What was announced at the Red Hat Summit 2025 keynote?

Image mode for RHEL (Announced), RHEL Lightspeed (Announced), OpenShift Virtualization on Oracle Cloud (Announced).

What were the highlights of the Red Hat Summit 2025 keynote?

Hicks framed Red Hat's job as bringing AI to the infrastructure companies already run: new "superpowers" for Linux admins in RHEL (image mode, RHEL Lightspeed, security), OpenShift Virtualization as the landing spot for VMs with more cloud partners, and Red Hat AI with vLLM for inference, with AMD, Ford and Visa on stage. Red Hat promised to change how Linux is managed, which it said has barely changed in 10 to 15 years. Image mode for RHEL manages servers like smartphone updates to control drift and simplify CVE fixes. RHEL Lightspeed is an optional gen AI assistant inside the RHEL command line.

Who gave the Red Hat Summit 2025 keynote?

Matt Hicks, on 2025-05-20 in Boston, Massachusetts.

Red Hat Summit 2025: Day 1 Keynote: Enterprise AI & modern infrastructure

Matt Hicks · May 20, 2025 · 1h 47m · Boston, Massachusetts

Watch on YouTube

Hicks framed Red Hat's job as bringing AI to the infrastructure companies already run: new "superpowers" for Linux admins in RHEL (image mode, RHEL Lightspeed, security), OpenShift Virtualization as the landing spot for VMs with more cloud partners, and Red Hat AI with vLLM for inference, with AMD, Ford and Visa on stage.

Key themes: Open source, AI infrastructure, Cloud · AI-written summary from the published transcript

Where each highlight is said1:46:57

1 2 3 4 5 6

16:09
Red Hat promised to change how Linux is managed, which it said has barely changed in 10 to 15 years
17:55
Image mode for RHEL manages servers like smartphone updates to control drift and simplify CVE fixes
20:48
RHEL Lightspeed is an optional gen AI assistant inside the RHEL command line
33:04
OpenShift Virtualization Engine is an edition of OpenShift focused on running virtual machines
43:00
OpenShift Virtualization gained support on Oracle Cloud and other clouds
53:38
Red Hat AI and vLLM were presented as the way to run inference on any model and accelerator