This Week in Data Center News: 11.3.25

This week in data center news, the relentless demand for AI infrastructure is fundamentally reshaping the market, driving massive investments, key acquisitions, and unprecedented technological shifts for developers.

The necessity of advanced cooling and power took center stage: Eaton Corp's nearly $10 billion acquisition of Boyd Thermal signals a major consolidation around liquid cooling solutions. Simultaneously, the urgent need for high-density power is being met by innovative solutions like SuperX's 800VDC architecture, which supports up to 3.6 MW per rack and offers critical retrofit options for existing facilities.

Meanwhile, hyperscalers are securing capacity and future-proofing their AI strategies through massive deals and buildouts. Microsoft locked in a $9.7 billion partnership with IREN for access to chips and AI-ready capacity, while the $7 billion OpenAI/Oracle "Stargate" campus in Michigan highlights a significant investment trend into the American Midwest, emphasizing gigawatt-scale capacity and renewable energy commitment.

These developments underscore an unequivocal message for data center developers: success now hinges on integrating advanced thermal management, extreme power density, and robust sustainability from the earliest stages of site selection and design.

Eaton Corp agrees to buy Boyd Thermal for $9.5 billion to tackle growing demand for AI data centers

Power and cooling company Eaton has announced a definitive agreement to acquire Boyd Thermal, a division of Boyd Corporation, for $9.5 billion. The deal, expected to close in the second quarter of 2026, signals a significant consolidation in the data center cooling market. Boyd Thermal, with forecast sales of $1.7 billion for 2026, is a major player in liquid cooling solutions, including high-capacity Coolant Distribution Units (CDUs) capable of supporting dense AI workloads like multiple Nvidia NVL72 racks.

For data center developers, this acquisition underscores the accelerating shift towards liquid cooling as a necessity for managing high-power AI infrastructure. Eaton, a dominant force in power management, is now positioning itself to offer an integrated solution covering power and advanced liquid cooling "from the chip to the grid." This move will likely increase competition and innovation in the liquid cooling sector, potentially leading to more standardized and scalable solutions for new data center builds. Developers should monitor how this integration impacts product availability, service models, and the overall cost of implementing next-generation cooling technologies.

Crusoe Technologies and StarCloud plan to deploy the first public space data center in 2027, with xAI pledging to follow suit

Crusoe, a vertically integrated AI infrastructure provider, has partnered with Starcloud to become the first public cloud operator in space, with plans to launch GPU capacity from orbit by early 2027. The partnership addresses a critical constraint in AI infrastructure: sourcing abundant, clean energy. By deploying data centers on satellites powered by dedicated solar arrays, Crusoe eliminates dependence on terrestrial energy grids, cooling systems, and land use while co-locating compute resources directly with the sun's limitless energy output. Starcloud's first satellite, launching in late 2026, will feature a dedicated module running Crusoe Cloud and will build on their November 2025 launch of an NVIDIA H100 GPU—100x more powerful than any GPU previously deployed in space.

For data center developers, this partnership signals a transformative shift in how the industry approaches energy-intensive AI workloads. Traditional terrestrial data centers face mounting pressure from grid capacity constraints, water usage for cooling, and land availability—particularly in high-demand markets. Crusoe's space-based model bypasses these limitations entirely, offering a scalable template for future development that prioritizes energy abundance and sustainability. While orbital data centers won't replace ground-based facilities in the near term, this proof of concept could accelerate innovation in modular, energy-efficient designs and influence site selection strategies that prioritize proximity to renewable energy sources. Developers should monitor how this model performs and whether it creates competitive pressure to rethink traditional infrastructure dependencies.

Microsoft closes deal with IREN at $9.7 billion, opening access to NVIDIA chips and related AI infrastructure

Microsoft has signed a five-year, $9.7 billion deal with data center operator IREN to gain access to NVIDIA's advanced chips and infrastructure for its AI operations. This strategic move allows Microsoft to rapidly expand its AI computing power without immediately building new data centers, addressing one of the main constraints in the growing demand for AI services. The partnership underscores a significant trend where major tech companies are leasing capacity from specialized data center operators to quickly scale their AI capabilities.

