Significant news is emerging from the AI sector. A major negative development has been reported concerning Anthropic, the primary competitor to OpenAI. The company is facing challenges following a dispute with the Pentagon over artificial intelligence safety. Anthropic has informed a judge that it could face losses amounting to billions of dollars if the U.S. government prohibits the use of its AI tools. Concurrently, OpenAI is making a strategic shift by planning to integrate its Sora AI video generation capability directly into ChatGPT. Analysts suggest this integration could help OpenAI increase its weekly active user count. Separately, NVIDIA has announced a new $2 billion investment in the AI cloud services company Nebius.
The challenges for Anthropic, which is considered OpenAI's most direct and capable rival, are mounting. The company has stated in court that due to its disagreement with the Pentagon on AI safety protocols, it risks losing up to billions in revenue this year if a Trump administration decision labeling it a supply chain risk is not blocked. Anthropic is urging the court for swift action.
According to Bloomberg, during a hearing in San Francisco, lawyers for the startup emphasized the urgency of the case to U.S. District Judge Rita F. Lin. The core of the dispute revolves around Anthropic's demand for assurances that its AI technology will not be used for mass surveillance of Americans or the deployment of autonomous weapons.
Anthropic's lawyer, Michael Mongan, argued that the federal government's actions have prompted over 100 enterprise clients to question their continued partnership with the company. He provided specific examples: a financial services firm paused negotiations on a $50 million contract; a pharmaceutical company requested a contract term reduction by ten months; and a fintech company explicitly cut its contract value from $10 million to $5 million, citing the issues between Anthropic and the government. Mongan stated that Anthropic's CFO estimates potential revenue losses ranging from hundreds of millions to billions of dollars by 2026.
A hearing on Anthropic's request, initially scheduled for April 3rd, has been moved up to March 24th by the judge. Mongan has asked the federal government to commit to refraining from any retaliatory actions, such as issuing an executive order impacting the AI startup, before the next hearing. Anthropic seeks to have the supply chain risk designation overturned and wants relevant government agencies to withdraw associated directives. The company claims it is being penalized for disagreeing with the government and argues that the legal principles involved could affect any federal contractor that falls out of favor.
Anthropic has garnered support from the tech industry. In a joint letter to the judge, dozens of AI scientists and researchers from companies like OpenAI and Google expressed their backing. They stated that current AI systems cannot "safely or reliably handle fully autonomous lethal targeting and should not be used for domestic mass surveillance of Americans."
In a significant strategic pivot, OpenAI plans to integrate its AI video generator, Sora, into ChatGPT soon, according to a report by The Information citing people familiar with the matter. This move is part of a broader strategy to further grow its user base, though it may also increase the operational costs of the chatbot. Sora represents OpenAI's expansion into multimodal AI technology, competing with text-to-video tools from companies like Meta and Google's Alphabet.
While text-based AI models have found applications in homes and workplaces, models specialized in generating video and images represent the next frontier for the technology's transformative potential. OpenAI launched the Sora mobile app in late September 2025, allowing users to create and share AI-generated videos in a TikTok-like format. Although the app gained some traction after launch, its downloads and user spending declined by early 2026. It is reported that OpenAI will continue to operate the standalone Sora app, though its long-term future is unclear. The integration with ChatGPT is expected to help boost OpenAI's weekly active users, which currently stand at approximately 920 million, short of last year's target of 1 billion.
By adding convenient video generation directly to ChatGPT, OpenAI hopes to replicate the viral success of last March when users flocked to generate personal photos in the style of Studio Ghibli, pushing the company's computing resources to their limits. OpenAI anticipates that its inference costs—the expense of running AI models for ChatGPT and other products—could exceed $225 billion from now until 2030. The company expects to need sufficient computing capacity to handle potential usage spikes from hit features.
In a separate development, OpenAI announced on Monday that it is acquiring the AI safety platform Promptfoo, pending customary closing conditions. OpenAI stated that Promptfoo helps enterprises identify and fix vulnerabilities in AI systems during development, with its toolkit trusted by over 25% of Fortune 500 companies. Upon completion of the acquisition, Promptfoo's technology will be integrated directly into the OpenAI Frontier agent collaboration platform.
NVIDIA has also made a major move. On Tuesday, the company announced a collaboration with AI firm Thinking Machines. As part of the agreement, NVIDIA will provide Thinking Machines with next-generation Vera Rubin chips offering over 1 gigawatt of computing power. Simultaneously, NVIDIA will make a "significant investment" in Thinking Machines. In a joint statement, the companies said these processors will be deployed early next year. The agreement also calls for the joint design of training and inference service systems based on NVIDIA's architecture, providing broader access to cutting-edge AI and open-source models for enterprises, research institutions, and the scientific community.
Thinking Machines was founded in 2025 by its CEO, who left the position of Chief Technology Officer at OpenAI in 2024. The CEO stated in the announcement, "NVIDIA's technology is the cornerstone of progress across the field. This collaboration will accelerate our ability to build AI that users can shape and personalize, which in turn will expand human potential." According to the agreement's terms, NVIDIA's "significant investment" will support Thinking Machines' long-term development, though the specific financial amount was not disclosed.
Notably, NVIDIA has recently engaged in a series of partnership deals. On March 2nd, it announced agreements with Coherent and Lumentum to jointly develop optical technologies. In February, NVIDIA indicated it would enter a large-scale, multi-generational deep collaboration with Meta. OpenAI had previously disclosed that NVIDIA would invest $30 billion in its $110 billion funding round. On March 11th, NVIDIA stated it would invest $2 billion in AI cloud services company Nebius, marking another expansion in its portfolio of investments within the AI enterprise sector.
Comments