AS400304UP Fabric100 GbE GPUONLINE StorageNVMe Uptime 99.99% · Toronto
Toronto · Canada · HPC Infrastructure

Compute, at the edge of the possible GPU. Bare metal.
Storage.

Synteq HPC CA Inc. is the Canadian entity of Synteq — delivering high-performance compute infrastructure to developers, researchers, and enterprises. GPU clusters, bare metal, virtualized instances, and enterprise storage, on our own network.

Network AS400304
Fabric 100 GbE
Region CA · Toronto
FIG. 01 — NETWORK TOPOLOGY LIVE
YYZ TORONTO PRIMARY DFW DALLAS SOF SOFIA POP · 04 AS400304 · 100 GbE FABRIC
Points of presence: 4 Status: nominal
01 Infrastructure

Four products, one platform.

Compute, storage, and network — provisioned together, not sold separately.
Product · 01GPU

GPU Compute

Dedicated and shared GPU instances for training, inference, and rendering. Blackwell and Hopper generations, provisioned with the interconnect and storage that actually feed them.

Blackwell RTX Hopper NVLink
Product · 02Bare Metal

Bare Metal

Single-tenant servers, no hypervisor, full root. Latest-generation AMD EPYC and Intel Xeon platforms with the memory channels and NVMe lanes the workloads require.

EPYC 9355P Zen 5 Single-tenant
Product · 03Virtualized

Virtualized Compute

Elastic instances on the same fabric as our bare metal and GPU products — for workloads that need to scale quickly, or that do not justify dedicated hardware.

Elastic API-driven Same fabric
Product · 04Storage

Enterprise Storage

NVMe-backed block and object storage, attached locally to the compute that uses it. Built for datasets that do not fit comfortably in memory and cannot wait for a network round-trip.

NVMe Block + Object Local attach
02 Network

Our own ASN. Our own fabric.

AS400304 — operated end to end, not resold.
YYZ Toronto · Primary AS400304 Operational < 1 ms
DFW Dallas · US AS400304 Operational ~ 28 ms
SOF Sofia · EU AS400304 Operational ~ 95 ms
POP Toronto · Edge AS400304 Provisioning —
03 Method

How we build it.

Four operating principles.
01

Own the fabric

We operate our own network — AS400304 — rather than reselling transit. When latency moves, we see it first. When a route needs changing, we change it ourselves.

02

Provision at the metal

GPU, bare metal, and storage are attached to the same fabric. A workload does not lose half its throughput crossing a vendor boundary we do not control.

03

Price the whole stack

Compute is cheap until it is not. We quote the network, the storage, and the interconnect alongside the CPU and GPU — so the number you plan against is the number you pay.

04

Operate in the open

Status, maintenance windows, and incident reports are published to customers in real time. An outage we did not tell you about is a worse outage than the one you noticed.

04 Contact

Correspondence.

Sales, engineering, and general enquiries.

For deployment planning, pricing, or general correspondence relating to Synteq HPC CA Inc., please write to us directly.

contact@synteqhpcca.site
Registered Office Synteq HPC CA Inc.
600-55 University Ave
Toronto, ON M5J 2H7
Canada
Corporation No. 1819238-3
Network AS400304