OpenAI Strategy Moves to Reshore Data Centre Supply Chains

OpenAI is extending its ambitions beyond software as it launches a long-term programme to restructure the physical supply chains that underpin large-scale AI systems.

Announced on 15 January 2026, the company has launched a Request for Proposals (RFP) to reimagine its hardware supply chain. 

The programme outlines a decade-long plan to reshore advanced manufacturing for robotics, consumer hardware and data centre infrastructure. The initiative signals a shift towards tighter control of the physical components required to train and deploy increasingly complex AI models at scale.

Rather than a short-term procurement exercise, the RFP sets out a framework for building a US-based manufacturing ecosystem capable of supporting OpenAI’s expanding compute footprint and long-term infrastructure requirements.

Manufacturing as a strategic lever

The RFP frames domestic manufacturing as a prerequisite for supply chain resilience and operational certainty.

OpenAI has positioned the programme as a move away from globally fragmented sourcing towards a more integrated industrial base aligned with national priorities.

According to the RFP document, “OpenAI has a long-term ambition to establish US-based hardware manufacturing and assembly that reflects US values, supports resilient supply chains and fosters national innovation leadership”.

For data centre operators, the emphasis on local production reflects growing concerns around lead times, geopolitical exposure and the availability of critical equipment. As AI workloads drive higher density and power demand, dependencies on overseas manufacturing for electrical and mechanical systems have become a material risk.

Data centres at the core

While the RFP covers consumer devices and robotics, data centres represent the most capital-intensive and infrastructure-heavy focus area. 

OpenAI identifies power systems and thermal management as priority categories, including generators, transformers, UPS equipment, chillers and liquid cooling components.

These systems are foundational to the company’s compute strategy as it expands training and inference capacity across hyperscale facilities. By localising production of core modules and assemblies, OpenAI is aiming to reduce bottlenecks that could slow deployment timelines.

The RFP states: “Over the next 10 years, OpenAI seeks to localise significant portions of the manufacturing for its hardware devices and data centres, including key components, modules and final assembly.”

This approach aligns with a broader trend among hyperscalers and AI developers seeking closer integration between facility design and supply chains, particularly for bespoke power and cooling architectures.

Scaling power and capacity

The manufacturing initiative follows OpenAI’s progress on the Stargate Project, launched in March 2025, which includes a target to secure 10GW of power capacity. The company has already passed the halfway mark on planned capacity, underscoring the scale of infrastructure required to support frontier AI models.

OpenAI argues that control over manufacturing is inseparable from control over energy and compute delivery. The RFP notes that infrastructure decisions will shape long-term competitiveness, stating that “infrastructure has long been destiny when it comes to America’s economic success, and that will be especially true in the Intelligence Age”.

For the data centre sector, this reinforces the link between energy strategy, equipment availability and site deployment schedules.

Beyond chips and servers

The document also challenges the narrow focus on semiconductors that often dominates AI infrastructure discussions. 

OpenAI highlights the breadth of physical systems required to bring data centres online at scale.

The company notes that “advanced AI depends on a much broader ecosystem of physical components: the racks, cabling, networking gear, cooling systems, power systems, power electronics, electromechanical modules and testing and assembly capacity are all required to bring it all online at scale”.

This perspective places data centre construction, fit-out and operations at the centre of AI development rather than as a supporting function.

Implications for operators and suppliers

For equipment manufacturers and engineering firms, the RFP opens the door to long-term partnerships tied to predictable demand and standardised designs.

For data centre operators, it signals a future where AI developers exert greater influence over how facilities are built, powered and supplied.

OpenAI’s plan does not replace existing global supply chains overnight, but it establishes a clear direction of travel.

By anchoring manufacturing closer to its data centre footprint, the company is seeking to align physical infrastructure with its compute ambitions and deployment timelines.