Many companies hold data that could be highly valuable to AI systems but is too sensitive to share with external model providers. Banks, pharmaceutical firms, and cybersecurity businesses worry about who can access the data, where it is processed, and what happens to it afterwards. Regulatory requirements under GDPR and industry rules make those concerns especially acute, so the data often stays locked away.
Confidential computing is emerging as a way around this. Instead of relying only on access rules, it uses hardware-backed protections to isolate data while it is being processed, creating a kind of locked room for computation. The infrastructure operator, and sometimes even the model provider, cannot see inside. VAST Data's new DataEnclave technology applies this idea by allowing companies to bring AI models into confidential environments containing sensitive data, rather than moving the data to an external AI service.
The approach could help financial firms analyse customer records, cybersecurity teams work with threat intelligence, and governments use AI on classified information. Startups with specialised datasets in finance, healthcare, and defence might also benefit, since they would no longer have to choose between using advanced AI and protecting their intellectual property. The source notes that governance decisions still remain, but confidential computing removes one major barrier to putting proprietary data to work. The article is based on a single source, so no differing views are reported.