SHARE

Add TechRepublic as a preferred source

How InstaDeep Scales Agentic AI for PCB Design With Dell AI Factory With NVIDIA

Sponsored By Dell AI Factory with NVIDIA Logo

The following is sponsored content. It may not reflect the views of our editorial staff.

Sep 29, 2026

4 minute read

Transcription

AI is actually delivering. You see it every day. At Instadep, the mission is to uh build AI that benefits everyone. The Deli factory with Nvidia is the fundamental building block that allows Instad to deliver its magic. Printed secret boards, PCBs, are everywhere. They're in your smartphone. They're in your fridge. They're even in the the screens behind me. This is the basic component of all electronics. But incredibly, they are designed manually.

And I was like, we can do this with AI. AI innovation requires a robust ecosystem of hardware, software, and expertise. By partnering with Dell Technologies and leveraging their AI factory solution with Nvidia, we have helped Instadip accelerate their AI model training by up to 40%. We've developed AI based agents for the placing and routing of PCBs using reinforcement learning that help automate the complex design process of printed circuit boards.

PCB designers can really make their vision come to life in a much more accelerated time frame. SEC, Dell Technologies and Instadip form a powerful AI ecosystem. Having the right partners is critical for us because we are focused on innovation and building world first products. This is not about LLMs. It's about agents AI systems that are so smart that they can evolve in a complex universe make decisions see the outcome and learn from there. This is a fundamental revolution.

I believe that this will generate insane economies of scale. This will allow smaller, leaner teams to accomplish the work that normally would be done by larger teams. This is really like a new phase of the AI revolution and we're very excited to be part of it.

This transcript was generated automatically from the video's captions and may contain errors.

InstaDeep is applying agentic AI to one of electronics manufacturing’s most complex tasks: designing printed circuit boards. Its AI-based agents use reinforcement learning to automate PCB placement and routing, helping designers move from concept to finished designs faster. Supporting those workloads requires substantial compute capacity and an infrastructure stack built to train AI models efficiently. InstaDeep uses Dell AI Factory with NVIDIA, including Dell PowerEdge XE servers and NVIDIA GPUs, with SCC France supporting the deployment. InstaDeep’s deployment has accelerated AI model training by up to 40%, reduced infrastructure costs by 30% and helped cut PCB design time by 10x.

Key takeaways
How InstaDeep is scaling agentic AI in production
Technologies behind InstaDeep’s agentic AI deployment
What the InstaDeep deployment shows about moving AI beyond experimentation
How Dell and NVIDIA extend the AI infrastructure stack

Key takeaways

InstaDeep uses AI agents to automate PCB design. The company developed reinforcement learning-based agents that handle the placement and routing of printed circuit boards, tasks that have traditionally required significant manual design work.
The project is built around agentic AI rather than an LLM use case. InstaDeep CEO and co-founder Karim Beguir describes systems that make decisions, observe the results and learn from those outcomes as they operate in complex environments.
Dell and NVIDIA infrastructure provides the compute foundation. Dell AI Factory with NVIDIA, including Dell PowerEdge XE servers and NVIDIA GPUs, supplies the accelerated computing environment used for the workload, while NVIDIA InfiniBand provides high-speed networking within the infrastructure.
The deployment has produced measurable improvements. InstaDeep has accelerated AI model training by up to 40%, reduced infrastructure costs by 30% and shortened PCB design time by 10x.

How InstaDeep is scaling agentic AI in production

Printed circuit boards sit inside nearly every category of electronic device, but designing them involves a complicated placement and routing process. InstaDeep developed AI-based agents that use reinforcement learning to automate that work, allowing the system to make decisions, evaluate their results and learn as it works through the design problem.

Running that kind of system at scale requires more than the AI model itself. InstaDeep needed additional compute capacity to train its models and support increasingly complex PCB workloads. Its deployment uses Dell AI Factory with NVIDIA, with Dell PowerEdge XE servers and NVIDIA GPUs providing accelerated compute and NVIDIA InfiniBand supporting the high-speed network fabric. SCC France worked with Dell and InstaDeep on deployment and integration.

The infrastructure is supporting a working AI application with measurable gains in training efficiency, infrastructure costs and PCB design speed. InstaDeep increased compute power by 10x, improved AI model training efficiency by up to 40% and reduced infrastructure costs by 30%. Its agentic AI approach has also helped reduce PCB design time by 10x.

Beguir sees the same agentic model extending beyond PCB design. These systems can operate in complex environments, make decisions, evaluate the results and learn from those outcomes. He argues that approach could allow smaller teams to accomplish work that previously required much larger groups.

Technologies behind InstaDeep’s agentic AI deployment

Technology	Role in the InstaDeep deployment
Dell AI Factory with NVIDIA	Provides the integrated Dell and NVIDIA infrastructure foundation supporting InstaDeep’s AI workloads.
Dell PowerEdge XE servers	Supply accelerated compute capacity for AI model training and agentic AI workloads.
NVIDIA GPUs	Provide GPU acceleration for InstaDeep’s AI training environment.
NVIDIA InfiniBand switches	Provide the high-speed networking fabric connecting the AI infrastructure.
InstaDeep AI agents	Use reinforcement learning to automate PCB placement and routing and learn from the outcomes of their decisions.

Dell PowerEdge XE servers, NVIDIA GPUs and NVIDIA InfiniBand provide the compute and networking foundation for InstaDeep’s AI workloads. On top of that infrastructure, InstaDeep’s reinforcement learning-based agents automate PCB placement and routing by making decisions, evaluating the results and learning from those outcomes.

What the InstaDeep deployment shows about moving AI beyond experimentation

InstaDeep’s experience highlights several requirements that become more important when an AI project moves into operational use:

Enough compute capacity to support repeated model training. InstaDeep expanded its compute resources as it scaled its AI work, with Dell reporting a 10x increase in compute power.
Accelerated hardware suited to the workload. Dell PowerEdge XE servers and NVIDIA GPUs provide the computing foundation for training InstaDeep’s models.
Fast networking between systems. NVIDIA InfiniBand supports the high-speed fabric connecting the Dell AI Factory environment.
Infrastructure integration and deployment expertise. SCC France worked with Dell and InstaDeep to integrate the environment rather than treating compute, networking and deployment as separate projects.
A workload with measurable operational results. The system is being used to automate PCB design, with Dell reporting faster model training, lower infrastructure costs and shorter design cycles.

How Dell and NVIDIA extend the AI infrastructure stack

Dell and NVIDIA’s AI infrastructure portfolio extends beyond accelerated compute and networking to include production reference architectures and certified storage systems.

The Dell AI Factory with NVIDIA includes reference architectures based on Dell PowerEdge XE servers with NVIDIA accelerators such as the H200 and RTX PRO 6000 Blackwell Server Edition. These architectures bring together compute, networking, storage and infrastructure software in integrated configurations for enterprise AI workloads.

For storage, Dell PowerScale F710 is NVIDIA-Certified Storage and is also certified for NVIDIA DGX SuperPOD. The certification covers requirements tied to AI workload performance, system integration and compatibility with NVIDIA accelerated computing environments.