Coming to Chicago for #SC26? Start the week by building.
Join the CIQ Supercomputing Local AI Hackathon, November 13–15, hosted by CIQ, NVIDIA and Drive Capital. Get access to two Lenovo ThinkStation PGX systems with NVIDIA GB10, orchestrated by Fuzzball, and build a real AI workload.
Apply: https://t.co/XBWMplAwbB
@nvidia@drivecapital
Supported by @Lenovo, VAST Data, @CambridgeComp, Moonlite AI.
#HPC #LocalAI
“Never write your own cryptography.”
Samba co-creator Jeremy Allison now leads FIPS 140-3 and Common Criteria work at CIQ, while helping prepare Linux for post-quantum cryptography.
Read his story: https://t.co/PgOzIGWBp7
#OpenSource#Cybersecurity
“Today, such vulnerability exploitation requires a frontier model. Tomorrow that won’t be the case.” Our CTO Peter Nelson on PHP2Shell and what’s next, via
@thenewstack: https://t.co/lVqdbVTavA #PHP2Shell#OpenSource
Starting in one hour.
Join Brady Dibble and Brian Rieb to learn how an update, upgrade, conversion, and migration differ, and what each means for your Linux project.
11 AM PT | 2 PM ET
https://t.co/sW3rdXEZkM
“Finding bugs is getting cheaper. Fixing them needs to keep up,” writes Nathan Blackham. CIQ engineers shipped backported PHP2Shell fixes about 3.5 hours after disclosure. Read the full write-up: https://t.co/flww420b1r
#PHP2Shell#OpenSource#CybersecurityNews
Before your team commits to a timeline, agree on whether the Linux project is an update, upgrade, conversion, or migration.
Tomorrow, Brady Dibble and Brian Rieb explain what each one requires.
11 AM PT | 2 PM ET
https://t.co/Ga6sLqc4OA
Can Enterprise Linux compete on a DGX Spark?
On Oct. 28, CIQ breaks down the community harness, pinned container, and results behind its RLC Pro AI tests.
10 AM PT / 1 PM ET
https://t.co/K6BUv07GU2
Change a queued Fuzzball workflow’s priority without resubmitting it. When a multi-node job fails, see the failed rank, node, and exit status in one error.
Fuzzball 4.3 also adds cluster diagnostics, AMD GPU fault signals, clearer workflow insights, and retention controls.
Read: https://t.co/SND5hnTcAt
Sovereign AI gives organizations verifiable control from the GPU to the model and agent.
At #LATechWeek, CIQ and Phinite demonstrated private model serving, governed agents, and a live model upgrade with rollback and an audit trail.
Thanks to everyone who joined us.
Don’t take a benchmark result on faith. Rerun it.
We’ve published every step behind our Spark Arena runs on RLC Pro AI: setup, driver, pinned container, the recipe, and checks needed to reproduce CIQ’s DGX Spark tests.
Follow them on your own Spark: https://t.co/CwBsZ7qAsy
Open source moves forward when the community comes together.
Meet Rocky Linux users, contributors, and maintainers at #AllThingsOpen2026 in Raleigh, October 19 and 20. Ask questions, learn how to get involved, and hear what’s ahead.
Details: https://t.co/Y0gnHMTPi3
#RockyLinux #OpenSource
At #FutureForum2026, a clear point of global alignment emerged: sovereign AI matters to governments and enterprises alike.
CIQ President Bjorn Hovland joined conversations shaping why that control matters and how open, auditable infrastructure can make it achievable.
#SovereignAI #LeaveYourLane
Don’t take a benchmark result on faith. Rerun it.
Andrew Jorgensen published every step behind our Spark Arena runs on RLC Pro AI: setup, driver, pinned container, the recipe, and checks needed to reproduce CIQ’s DGX Spark tests.
Follow them on your own Spark: https://t.co/YQgl7KVRcf
Build on your DGX Spark then run the same OS in production.
RLC Pro AI puts the newest long-term kernel and validated NVIDIA drivers on your Spark, and your prototypes move to production without changing the OS.
Free developer licenses are available now: https://t.co/WocPJLvvSY
#DGXSpark #AIInfrastructure
The cheapest model for easy work can cost 37x more per correct answer when the questions get harder.
Damen Knight measured 27 local LLMs on a Strix Halo using real power data.
Read the first post in his new series on cost, accuracy, and workload fit: https://t.co/7B7rL9vDvu
Good news for Rocky Linux from CIQ Pro (RLC Pro) teams: Tenable Nessus scans can now accurately recognize the CVE fixes CIQ delivers, including on pinned LTS releases.
New plugins use CIQ advisories to provide more accurate findings, release by release.
Read: https://t.co/yNGlsNv5BC
AI may feel weightless, but its infrastructure is not.
At #FutureForum2026, CIQ President Bjorn Hovland will discuss AI sovereignty in “Who Owns the Intelligence?” at 10:35 AM at Network Central, then join the 4:45 PM debate: “This House Believes AI Will Do More to Heal the Planet Than Harm It” in Vision Hall.
Details: https://t.co/2Z5Wc41JMT
See sovereign AI move from GPU to governed production during #LATechWeek.
Join CIQ and @PhiniteAi at CIQ HQ for private model serving, agents with traces and approvals, and a live model upgrade with rollback and an audit trail.
For startup, VC, and enterprise leaders.
Oct. 17 | 1 to 6 PM | Los Angeles
Register: https://t.co/AFPaLgMypg
Scientists should not need a second career in computer science to use HPC.
From neuroscience and NIH’s Biowulf to Singularity/Apptainer and Fuzzball, David Godlove has built a career removing technical roadblocks to discovery.
Read his CIQ employee spotlight: https://t.co/ITo9OX1fUW
#HPC #Fuzzball
Updating Linux is a solved problem, right up until you have a hundred servers.
Brian Rieb breaks down where the usual models creak at fleet scale, and how CIQ Enterprise Linux Manager (ELM) lets your team pin production to what you tested and promote when you're ready.
Full walkthrough video included: https://t.co/5qFAj956mj
#EnterpriseLinux #Linux