Operate · GPU Fleet
One failed GPU stops a whole job. Know which, whose and what it costs.
Telemetry per GPU in the shape of the NVIDIA DCGM exporter (and Intel Gaudi / XPU Manager), joined to the job, the customer and the building. See the job with the biggest blast radius, the slowest job, GPU time lost per hour and the work to redo since the last checkpoint.
Why teams open GPU Fleet
Three questions GPU Fleet answers.
What people ask in the first minutes, and who asks it. GPU Fleet puts the answer on one screen, for NeoCloud operators.
Capabilities
What GPU Fleet does.
Jobs needing attention
The biggest blast radius and the slowest job, with lost now, lost so far, work to redo and the cause (Xid 79, CDU flow low…).
DCGM-style telemetry
Utilisation, SM and tensor activity, memory, temperatures, power, clocks, throttle reasons, ECC, row remap, PCIe and NVLink errors, last Xid, plus 24-hour history.
Clusters, nodes and models
NVIDIA B200 / H200 / H100 / A100 / L40S / T4 and Intel Gaudi 3 / Max 1550, with vendor-aware labels.
Network rails
InfiniBand, Ethernet front-end and OOB planes, with ports and errors per rail.
Cooling and power cost
Power per installed and per sold GPU-hour against the price, energy per job and per tenant, and the cost of idle allocations.
Where to place
Pods ranked by free healthy GPUs, rack headroom, cooling, contracted power and the marginal power cost per GPU-hour.
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.