition with rival Moonshot AI’s popular Kimi K3 open-source model.

5.	Google Outbids Mercor for Spirit Airlines’ Corporate Data By Alix Coutures Source: The Information

Google won a bankruptcy auction for its internal business data and software code of the collapsed low-cost carrier Spirit Airlines, bidding $10 million and beating a $7.5 million offer from Mercor, a company that pays contractors to train AI models for labs, according to a filing Friday in the US Bankruptcy Court for the Southern District of New York.

The assets could help improve its products and AI models, Google said in a statement, adding that “any data we receive will be rigorously scrubbed of any personally identifiable information by a third party.”

“Companies are sitting on decades of records that show how real work gets done,” said Mercor. “Spirit was that same process applied to a bankruptcy estate.“

The sale requires approval at a hearing scheduled for Wednesday. Google is buying Spirit’s internal corporate data but not its customer or credit card information, a person close to the company said. The source said the court would appoint an independent expert on Wednesday’s hearing, to ensure the data is stripped from personal information by a third party.

In recent months, AI companies including Google and data suppliers such as Mercor have been racing to collect corporate data from failed or sold companies. These Slack messages, GitHub and Asana content, Google Drive documents, and years of transcripts of staff meetings to train and run their AI models. Driving this demand is increased use of AI agents that carry out tasks such as answering customers’ technical questions, managing clinical drug trials and verifying invoices on behalf of users.

Spirit Airlines shut down in May, laying off roughly 17,000 people. The employee emails and chat history Google are being sold as an asset.

The story was updated to add a statement from Mercor.

Popular articles