Xcelerated Compute Show 2026: Bare Metal for AI Infrastructure

The Xcelerated Compute Show brought the AI infrastructure industry to the Marriott Marquis in Times Square on 23 and 24 March 2026, and velia.net was there. The organiser gathers hyperscale, neocloud and enterprise infrastructure teams in one room.

AI demand is mostly discussed as a capacity problem. Tenancy is the part that gets skipped: who else is on the machine, and in which city it stands. We run servers in New York, and we went to have that conversation at home.

Networking and Innovation

A conference does what a call cannot: it puts the team with the problem and the people with the hardware at the same table. These were the conversations we went to New York to have.

• Dedicated servers in New York for inference, data pipelines and platform teams.
• What one physical machine with no other customer on it changes for consistency.
• One operator across Frankfurt, Strasbourg, New York, Miami, St. Louis and Singapore.
• Monthly terms and zero setup fee for teams that are still finding their load.

Key Points of our Attendance

Our focus at the Xcelerated Compute Show stayed on four themes:

• The worst minute matters more than the average one.
• No hypervisor between the workload and the metal.
• No shared tenancy, so nobody else's workload sits next to yours.
• A named city on the contract, not a region in a console.

Looking ahead from the Xcelerated Compute Show

See where we go next on our events page. If you are there, let us grab a coffee and talk shop. You can find more updates on our LinkedIn profile.