While artificial intelligence was a major focus of Red Hat’s Linux announcements this morning, the just-released RHEL 10 operating system now also includes support for quantum-proof encryption and better containerization, among other improvements.
The shift to AI is the “next evolution of our platform,” Joe Fernandes, Red Hat’s vice president and general manager of the AI business unit, told reporters in advance of the Red Hat Summit in Boston. “Ultimately, AI models will run, together with applications, on Linux, managed by Kubernetes.”
This will require new AI capabilities, like support for inference servers, language models, tuning, and agentic AI tools, he says.
Today, most generative AI deployments happen within the AI infrastructure offered by the big providers – Azure, AWS, and GCP. But as AI becomes more operationalized, and integrated with more enterprise systems, Red Hat is betting that its dominance in enterprise Linux will translate into AI.
Running AI workloads within the enterprise’s own infrastructure, whether on-prem or in private cloud, offers flexibility, security, and cost advantages.
It’s hard to say who’s in the lead when it comes to enterprise-grade AI because things are changing so quickly, says IDC analyst Jim Mercer. “This is a very dynamic space.” But Red Hat is definitely taking a lead when it comes to leveraging open source to help organizations build and deploy modern AI solutions, he says.
AI in RHEL 10
In particular, Red Hat Enterprise Linux 10 now supports Lightspeed, which integrates generative AI directly into the command line, signaling the beginning of a shift to AI-assisted operations.
“You just type in a plain English query saying, ‘Hey, why is this not working?’,” Raj Das, Red Hat’s senior director of product management, told reporters. “And what the RAG application does effectively is it goes out, searches our knowledge base, searches our manuals, searches our documentation, searches our support cases, and comes back with effectively what you need to do.”
RAG – retrieval-augmented generation – is when relevant data and context is added to a user’s question before it’s sent out to a large language model. It is one of the most popular ways in use today for improving the relevance and accuracy of AI responses.
“It makes the administrator’s job much easier and also lowers the bar of the skills set level that you need,” Das says.
Scott McCarty, Red Hat’s global senior principal product manager for RHEL server, added that Lightspeed leverages Red Hat’s 25 years of expertise and documentation. “Historically, the challenge has been sifting through that documentation,” McCarty told reporters. “People with a PhD in RHEL are fine. They’re great. They love it. But this really expands it and makes it easier for less-trained people to access that knowledge quicker and easier.”
Speaking of RAG, RHEL is now integrated with Postgres vector databases, which are commonly used for RAG applications.
RHEL 10 also supports MCP, or model context protocol, a standard developed by OpenAI competitor Anthropic. Anthropic open-sourced MCP late last year and it has since been adopted by OpenAI, Google, and many other companies. It creates a common way for large language models to interact with tools and data sources.
RHEL 10 security
Red Hat’s latest release also includes some security improvements.
In the area of AI security, RHEL 10 now allows companies to use confidential computing to run AI modules in isolated secure enclaves, Das told reporters. Secure enclaves ensure that attackers can’t listen in on applications and the unencrypted data they use while processing it, during runtime.
Another data security problem is that, even when the data is encrypted, attackers could be vacuuming it up in order to decrypt it later, once quantum computers get powerful enough to break current encryption. RHEL now supports quantum-proof encryption, Das says, including lattice-based encryption algorithms for key encapsulation and digital signatures.
According to Red Hat, RHEL 10 is the first enterprise Linux distribution to integrate Federal Information Processing Standards (FIPS) compliance for post-quantum cryptography.
Red Hat also announced extended update support and lifecycle support for CVEs to include those with a score of 7.0 or higher.
Linux in containers
RHEL has long supported containers, but HREL 10 has a new image code that combines the operating system with the drivers and applications that run on it. This makes it easier and faster for customers to deploy their applications, and to roll them back or update them as needed, Das told reporters. And the containers are immutable, he adds. “So we reduce the attack surface.”
In addition to the Lightspeed AI announcement, image mode is one of the two most impactful announcements that Red Hat made this week, says IDC’s Mercer.