Spectro Cloud at NVIDIA GTC Berlin 2026
We've taken over Gibson Bar for two days, a few minutes' walk from the expo hall and the keynote. Come and see PaletteAI running on NVIDIA infrastructure, talk through your AI factory plans with our engineers, and get a proper coffee. Or something stronger.
Reserve your spot
Skip the queue. Come to the suite.
STATION Berlin will be packed and the coffee lines long. Our suite is about 10 minutes' walk from the expo hall and six from the Tempodrom, so you can catch the keynote and be with us before the crowd clears. Drop in between sessions, or book a slot with our team and we'll have a room ready.
What's in the suite
Two demo stations, four meeting rooms, one very good bar
The demo stations run PaletteAI live: designing a governance-approved AI stack, deploying it onto NVIDIA GPUs, and watching utilization from a single console. No slideware.
The meeting rooms are bookable for private sessions with our solution architects. Bring your AI factory design, your GPU-as-a-service plans or your awkward migration questions. The lounge is open all day, with food and drinks on us.
Book a meeting roomWednesday evening reception
October 21, 19:00–22:00
Once the keynote day winds down, join us at Gibson Bar for drinks, canapés and a DJ who has been told, firmly, to keep it at conversation volume.
It's an invitation-only gathering of around 50 people building AI infrastructure across Europe, so you'll leave with a few useful numbers in your phone. Places are limited by the venue, so reserve early.
Reserve your placeSuite hours
Two days, three moments
When to find us
What we'll be showing
PaletteAI on NVIDIA, from data center to edge
PaletteAI is validated for NVIDIA's Enterprise AI Factory validated designs and covers the full stack: Blackwell GPUs, BlueField-3 DPUs, Spectrum-X networking, NIM microservices and NeMo. Here's what will be running on the demo stations.
Design a repeatable, governance-approved AI stack once and deploy it across your NVIDIA estate. Platform teams keep control; data scientists get self-service environments in minutes.
For service providers and sovereign clouds: multi-tenant GPU pools, metering, and model serving with NVIDIA NIM, run from one platform with the audit trail your regulator expects.
The same blueprints and the same console, deployed to small sites and constrained hardware. Palette has managed Kubernetes at the edge for years; PaletteAI puts inference on top.
Can't make it to the suite?
Our team is in Berlin all week. If a 30-minute conversation somewhere near STATION Berlin suits you better, tell us when and we'll come to you.
Book a meeting →