Compute, chips, edge and infrastructure for AI

Hardware track

Every model runs on something, and that something has a price, a power draw and a ceiling. This track is about the physical layer: GPUs and chips, edge devices, data centres, energy, deployment, performance and the cost choices nobody puts on the slide.

📅 November 13, 2026 📍 Forum Groningen · Rabo Studio
About the track

What AI actually runs on.

This track is aimed at hardware engineers, infrastructure teams, data-centre specialists, chip and edge specialists, technical buyers and researchers — the people who get the bill and the thermal budget.

The speakers are engineers, infrastructure specialists and hardware companies explaining compute and acceleration, edge deployment, energy, cost and the scaling choices behind them. Numbers, benchmarks and trade-offs rather than product tours.

Seven sessions in aiGrunn's central 3-2-2 day format, in the Rabo Studio at Forum Groningen.

GPUs Chips Edge devices Data centres Energy use Model deployment Performance Cost
Track at a glance
  • Theme Compute, chips, edge and infrastructure for Artificial Intelligence
  • For Hardware engineers, infrastructure teams, data-centre specialists, chip and edge specialists, technical buyers and researchers
  • Format Seven sessions in the central 3-2-2 day format
  • Date Friday November 13, 2026 · Rabo Studio, Forum Groningen
  • Track leads Jaap Kommandeur, Jeroen Bos and Tim de Zwart (Bossers & Cnossen)
Tickets

Join 400+ professionals
at aiGrunn 2026

Call for proposals

Do you know what it costs to run?

We are looking for engineers, infrastructure specialists and hardware companies who can explain compute, acceleration, edge deployment, energy, cost and scaling choices with real numbers attached.

Benchmarks, failure modes and the trade-off you eventually settled on. Submit your proposal for aiGrunn 2026 and join a lineup of speakers from across the Netherlands and beyond.

Submit your talk →
What we are looking for
  • Compute, chips, edge devices, data centres, energy, deployment, performance or cost
  • Bring the numbers: benchmarks, power draw, throughput, total cost of running it
  • No product tours. Vendors are welcome when the talk teaches something
  • No sales pitches. All talks must deliver genuine learning value
  • Slots: 30 minutes including Q&A
  • English submissions only
  • All talks are recorded and published on YouTube
Blast from the past

Compute and constraints,
from previous editions.

A selection of infrastructure- and performance-minded talks from aiGrunn 2025, 2024 and 2023. Looking for the broader technical ones? They are on the tech track page.

2025 3 talks
2024 3 talks
2023 2 talks