Teams will use Lenovo ThinkStation PGX systems powered by the NVIDIA GB10 Grace Blackwell Superchip, orchestrated by CIQ Fuzzball, to train, serve, and benchmark AI workloads in Chicago from November 13–15.
RENO, NV, October 7, 2026 - CIQ, the founding commercial sponsor of Rocky Linux, today announced the CIQ Supercomputing Local AI Hackathon, a hands-on technical build weekend in Chicago right before Supercomputing 2026 (SC26). The hackathon is presented with NVIDIA and Drive Capital, and supported by Lenovo, VAST Data, and Moonlite. Up to 20 teams will each get their own multi-node cluster and one challenge: build a real AI workload that makes full use of it.
Most AI work starts on a single machine and gets stuck when it needs more memory, throughput, or concurrency than that machine can give. The hackathon puts developers right at that step. Each team gets two Lenovo ThinkStation PGX systems powered by the NVIDIA GB10 Grace Blackwell Superchip, connected into a cluster and orchestrated through CIQ Fuzzball.
"Some of the best work in this industry happens when you hand talented people real compute and a hard problem," said Bjorn Hovland, President of CIQ. "Fuzzball handles the orchestration so teams can focus on models and performance. We want to see what that second node unlocks, and we want the SC26 community to see it too."
Teams will compete in one of four tracks:
- Multi-Node Training: Train or fine-tune a model across multiple nodes.
- Inference at Scale: Increase throughput, concurrency, model capacity, or performance with multi-node inference.
- Scientific AI: Apply distributed AI to a scientific or research workload.
- Cluster and Scheduling: Build around orchestration, scheduling, resource allocation, or utilization.
The top three teams win SC26 conference passes and will be recognized at the CIQ booth during SC26 and featured on an NVIDIA Developer Livestream.
Participation is in person, with teams of three to five members who have multi-node or HPC experience. Registration requires approval. To apply, visit luma.com/local-ai-hack-chicago.