From Local Firm to Global Design Force: NVIDIA vGPUs Power Cuhaci Peterson’s Evolution

After exploring various virtualization options, the company gravitated toward the Azure Virtual Desktop platform.

CP’s current solution includes NVIDIA A10-powered NVadsA10v5-series virtual machines running on Azure. Each machine features 36 cores and 880 GB of RAM, providing the necessary horsepower for graphics-intensive applications like Autodesk Revit and AutoCAD.

The environment supports 12–13 users per virtual machine, while maintaining performance through full GPU allocation. “We have 13 machines in the host pool right now for our United States users and another five for our team in India,” Robinson says.

While the path to full implementation wasn’t without challenges and learning opportunities, an initial pilot program allowed the IT team to identify and resolve issues before full deployment. “Any AEC firm jumping into VDI should set up a pilot program for testing,” Carpenter says. “You’re really going to find out quickly what obstacles you’re going to run into and how to overcome them.”

At Cuhaci Peterson, a 20-person test group provided critical feedback during a two-month trial period, highlighting issues with disk types, storage accounts, and virtual machine family selections. Carpenter and Robinson discovered, for example, that premium SSDs offered significantly better performance than standard hard drives, despite the higher cost. They also experimented with different approaches to FS Logix containers and storage accounts to find the optimal configuration.

Fortunately, the team was ready to rapidly scale their virtualization efforts when the pandemic hit. “Our plan was to gradually transition from Citrix to Windows Virtual Desktop, but all of a sudden we had to go into overdrive,” Carpenter recalls.

Microsoft representatives helped the team secure the needed NVIDIA GPU resources during a period of high demand. This relationship proved invaluable as the team navigated the transition during a time when virtualization resources were strained industry-wide.

“Back then we used NV16 series GPUs with NVIDIA® Tesla® M60s, which were the only thing available for years through Azure,” recalls Robinson. “The NVadsA10v5 series we’re using now has the A10 GPU, which provides a 60% higher user density.”

The move to NVIDIA GPU-powered cloud infrastructure has transformed how Cuhaci Peterson operates across multiple dimensions of its business. Perhaps most significantly, the firm’s geographic reach has expanded dramatically, freeing it from the constraints of physical office locations for both hiring and operations.

“Before implementing NVIDIA virtual GPUs, we were limited to hiring talent near our physical offices,” Carpenter explains. “Now we can recruit specialized expertise nationwide. Refrigeration engineers are hard to come by in the Orlando area, for example, so we’ve been recruiting talent from Kansas, the Carolinas, and Texas.”

This expanded hiring reach has even extended beyond US borders. In 2021, the firm established a 55-person team in India—something that would have been impractical without its virtual GPU infrastructure. “Sharing Revit and AutoCAD files would be a much bigger problem,” Carpenter says. “The fact that everyone just comes into one host pool, one file server, really allows for fluid connectivity.”

The 10.5-hour time difference between the U.S. and India has become an advantage, enabling round-the-clock productivity with effective handoffs. While the U.S. team focuses on client and project management, the India team concentrates on production work in Revit models and AutoCAD drawings, creating a complementary workflow that maximizes productivity.

The shift to NVIDIA GPU-powered virtual desktops has dramatically simplified IT management across the organization. Every desk now has an identical setup, allowing employees to work from any location by simply plugging in their laptop. Software updates are streamlined through a master image that can be pushed to all machines simultaneously, eliminating the labor-intensive process of updating individual workstations.

Remote employees receive standardized kits that include a laptop, docking station, monitors, and peripherals—replacing expensive specialized workstations and reducing hardware costs. “Because we mostly have firewalls and switches at the offices, maintenance and expansion are a lot easier,” Robinson explains. “When we open up new offices, we configure everything here, send it out, and hire a local contractor to set things up.”

This approach has generated substantial cost efficiencies throughout the organization. After identifying their optimal configuration, the team committed to three-year reserved instances for their NVIDIA-powered VMs. By leveraging other management tools, Cuhaci Peterson has reduced capital expenditures while gaining flexibility to scale resources on demand.

The team emphasizes that disciplined resource management is critical to maintaining these cost benefits—they continuously monitor and clean up unused resources to prevent unnecessary charges in their cloud environment.

While the firm’s virtual infrastructure primarily supports Revit modeling and AutoCAD drawings, Cuhaci Peterson is actively exploring how AI can further transform operations. The implementation of Windows 11 with Microsoft Copilot has already sparked excitement among end users, providing immediate productivity benefits.

More significantly, the firm recently participated in an AI workshop focused on leveraging artificial intelligence for proposal development. “One of the key areas we focused on was how Copilot and similar tools can be used for writing proposals and RFPs,” notes Carpenter. “It really gave us insight into the capabilities we can move forward with.”

Looking ahead, the company is investigating how AI can automate aspects of the design process itself. “We’ve been thinking for a long time about how automation can help designers and production staff with efficiency in their drawings and model renderings,” Carpenter explains.

The NVIDIA RTX Virtual Workstation and NVIDIA A10 GPUs powering its virtual infrastructure provide the computational foundation needed for these AI initiatives, positioning Cuhaci Peterson to leverage emerging technologies as they continue to evolve.

With the foundation of its NVIDIA GPU-powered infrastructure in place and AI initiatives underway, Cuhaci Peterson continues to refine its technology strategy. 

“Our next iteration of VDI will focus on application optimization,” Carpenter says. “When we first deployed, every user received the same resources and software. We’re now implementing discipline-specific resource allocation, so structural engineers, project architects, and other specialists only see the tools relevant to their work.”

Geographically, the firm is exploring additional Azure regions to better serve their international operations. “Currently, our India team connects to VMs in South Central US, which introduces some latency,” notes Carpenter. “As new fiber connections become available in India, we’re evaluating options to improve performance while maintaining our seamless collaboration model.”

These infrastructure improvements, combined with their AI initiatives, position Cuhaci Peterson to continue pushing the boundaries of what’s possible in architectural design and collaboration.