Plan · Build · Operate
One platform for the whole life of a datacenter.
SynapseDC plans the capacity, builds it under change control and operates it live, with power, cooling, racks, GPUs and network connected to the customers they serve. When something fails, you see who is affected, what it costs and what to do first.
What SynapseDC is
Plan, build and operate,
on one estate model
SynapseDC is the datacenter platform for colocation and GPU cloud operators. It plans the capacity, builds it as records under change control, and operates it live, with the building, the racks, the GPUs and the network connected to the customers and contracts they serve. Every fault arrives with who is affected, what it costs and the next action, and every number is the same on every screen.
Why it exists
Planned in spreadsheets, built in silos, operated across many tools.
A spreadsheet for capacity, a project tracker for the build, a DCIM for records, a BMS for the plant, an NMS for the network and a ticketing tool for incidents. None of them share a record, and none can say who is affected and what it is costing.
The building and the business never meet
The BMS knows the UPS is on battery. The CRM knows who the tenants are. Nothing in between can say which contracts are now at risk.
Impact is worked out by hand
When a link drops or a CRAH fails, the affected services, customers and money are reconstructed from memory and spreadsheets, after the event.
GPUs are blind to the facility
A CDU flow drop or a power cap shows up as a slow job. The GPU team sees the symptom; the facility team sees the cause. Nobody sees both.
Plan and build never reach operations
Capacity is planned in a spreadsheet, built from a project tracker and handed over as a PDF. The live estate starts from a record nobody trusts.
What you get
From the first plan to the live building, and the money.
Plan · Planner and capacity
From a capacity question to an approved plan.
Sellable space, power, cooling and GPUs net of what is already held, and a planner that sizes a new hall, prices it with payback, schedules it from lead times, checks the rules and routes approval by CapEx band.
- Templates and benchmarks for halls, plant, refreshes and new sites
- Plan pack, drawings and a review portal for sign-off
- Power economics: the bill, contract findings, cost per sold GPU-hour
Build · DCIM and delivery
Built as records, under change control.
Racks, devices, power chains and cabling down to the outlet and the patch cord. Every write carries a change request, and approved plans flow into projects, work orders, purchase orders and commissioning gates.
- Site → hall → row → rack → device → port
- Standard, open or new change request on every edit
- Commissioning L1 to L5 and earned value per project
Command Center
Every datacenter, ordered by what matters.
An estate health score you can explain, a live map, and every site worst-first. Then three columns: what is happening, who is affected and the recommended next step.
- Status follows the worst single issue, not an average
- Every figure opens the list behind it, with the same count
- Recommendations are ranked by money and customers at risk
3D datacenter layout
Building, floor, room, rack, device.
A 3D layout of every site with health beacons on the racks that need attention. Turn on power paths to see which UPS feeds a rack, or cable paths to follow a service to its trunk.
- Colour by health, temperature, rack power or connectivity
- Walk the aisle at eye level
- Trace one device's cords and cables end to end
GPU Fleet
What one bad GPU costs, per hour.
Per-GPU telemetry joined to jobs, customers and the building. The job with the biggest blast radius, the slowest job, GPU time lost and the work to redo since the last checkpoint.
- DCGM-style telemetry for NVIDIA and Intel accelerators
- Cooling and power cost per sold GPU-hour
- Where to place the next job
Revenue Impact
Which issue is costing the most right now?
Each impacted service is charged to exactly one issue. Revenue lost so far, SLA credits by tier, the next credit cliff and what is at stake, reconciled across issues, customers and sites.
- Down, degraded or exposed to a critical fault
- Credits net of those already locked in
- A cost-of-delay curve for the order of work
Facilities
Know how long the building holds.
The single-line power path from the utility to the rack, UPS minutes, generator kW and fuel hours, the cooling heat balance, and every single point of failure with a what-if.
- Hover any node to trace upstream and downstream
- Single points of failure found before a fault finds them
- Maintenance planned and booked as changes
One fault, end to end
What the first sixty seconds look like.
Integrations
Works with the systems you already run.
SynapseDC reads the building, the network, the servers and the GPUs through a read-only collector, and syncs incidents, customers, addresses and security events with the tools your teams use.
All integrationsCommon questions
Before you begin.
The questions that come up in nearly every first conversation.
We already have a DCIM. Why do we need this?
A DCIM keeps asset records and plans space and power. SynapseDC runs live operations on top of the records: incidents with impact, revenue, access, security and change control in one view. It can import your DCIM.
Will you control or change our plant?
No. SynapseDC is read-only by design. It reads the BMS, UPS, PDUs, meters, network and BMCs and never changes a set point or switches anything. People act through a change request.
Does data leave our site?
An on-site collector (active + standby) reads the equipment and sends outbound only, over mTLS on port 443. Nothing connects into the site.
Do you look inside tenant equipment or customer jobs?
In colocation, tenant gear is sealed: it is never polled. In a GPU cloud the GPUs are the operator's, so GPUs and jobs are monitored, but never a job's data, model or code.
How long does onboarding take?
A pilot on one site typically takes 6–10 weeks. Further integrations follow in waves of 4–12 weeks.
See your own estate in it.
Bring the fault your team keeps dealing with. We walk it end to end on a demo estate shaped like yours, then agree what a one-site pilot should prove.