Oktoberfest Tech Edition — Build AI-Ready HPC Data Centers That Scale

23 oktober 2025 13:00 - 23:30 @ Hälsingegatan 10, 414 63 Göteborg, Sverige

Oktoberfest Tech Edition — Build AI-Ready HPC Data Centers That Scale

If you’re planning, building, or funding an AI/HPC data center, this is your afternoon. Join practitioners from NVIDIA Networking, ProLabs, and Aixia for a fast-paced, vendor-neutral look at the choices that make-or-break AI performance and your TCO. We go beyond buzzwords to the engineering that matters lossless Ethernet (RoCEv2), leaf-spine/CLOS fabrics, 400/800G transitions, ECMP and congestion control, and the optics/cabling strategy (DAC vs AOC vs pluggables, link budgets, MPO/MTP) that keeps you from stranding capacity.

You’ll see concrete patterns for scaling from a single GPU pod to multi-cluster deployments, how to budget for bandwidth vs power and cooling, and where observability pays back in days, not months. Most importantly, you’ll learn how the right network and Layer-1 decisions de-risk timelines, shorten lead times, protect upgrade paths to 800G/1.6T, and deliver predictable ROI.

Why attend

Translate AI workload demands into clear network and optics requirements

Cut capex/opex with the right fabric, transceivers, and fiber plan

Avoid interoperability traps and supply-chain surprises

Walk away with sample BOM ideas and a practical checklist for your next phase

We’ll wrap the knowledge share with an Oktoberfest mingle—food, drinks, and great conversations with peers who build and run advanced data centers.
Seats are limited. Save your spot and take the guesswork out of building an HPC data center that can handle real AI workloads.

Why attend?
AI workloads demand precision engineering: lossless fabrics, low latency, the right optics, and a network that scales without blowing up the budget. This afternoon zeroes in on how to build HPC data centers for AI — with a strong focus on networking, optics, and fabric design — and how these choices drive TCO, risk and returns for investors.

Key takeaways

AI-ready fabrics: leaf-spine/CLOS, 400/800G, buffering/queues, congestion management and practical lossless Ethernet (RoCEv2).
Optics & cabling that pay off: DAC vs AOC vs pluggables, link budgets, MPO/MTP strategy, and avoiding stranded capacity.
Scaling without surprises: from one cluster to many — capacity planning, oversubscription, ECMP, telemetry and observability.
CapEx/OpEx & upgrade paths: how switches, transceivers and fiber choices impact TCO, lead-times and future 800G/1.6T transitions.
Sourcing & risk: sample BOMs, supply chain realities, interoperability and minimizing lock-in.

Who should join?
Leaders in design, operations and investment of advanced data centers: Heads of Infrastructure/Operations, CTOs, DC Managers, Network/Infra Architects, and CFOs/Investment leads across AI/HPC, enterprise and colocation.

Speakers & focus

NVIDIA Networking – Gadi Godanyan (EMEA)
- Front-line insights on building AI-ready, scalable networks for training/inference; lessons from low-latency Ethernet storage fabrics and AI cloud networking.
ProLabs – Ray Hagen (VP, Product Line Management)
- Optical decisions that move the needle on cost, performance and reliability in GPU clusters: transceivers, fiber, and Layer-1 architecture to accelerate AI data centers.
Aixia — Host & moderator with Nordic delivery experience in advanced data center solutions.

At a glance
13:00–16:00 Seminar (NVIDIA Networking · ProLabs · Aixia)
16:00–late Oktoberfest — classic food, drinks and activities.

Thank you for registering!

We look forward to seeing you on October 23 at Aixia’s office.

Tack så mycket

Vi kontaktar er så snart vi kan.

Sorry, this event is full

Sorry, this event is canceled

Sorry, this event is no longer available

Ray Hagen

VP – Product Line Management

Ray Hagen is responsible for the global product strategy and development of ProLabs’ transceiver portfolio. With over 20 years of experience in telecom and technology—including roles in data communications, data centers, and managed hosting—he brings broad expertise spanning both technical design and business strategy. He has led AddOn’s efforts on new product initiatives for AI and ML cluster networking, with a focus on optical Layer 1 networks and fiber connectivity. By driving value-creating solutions that reduce total cost of ownership and strengthen GPU-based cloud strategies, he helps accelerate the development of next-generation data centers.

Gadi Godanyan

Senior AI Networking Lead, EMEA — NVIDIA

Gadi Godanyan is responsible for the strategy and sales of AI networking across Europe, the Middle East, and Africa for NVIDIA. With over a decade of experience in the networking industry—including key roles in Ethernet switch management at Mellanox and NVIDIA—he brings deep expertise in powering AI-driven data center infrastructures. Gadi is a frequent speaker on advanced networking topics. He has shared insights on low-latency Ethernet Storage Fabrics at Spectrum Scale User Group events and presented technical perspectives on AI Cloud networking during NVIDIA’s “Power Networks for AI Clouds” workshop. Join us as Gadi explores the future of AI-ready networking and its impact on high-performance, scalable infrastructure.

Aixia + Partner

From GPU-optimized architectures to green power and heat reuse, Aixia and partners share the playbook for power-hungry AI data centers—build AI the planet can afford by pairing peak performance with a low-carbon footprint.