By Colin Brix, Sr. Director, Product Marketing, AMI Business Unit, Lattice Semiconductor
AI infrastructure is scaling beyond individual servers to racks, pods, and entire data centers. Yet as infrastructure becomes more distributed and complex, the systems used to manage it remain highly fragmented across vendors, devices, and architectures.
At the 2026 Open Control Plane Summit, hosted by AMI, a Lattice Semiconductor company, during the 2026 OCP Global Summit in San Jose, we brought together leaders from across the infrastructure ecosystem to explore a bigger question: What will it take to create a common, open, and interoperable control plane for AI infrastructure?
The conversations from our Summit are now available to watch on our 2026 OCP Global Summit mini-site, offering perspectives from technology providers, infrastructure operators, and the open-source community on some of the industry’s most pressing management challenges.
Across the Summit sessions, speakers examine questions including:
- How can organizations gain greater visibility and control across increasingly heterogeneous infrastructure?
- What do operators need to simplify firmware management and overcome inconsistent management interfaces?
- How can open standards, telemetry, and attestation build the foundation for more autonomous infrastructure?
- What role can open-source development, including AMI MegaRAC OneTree Community Edition and OpenBMC, play in creating more consistent management building blocks?
Here are five conversations from the Open Control Plane Summit worth watching, and the key questions each one helps address.
1. How Do We Control Infrastructure Beyond the Server?
As compute moves beyond individual servers to racks, pods, and entire data centers, traditional server-level management is no longer enough.
In their opening remarks at the Summit, Lattice’s Sanjoy Maity and Mamta Gupta examine three challenges created by that transition: coordinating disaggregated components across fleets, managing complexity across processor and firmware architectures, and securing systems assembled from components supplied by many vendors.
The discussion also shows why the control plane can’t be treated as software alone. FPGAs, BMCs, firmware, telemetry, power systems, and security all have roles to play in determining how infrastructure responds.
“Silicon to Software: Effective Control for AI Factories”
Explore how the control plane must evolve as AI infrastructure scales from servers to racks and data centers: Watch: Silicon to Software: Effective Control for AI Factories →
2. Why Systems Management is Becoming OCP’s Next Frontier
Complex infrastructure becomes manageable when it can be broken into understandable, interoperable components.
In “Systems Management as OCP’s Next Frontier,” Zane Ball, Chief Technical Officer of the OCP Foundation, looks at what that principle means for increasingly complex data center infrastructure. His recommendations center on three priorities: making telemetry visible across devices, enabling attestation across hardware and software, and progressively allowing infrastructure to take more autonomous action.
“Systems Management as OCP’s Next Frontier”
Hear why visibility, attestation, and open systems management are becoming essential building blocks for increasingly complex infrastructure: Watch: Systems Management as OCP’s Next Frontier →
3. What Operators Actually Need from an Open Control Plane
Jacob Yundt of CoreWeave, Jonmichael Hands of FarmGPU, and Manoj Roge of Google joined our Summit operator panel, which surfaced issues around inconsistent management interfaces, firmware updates, and coordination across different vendors.
What happens when a firmware fix requires coordination across five vendors and a platform reboot? Or when nominal support for an industry-standard interface still varies widely from one piece of equipment to another?
The operator panel turns the abstract idea of a control plane into real infrastructure problems. These experienced speakers discuss inconsistent management interfaces, lengthy firmware-update processes, facility equipment management, and gaps in Redfish support outside traditional servers.
As Mamta Gupta summarized it: “This is not a data flow problem. All of these are control plane problems.”
“The Road to Autonomous Data Centers: What Operators Need”
Hear directly from infrastructure operators about the management challenges they face today, and what they need to move toward greater automation. Watch: The Road to Autonomous Data Centers: What Operators Need →
4. What “Open First” Means for AMI
An open control plane also requires a different approach to developing the firmware underneath it.
Zachary Bobroff and Pravinash Jeyapaul explain AMI’s move toward an Open First development model, where AMI’s Community Edition products become the active development tree and commercial long-term-support editions are built from it.
The discussion connects that strategy to AMI MegaRAC OneTree Community Edition, the OpenBMC project, and AMI’s broader participation in community development. The pair outline how AMI integrates multiple OpenBMC implementations into a single silicon-neutral stack through AMI MegaRAC OneTree Community Edition and the benefits that flow from this approach.
“AMI Community Edition Moves to an ‘Open First’ Model”
See how AMI’s Open First approach and AMI MegaRAC OneTree Community Edition are helping advance open, community-driven firmware development. Watch: AMI Community Edition Moves to an “Open First” Model →
5. Why Openness Matters Operationally
Interoperability isn’t simply an architectural preference. Fragmented management can create operational risk and prevent operators from using infrastructure as efficiently as possible.
In the Summit’s closing panel, Rebecca Weekly of brk tech made the point that having common lower layers of the control plane can reduce the “bespoke islands” that prevent systems from communicating with one another. The conversation also explored stranded power, infrastructure visibility, and the role of OCP S.A.F.E. in establishing trust across vendors.
Watch “Open Building Blocks to Autonomous Infrastructure”
Explore how interoperability, common management building blocks, and industry collaboration can help lay the foundation for autonomous infrastructure. Watch: Open Building Blocks to Autonomous Infrastructure →
In Closing: A Call for Collaboration
No single vendor can solve AI infrastructure management alone.
Across the Open Control Plane Summit conversations, one theme stands out: before autonomous operations become practical, infrastructure needs better visibility, more consistent interfaces, and shared management building blocks.
That means making telemetry accessible across devices, improving attestation and security, expanding interoperability beyond the server, and developing more of the underlying technology in the open. Those themes also underpin AMI’s Open First strategy and the development philosophy behind AMI MegaRAC OneTree Community Edition.
As B Parthiban, CVP of AMI’s Boot Firmware Group, put it in the Summit’s closing session: “If you want to go fast, you do it alone. But if you want to go far, you do it together with the community.”
The conversations from the Open Control Plane Summit make one thing clear: the future of AI infrastructure will be built through collaboration. Creating a more open, interoperable, and autonomous data center ecosystem requires participation across the entire industry.
AMI invites technology providers, operators, developers, and the open-source community to join us in building the open control plane that will power the next generation of AI infrastructure.
Watch: Open Control Plane Summit Closing Remarks: The Call to Action →












