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.

OpenAI said it fired three safety employees for mishandling corporate information, which a person familiar with the matter said included sensitive information with an outside organization that does AI evaluations.

“We have parted ways with three individuals for violating our policies on accessing and handling sensitive company information,” said a spokesperson for OpenAI in a statement. “Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work.”

The firings come amid widening fears about the risks posed by advanced AI models. OpenAI and other AI companies have worked closely with evaluation organizations in the past, and in response to the growing concern have recently expressed interest in “embedding” personnel from AI safety organizations into their operations who can audit statements about the risks from the companies’ technology. But these arrangements usually involve strict conditions on what the outside parties can access and share publicly.

One of the terminated researchers focused on AI safety, another on alignment and the third was a research program manager for alignment, according to the OpenAI spokesperson. OpenAI’s investigation found a pattern of misconduct, including mishandling research information in violation of company policies, the spokesperson added.

The Wall Street Journal first reported on the terminations.

Nvidia and SoftBank have each made the final $10 billion investment in each of their $30 billion pledges to OpenAI’s last funding round, according to a person with knowledge of the investments and a statement from SoftBank.

OpenAI in March said it had $122 billion in commitments in a round that valued it at $852 billion, including the money raised. In addition to SoftBank and Nvidia investments, Amazon committed to invest up to $50 billion in two parts. Amazon in July completed the entire investment, after OpenAI negotiated a new agreement with long-term partner Microsoft so that it could run its AI on competing cloud services.

Finalizing the March round could help pave the way for OpenAI to raise its next private financing. It’s targeting raising $30 billion and could seek a valuation of around $1.4 trillion.

SoftBank has entered into a complicated series of financings to fund its OpenAI investments. In a statement to the Tokyo stock exchange early Thursday, SoftBank said the final installment of its funding to OpenAI brings its total investment in the ChatGPT maker to $64.6 billion and its ownership interest to about 13%.

The tech and telecom conglomerate run by Masayoshi Son funded the final installment of its OpenAI investment with an $11.1 billion high-yield bond offering in late September.

A state-backed Chinese financing company has revealed in documents filed with regulators that it funded the purchase of Nvidia chips subject to U.S. export restrictions, Bloomberg reported on Friday.

Semi-Tech Leasing Group, owned by several local government entities, has provided another Chinese company called Glory View Technology with capital to purchase more than 700 servers, Bloomberg said, citing the documents. At least 32 of those servers were equipped with Nvidia’s advanced B300 chips, whose sales to China are prohibited under Washington’s export controls.

Semi-Tech’s filings indicate that government funds are helping facilitate China’s access to advanced Nvidia chips—even though Beijing has also been trying to accelerate the adoption of domestic AI chips to reduce the country’s dependence on U.S. technology.

The Information reported last week that China’s government had signaled it could allow some domestic companies including Alibaba and ByteDance to buy a new Nvidia chip built for high-end professional computers.

A federal judge on Wednesday dismissed lawsuits from education technology company Chegg and Penske Media, the owner of Rolling Stone, Variety and other publications, which had alleged Google had violated antitrust law when its AI Overviews summaries used their content. The court rejected the argument that Google trades search traffic in exchange for content, along with the publishers’ monopoly claims.

In separate lawsuits last year, both firms alleged that AI Overviews siphoned traffic from their sites and cut affiliate revenue. They also alleged that Google forced publishers to accept AI Overviews in exchange for showing up in search.

The judge for the District Court for the District of Columbia said that Google crawls public websites and that publishers allow it because they benefit from being found, which doesn’t amount to an agreement. Google has made the same argument for at least a decade, during the 2005-2015 Authors Guild Book lawsuit and the 2011 GTC antitrust investigation for example, arguing that companies built their business on the back of Google, benefiting from traffic from its search results but doesn’t owe them anything.

The decision follows a similar loss for publishers in March. In that case, a federal court dismissed an antitrust lawsuit against Google and Alphabet by digital news publishers Helena World Chronicle and Emmerich Newspapers. The publishers had argued that Google tied search traffic to the supply of news content for republishing and AI training. The court ruled that they lacked standing because they weren’t in the general search services market, and that Google had no monopoly power in online news.