For data center developers, this deal highlights the immense and growing demand for AI-ready infrastructure. The collaboration between Microsoft and IREN, particularly IREN's use of renewable energy and its significant 2,910-megawatt capacity across North America, signals key priorities for hyperscalers: speed, massive power availability, and sustainability. As AI workloads intensify, developers should focus on building or retrofitting facilities that can support high-density power requirements and offer sustainable energy solutions to attract major tenants like Microsoft. This leasing model could become a dominant strategy for tech giants, creating major opportunities for developers who can meet these specific infrastructural demands.

SuperX launches 8000VDC power solutions for data centers, deploying end-to-end native architecture

SuperX AI Technology Limited has launched two innovative 800VDC power solutions addressing critical infrastructure challenges for data center developers. The Panama-800VDC targets new-build AI hyperscale facilities with an end-to-end native architecture achieving up to 98.5% efficiency and supporting rack densities exceeding 3.6 MW. The Aurora-800VDC provides a modular retrofit solution enabling existing facilities to upgrade without disrupting operations, completing single-rack upgrades in hours while scaling from 200 kW to over 1 MW. Both systems reduce power conversion stages from 4-5 down to 1-2, eliminate transient current peaks through intelligent load management, and cut copper usage by over 45% compared to traditional AC systems—translating to approximately 500,000 tons of copper savings for a 1 GW data center.

For data center developers, these solutions directly address the power density bottleneck created by next-generation GPU systems like NVIDIA's GB300 NVL72, which demand 250kW+ per rack. The technology's compatibility with NVIDIA Kyber rack architecture and integration of compute, cooling, and power capabilities provides a full-stack infrastructure approach. The Aurora retrofit option is particularly valuable for developers with existing assets seeking to monetize facilities for AI workloads without extensive capital expenditure or downtime. Combined with the 98.5% efficiency rating, these systems offer improved operational economics through reduced energy costs while supporting the higher power densities essential for competitive AI data center positioning in today's market.

OpenAI and Oracle plan 1GW data center in Michigan with Related Digital under StarGate venture

OpenAI, in partnership with Oracle and Related Digital, has announced a $7 billion hyperscale data center campus in Saline Township, Michigan. This project, part of OpenAI's larger "Stargate" initiative, is set to break ground in early 2026. The 2.2-million-square-foot facility will consolidate multiple data centers into a single campus, delivering over a gigawatt of capacity to power the next generation of AI. For data center developers, this move signals a significant investment trend into the American Midwest, leveraging state tax incentives and large land parcels to support massive AI infrastructure buildouts.

A key aspect for developers to note is the project's approach to power and sustainability. The campus will draw an estimated 1.4 gigawatts from DTE Energy, supported by a developer-funded battery storage facility and grid modernization investments, which DTE claims will prevent rate hikes for existing customers. The design aims for LEED certification with a closed-loop cooling system to minimize water usage. Furthermore, Michigan's clean energy laws require the data center to source 90% renewable power to qualify for tax credits, setting a precedent for future large-scale developments in the region.

However, the project was not without its challenges, facing initial resistance from local residents concerned about noise, energy consumption, and the loss of rural character. The plan ultimately moved forward after a zoning dispute was settled in court, with developers committing $14 million to local investments and the preservation of 700 acres of farmland, wetlands, and forest. This highlights the growing importance for developers to proactively address community and environmental concerns as part of the site selection and approval process for new data center projects.

Tools & Solutions for Data Center Developers

Discover how we address critical challenges like power availability and project siting, and explore our range of available solutions. Book a demo with our dedicated team.LandGate provides tailored solutions for data center developers. 

You can also visit our library of data center resources.