InfraPartners and Phaidra Partner to Integrate AI Agents With Prefabricated AI Factories, Maximizing Power for Compute Through Advanced Cooling Control and Real-Time Facility Intelligence

Collaboration combines InfraPartners’ prefabricated AI data center solutions with Phaidra’s agentic technology to unlock greater productivity and speed across the design, build and operations of AI factories

HOUSTON & SEATTLE--(BUSINESS WIRE)--InfraPartners, a leader in prefabricated, fungible AI data center solutions, and Phaidra, a leading provider of AI agents for AI factories, today announced a partnership to integrate Phaidra’s AI agents into InfraPartners’ prefabricated data center solutions.

The rapid expansion of AI infrastructure is creating new challenges across the data center. AI factories are becoming larger, denser and more complex. These more elaborate facilities require power, cooling and compute to operate as a single interconnected system. Demand for the experienced engineering, construction and operations teams that can build and operate this more unified, modern data center continues to outpace supply.

By combining InfraPartners' standardized infrastructure with Phaidra's AI agents, customers gain continuous operational intelligence that analyzes infrastructure behavior and helps optimize AI factory performance. InfraPartners and Phaidra will also work together to use AI agents across design, manufacturing and commissioning stages, so AI factories operate more reliably and efficiently from day one.

Both organizations are focused on supporting the skilled labor building the world’s rapidly expanding AI infrastructure. InfraPartners uses offsite manufacturing to build 80% of the data center in its factories and 20% onsite, shifting assembly into controlled factory environments, improving quality and easing demand for scarce onsite trades, especially in remote locations where many AI factories are built and experienced workers are hardest to find.

The partnership is designed to accelerate what the industry can deliver. Demand today for electrical, mechanical and construction engineers already outstrips supply worldwide. By helping customers build and operate more data centers faster, InfraPartners and Phaidra expect the collaboration to grow industry capability across manufacturing, construction and operations while scaling at the speed AI and the world need.

Projections forecast data center demand growing to between 171 gigawatts (GW) and 219 GW by 2030, placing increasing pressure on engineering, construction and operations teams. Scaling existing ways of working is not sustainable or efficient; new methods, tools and thinking are needed to ensure AI factories are not the bottleneck preventing the world and its communities from achieving the full potential of AI — that is where the InfraPartners and Phaidra partnership is focused.

Through the partnership, Phaidra and InfraPartners are building world-class AI agents and digital infrastructure to optimize the whole data center lifecycle. Working alongside engineering and operations teams, Phaidra’s AI agents help optimize AI factories by streamlining complex infrastructure analysis and workflows, delivering more of the available power to compute, accelerating decision-making and improving operational consistency — enabling organizations to deploy and manage AI infrastructure more reliably as demand continues to grow.

One of the first areas of collaboration will integrate InfraPartners’ RapidNode™, a scalable AI factory solution, and its building management system (BMS) with two of Phaidra’s AI products: Phaidra Prism and Phaidra’s Liquid Cooling Agent.

Phaidra Prism is the operational intelligence layer for AI factories, connecting telemetry, alarms, BMS, equipment metadata and facility documentation into a unified system model that provides context for each specific facility. Prism’s large language model (LLM) interface allows operators to chat with the agent for root-cause analysis and faster time to resolution.

Phaidra’s Liquid Cooling Agent replaces reactive, temperature-based feedback loops with real-time, AI-driven, power-based, feed-forward controls. By monitoring real-time rack power data and coolant distribution unit (CDU) parameters, the self-learning agent identifies IT load spikes and predicts the resulting thermal response, meaning cooling is rightsized to the need. This proactive approach improves thermal stability, reduces cooling inefficiencies and enables operators to safely maximize the power available for revenue-generating AI workloads.

“AI factories are growing faster than any other asset class; the demand is significant and unlike anything we have seen before. The data center supply chain cannot scale quickly enough and risks becoming a massive bottleneck for the industry. InfraPartners and Phaidra are working together to solve this,” said Michalis Grigoratos, CEO of InfraPartners. “This is not about needing fewer people — the industry needs more electrical, mechanical and construction engineers than ever, and every facility we bring online creates high-value intelligence and tokens. It is about multiplying what those people can deliver and focusing on where they can have the highest impact. InfraPartners is bringing manufacturing principles to data center delivery, and Phaidra is building some of the most innovative AI agents for the design, build and operation of a data center. This is just the beginning of what we can achieve together, and I am very excited about this partnership.”

A Partnership Built on Standardization and Choice

Both companies are aligned to NVIDIA’s roadmap, and both take a vendor-agnostic approach to infrastructure. This provides customers with choice, capital flexibility and accelerated delivery times.

Phaidra will test, optimize and quality-assure its AI agents within InfraPartners’ factories, integrating its intelligence into the design, factory build and testing stages rather than retrofitting it after deployment. The result is faster testing cycles, more cost-efficient AI factory management and a further step toward software-defined building infrastructure.

“AI factories can’t be optimized as disconnected systems. Power, cooling and compute must work together as one operating system,” said Jim Gao, CEO and co-founder of Phaidra. “By embedding our AI agents into InfraPartners’ AI factory platform, we’re giving customers continuous intelligence and AI-driven control from the moment facilities come online. That’s how the industry will deploy AI infrastructure faster while getting more compute out of every watt.”

Beyond product integration, the two companies will collaborate on joint AI research and industry discussions, advancing the market’s understanding of AI-driven infrastructure optimization. InfraPartners and Phaidra are actively working with clients to design and implement the integrated solution.

About InfraPartners

InfraPartners is a global leader in digital infrastructure, delivering scalable, future-proof data center solutions. The company’s standardized design process and offsite manufacturing offer greater schedule and cost certainty, plus rigorous quality control. InfraPartners collaborates with hyperscalers, colocation providers, GPU-as-a-service (GPUaaS) providers and government entities to design and deploy AI-ready data centers. For more information, visit https://infrapartners.llc/.

About Phaidra

Phaidra is an NVIDIA DSX partner that builds AI agents for AI factories, helping operators optimize power, cooling and AI infrastructure performance at gigawatt scale. Designed to work alongside engineering and operations teams, Phaidra’s AI agents help maximize available compute capacity while improving operational efficiency, reliability and performance. Learn more at www.phaidra.ai.