Mark III Systems Secured Nvidia AI Factory Deal
The firm will deploy Vera Rubin systems for a Fortune 100 enterprise to bolster private AI capabilities.
Updated on Sept. 30, 2026 in Data Centers
Live Poll
Is now a good time for large companies to move AI infrastructure to private data centers?
Mark III Systems has signed an agreement to deploy an Nvidia AI factory for a Fortune 100 enterprise. The project utilizes the new Vera Rubin NVL72 platform, which helps businesses run open models within their own data centers.
Why it matters
Enterprises are increasingly seeking to manage their own AI infrastructure to control costs and budget requirements. This move allows companies to avoid dependence on the expensive and unpredictable frontier models offered by firms like OpenAI and Anthropic.
The Vera Rubin NVL72 system architecture links 72 Nvidia GPUs and 36 Nvidia Vera CPUs. Nvidia began shipping these high-performance platforms during the summer of 2026.
The players
Mark III Systems
Headquartered in Houston, this company specializes in delivering high-end enterprise IT infrastructure and Nvidia-based AI platforms.
TD Synnex
Based in Fremont, California, this global distributor provides supply chain and IT integration services to help companies deploy complex technology solutions.
Nvidia
This technology corporation designs the GPUs and central processing units that power modern artificial intelligence and data center environments.
The details
Mark III Systems will oversee the design, integration, deployment, and ongoing co-administration of the AI factory for its client. The company has a history of building Nvidia-based infrastructure, having successfully delivered eight AI server clusters alongside TD Synnex since 2023.
Timeline
Mark III began delivering AI server clusters with TD Synnex in 2023.
Nvidia started shipping Vera Rubin platforms in the summer of 2026.
TD Synnex officials revealed the agreement during a third-quarter earnings call last week in September 2026.
The Tech Race
This deployment follows the path set by the Nvidia DGX SuperPod, serving as an evolution in enterprise-grade AI infrastructure. It signals a shift where businesses move away from relying solely on external frontier models to operating their own high-performance data centers.
For enterprise clients, this shift allows for better control over the costs and data management associated with running complex AI operations. It reduces reliance on external subscription models by bringing powerful computing capabilities directly into private corporate data centers.
The takeaway
Companies are prioritizing self-hosted AI infrastructure to gain operational independence from major model providers. As Vera Rubin systems ramp up, businesses that control their own hardware may find themselves better positioned to optimize AI costs over the long term.
Further reading
Learn more about the infrastructure behind the latest AI developments in our Data Centers section.
Live Poll
Is now a good time for large companies to move AI infrastructure to private data centers?
