Deployment, cabling and migrations
We take rooms from empty racks to powered on.
Full room builds, GPU and high density compute, network gear, PDUs and structured cabling we design as well as install. Crews travel from Boston to Northern Virginia and anywhere in between, on a two day minimum.
100 rack builds GPU and high density Design and install
What we deploy
Whole rooms, not just the easy racks.
Most deployment work is sold by the hour by people who do it occasionally. We run datacenter infrastructure for a living, which is a different starting point.
Cabinets and full rooms
Empty cabinets levelled, bonded and populated, through to complete room builds handed over powered and labelled.
Servers and GPU systems
Standard compute at volume, and GPU and high density hardware where weight, depth and rail work matter.
Switches, routers and fabric
Switching and routing racked, cabled and brought up. We configure as well as install when you want that.
PDUs and power distribution
PDU installation and whip dressing, with circuits balanced across feeds rather than filled in the order things arrived.
Structured cabling, designed and installed
We design the cable plant and then install it, so the drawing matches what the cabinet can physically take.
Migrations and decommissions
Moving a footprint between facilities, or clearing one out properly, with an audit rather than a guess at what is in there.
Track record
What that looks like in numbers.
Deployment vendors all claim to be fast. These are the figures from real builds, with crew sizes, so you can compare them against your own last deployment.
100 racks, empty to powered on, in about a week
Four technicians. Hundreds of servers and 200 switches and routers racked, cabled and brought up. The room started as 100 empty cabinets.
25 racks of GPU infrastructure, under a week, two technicians
GPU compute, network gear, PDUs and servers, with the structured cabling designed by us and then installed by us.
Multiple 100 rack deployments
Comparable room builds in our home region, where we also hold colocation space and run our own network.
Available on request
The Virginia deployments are under NDA, so we do not name customers or buildings. We can introduce you to references.
Those two Virginia numbers are not inconsistent. The first works out at roughly 25 racks per technician week, the second at half that. GPU and high density hardware is heavier, denser and less forgiving than standard compute, and on the Sterling job the cable plant was ours to design as well as install. Any vendor quoting you the same throughput for both has not done the second kind of work.
How an engagement works.
- Two day minimum on deployments. That is 16 hours of crew time. A crew travels to site, so below that the economics do not work for either of us.
- Hourly rate depends on the facility. Access rules, escort requirements and dock procedures change how long identical work takes, so we quote against the building rather than publish one rate.
- Travel is quoted case by case, because the figure depends on how many technicians the scope actually needs. You get it in writing before anyone is dispatched.
- No emergency surcharge and no same day dispatch fee.
- You authorise our technicians on your access list at the facility. That is done per customer, not per building, and it has to be in place before we can get in.
Where we work
- Massachusetts and New Hampshire, our home region
- Northern Virginia, Ashburn and Sterling
- Charlotte, where we hold our own space at Segra CLT2
For hourly work in Massachusetts and New Hampshire rather than a deployment, see remote and smart hands, which has no two day minimum.
Straight answers
Deployment questions.
What size deployment do you take on?
Deployments carry a two day minimum, which is 16 hours of crew time, because a crew travels to the site. Below that the travel does not make sense for either of us. We have run jobs from a handful of cabinets up to 100 rack room builds. In Northern Virginia we also have a technician available for smaller tasks, with a four hour minimum.
How fast do you actually work?
On a recent Ashburn build, four technicians took 100 empty racks to powered on in about a week, including hundreds of servers, 200 switches and routers, and all the cabling. On a Sterling GPU deployment, two technicians completed 25 racks in under a week. That is a deliberate difference. GPU and high density hardware is heavier, denser and less forgiving, and on that job the structured cabling was ours to design as well as install.
Can you deploy GPU and AI hardware?
Yes. We have deployed GPU compute, the network fabric around it, PDUs and the structured cabling to support it. This hardware is heavy and dense, and it punishes rushed rail work and sloppy cable management in ways that standard 1U compute does not. It is worth asking any deployment vendor what they have actually racked before you hand them a room.
Do you design cabling or only install it?
Both. We have designed structured cabling for deployments and then installed it ourselves, which removes the gap between the drawing and what is physically possible in the cabinet. Most deployment crews execute someone else drawing. We are happy to work from yours if you have one.
Where do you travel?
Massachusetts and New Hampshire as our home region, and Northern Virginia including Ashburn and Sterling. We are in Charlotte regularly for our own infrastructure at Segra CLT2. Outside those areas, ask and we will tell you honestly whether it is worth the travel.
How is a deployment priced?
Hourly rate depends on the facility, because access rules, escort requirements and dock procedures change how long the same work takes. Travel is quoted case by case, because the figure depends on how many technicians the scope needs. You get both numbers in writing before anyone is dispatched. There is no emergency surcharge and no same day dispatch fee.
Can you provide references?
Yes. Our Northern Virginia deployments are under NDA so we do not name the customers or the buildings, but we can put you in touch with references on request.
Who actually turns up?
Our own technicians. Axcelx has been operating datacenter infrastructure since 1997 and runs its own network under AS33083, so the people on your deployment work in datacenters as their job rather than as a dispatch assignment.
Tell us what the room has to become.
Send us the scope, the facility and the hardware. We will tell you how many technicians it needs, how long it will take and what the travel costs, before anyone gets in a van.