Data-Security Concerns Reshape How Defence AI Firms Select Their Model Partners

Recent reporting by Yahoo Finance indicates that Palantir Technologies and chipmaker Nvidia have become more cautious about their use of Anthropic’s artificial intelligence models, with the concern centred on whether sensitive corporate information could be exposed through third-party AI systems.

The development highlights a growing tension in the enterprise and government technology markets: while large language models are increasingly embedded in software platforms sold to corporations and public-sector agencies, buyers in those markets apply unusually strict standards around data handling, confidentiality, and vendor dependency. For companies whose customer bases include intelligence and defence organisations, any perceived risk that proprietary or client data could leak through an outside model provider carries outsized commercial consequences.

Why the defence-adjacent market is especially sensitive

Palantir has built its business on software platforms — including its Gotham platform, which integrates with other systems used by intelligence and government customers — where trust in data governance is central to the product proposition. Firms in this sector typically market themselves on their ability to keep classified and commercially sensitive information siloed and secure. Reliance on externally developed AI models therefore introduces a vendor-risk question that goes beyond ordinary procurement diligence: customers may scrutinise not only what a platform does with their data, but whose underlying models it calls.

Nvidia, meanwhile, supplies the accelerated computing hardware that underpins most modern AI deployments, and also ships its own software stack. A reported pull-back in model partnerships underscores that even companies positioned upstream in the AI supply chain are weighing reputational and contractual exposure when associating with specific model developers.

Market context

Shares of Palantir Technologies Inc. (NASDAQ: PLTR) were essentially flat in the most recent session, trading at $170.50 against a previous close of $170.62, a decline of roughly 0.07%. The company carries a market capitalisation of approximately $439.7 billion and is classified in the Technology sector, within the Software – Infrastructure industry. No specific financial impact from the reported change in model usage has been disclosed by the company.

Neither the reported scale of the shift nor its effect, if any, on existing customer contracts is clear from the reporting, and the companies involved have not announced formal changes to their product architectures in connection with it.

What to watch

• Any disclosure from Palantir on its AI model partnerships in upcoming quarterly earnings materials or investor presentations.
• Formal statements from Nvidia, Palantir, or Anthropic clarifying their commercial relationships.
• Broader enterprise and government procurement language around third-party AI model use, which may signal how data-governance expectations are evolving across the sector.

Source: original release (Yahoo Finance).