Dedicated GPU Infrastructure
NVIDIA Blackwell compute on dedicated hardware.
MPI provides bare-metal GPU infrastructure powered by NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs.
Our systems operate from DataBank HOU1 in Houston and can scale from dedicated servers to larger multi-node deployments.
Dedicated compute without a shared cloud layer.
MPI customers receive direct access to complete physical servers. Your team controls the operating system, software stack, models, data, and workload scheduling.
There is no shared GPU tenancy and no virtualization layer between your workloads and the hardware.
- Dedicated bare-metal access
- Reserved GPU capacity
- Multi-node deployments
- Direct support from MPI’s infrastructure team
- Optional deployment and management assistance
Server configuration
MPI’s bare-metal servers are configured for sustained multi-GPU workloads in a production data-center environment.
- 8 NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs
- 96 GB of GDDR7 ECC memory per GPU
- 768 GB of total GPU memory
- 2 AMD EPYC 9555 processors
- 128 physical CPU cores and 256 threads
- 768 GB of DDR5-6400 ECC system memory
- 2 × 3.84 TB U.2 PCIe 5.0 NVMe data drives
- 960 GB dedicated NVMe boot drive
- PCIe 5.0 x16 connectivity for each GPU
- Dual 10 GbE network interfaces
- Redundant hot-swappable power supplies
- Dedicated out-of-band server management
The systems support AI training, fine-tuning, inference, rendering, simulation, scientific computing, video processing, and other GPU-intensive workloads.
NVIDIA RTX PRO GPU specificationsCapacity that can grow with the workload.
MPI supports individual dedicated servers and larger reserved deployments.
Our team works directly with customers to plan capacity, procure additional hardware, and add nodes as requirements change. Customers work directly with the people responsible for deploying and maintaining their infrastructure.

Houston, Texas
Deployed at DataBank HOU1.
MPI infrastructure operates from DataBank’s Westway Park I data center in Houston.
HOU1 is an HPC-capable facility with N+1 power and cooling design, 15.6 MW of critical IT load, multiple connectivity options, and around-the-clock operations and remote hands.
5150 Westway Park Blvd
Houston, Texas 77041DataBank HOU1 facility specifications
Based in Austin, Texas
Infrastructure operated by people who know it.
Machine Path Industries is an Austin-based compute infrastructure company with experience in data centers, servers, networking, and production systems.
We provide companies with dedicated GPU capacity and the operational expertise required to deploy and expand it.
301 Congress Ave, Suite 1200
Austin, TX 78701

Looking for dedicated GPU capacity?
Contact MPI to discuss availability, hardware requirements, and deployment options.