The court losses leave licensing deals with Google as news publishers’ current best option for getting paid by the search giant for its use of their content in AI. Google also started to pay roughly 100 digital publishers based on how much their content contributes to AI-powered answers in search and other products, The Information reported this week.

Australian neocloud Sharon AI has secured $356 million through its first loan backed by its graphics processing units, making it the latest cloud provider to finance their expansion by borrowing against chips and cash flows from customer contracts. The senior secured facility carries an interest rate of 9.95% before fees.

Sharon AI said it will be used to finance more Nvidia GPUs as the company expands its data center footprint across Australia, New Zealand and other parts of Asia-Pacific. The template for this deal was popularized by CoreWeave in August 2023, when the neocloud secured a $2.3 billion loan backed by Nvidia H100s. More recent companies adopting these types of financing structures for GPUs include Nscale and Iren.

The Sharon AI deal includes participation from Goldman Sachs, among other investors. In total, the company has raised more than $2.6 billion over the past 10 months. The company’s stock dropped more than 8% following the announcement.

U.S. authorities arrested a California man on Thursday for smuggling more than $300 million worth of servers containing restricted Nvidia AI chips to China. Greg Lui, the owner of Earthmade Computer Inc, a privately held company based in San Gabriel Valley, allegedly worked with conspirators from 2023 to 2024 to trade export-controlled items without the required licenses.

Authorities said the servers were routed through countries such as Malaysia and Singapore, where no such license is required, then shipped to the ultimate buyers in China. From January 2024 to October 2024, Earthmade received more than $176 million from two Malaysia-based shipment companies, the Justice Department said. Lui is charged with one count of conspiracy to violate export controls, one count of outbound smuggling, and one count of conspiracy to commit money laundering.

The charges against Lui are the latest move by federal prosecutors to clamp down on illegal sales of advanced US-made AI chips to Chinese buyers. In March, the U.S. charged a co-founder of Super Micro Computer for diverting billions of dollars of Nvidia chips to China via a Southeast Asian company.

Over the past few years, smuggling of advanced Nvidia chips into China has become an organized, multinational enterprise, as Chinese AI firms continue to rely on such chips for model training. In December, The Information reported that DeepSeek was training its model using Nvidia’s Blackwell chips that were smuggled into China.

Infinigence AI, a Chinese provider of AI cloud infrastructure, has filed confidentially for an initial public offering in Hong Kong to raise several hundred million U.S. dollars, Bloomberg reported.

The Shanghai-based company is currently valued at 14.3 billion yuan ($2.1 billion). It has raised 4.3 billion yuan from backers including tech giants Tencent and Baidu, and AI lab Z.ai.

Infinigence would be one of the first Chinese neocloud startups to go public, joining a wave of global neocloud listings this year that includes companies like Sharon AI and Nscale.

Infinigence is planning to list as early as the first half of next year. Details of the offering are still being finalized and may change, Bloomberg said.

Infinigence was founded in 2023 by Wang Yu, a professor at Tsinghua University and the former chairman of the school’s electronic engineering department. The company’s AI infrastructure uses chips from domestic suppliers such as Huawei and Moore Threads Technology.

Data center infrastructure provider Accelevation’s shares were trading down roughly 5% on Thursday morning, as the stock continued to sink following its initial public offering earlier this week.

The company on Sept. 29 raised $540 million, moving ahead with the offering even as some AI-infrastructure companies have pulled back on public debut plans amid broader market volatility. The private equity-backed company sold 30 million shares at $18 each, less than the marketed range of $20 to $24.

Accelevation offers power and infrastructure products and services primarily to data centers. Olympus Partners acquired Accelevation last January from private equity firm LFM Capital. The listing marks the latest test of the broader AI IPO thesis amid a run of AI-related setbacks, including a pickier debt market, power constraints and growing political backlash.