Ampere Insights

May 14, 2025

The U.S. Department of the Interior’s expedited permitting of the Velvet-Wood uranium and vanadium mine in Utah, announced on May 12, 2025, signals a bold response to the national energy emergency declared by President Donald J. Trump. Coupled with discussions around thorium-based nuclear reactors and the potential revival of coal plants, this move aims to secure energy for AI data centers, strengthen the grid, and enhance national security. The Velvet-Wood project, as reported by ZeroHedge and the DOI, underscores America’s push for energy independence amid surging AI-driven power demands. This article explores the Velvet-Wood initiative, its implications for AI data centers, the role of thorium and coal, and the broader impact on the U.S. energy landscape. Read the DOI press release.

Velvet-Wood Mine: Fueling Nuclear Power

The Velvet-Wood mine in San Juan County, Utah, is set to produce uranium and vanadium, critical for nuclear reactors and advanced alloys, respectively. The DOI’s emergency permitting, slashing environmental reviews to 14 days under the National Environmental Policy Act and other regulations, addresses America’s “dangerous reliance” on imported uranium, with 99% of U.S. nuclear fuel sourced abroad in 2023, per ZeroHedge. The mine, operated by Anfield Energy, will leverage existing Velvet Mine infrastructure, minimizing surface disturbance to three acres. If approved, it will feed the nearby Shootaring Canyon uranium mill, reducing import dependence and supporting AI data centers’ energy needs. Explore ZeroHedge’s coverage.

[](https://www.zerohedge.com/commodities/us-fast-tracking-uranium-mine-permit-meet-urgent-energy-demands)[](https://www.doi.gov/pressreleases/interior-expedites-permitting-critical-energy-project-address-national-energy)

AI Data Centers and Uranium Demand

AI data centers, projected to consume 9% of U.S. electricity by 2030, require reliable, carbon-free power, making nuclear energy a prime solution. Projects like OpenAI’s 200 MW Stargate facility and Google’s 1 MW rack designs demand robust power infrastructure, with uranium as the backbone for nuclear reactors. Velvet-Wood’s output will support facilities like Amazon’s nuclear-powered Pennsylvania campus, ensuring domestic fuel for AI-driven hyperscale growth. Posts on X highlight the urgency, with Anfield Energy’s stock soaring 42% after the permitting announcement, reflecting market confidence in uranium’s role. Learn about Stargate’s scale.

[](https://x.com/quakes99/status/1922022162333868441)

Thorium: A Potential Nuclear Alternative

While uranium dominates nuclear energy, thorium-based reactors are gaining attention as a safer, more sustainable alternative. A ZeroHedge report notes China’s 2025 unveiling of a meltdown-proof thorium reactor, leveraging U.S. research from the 1950s. Thorium offers higher fuel burn-up rates (55,000 MWd/T vs. uranium’s 7,000 MWd/T) and reduces plutonium waste by over 80%, making it ideal for AI data centers’ long-term needs. The U.S., through partnerships like Clean Core Thorium Energy with Texas A&M and Idaho National Laboratory, is developing thorium-based ANEEL fuel for existing reactors. Velvet-Wood’s uranium focus could complement thorium initiatives, diversifying nuclear options for energy-intensive AI applications. Read about thorium reactors.

[](https://www.zerohedge.com/energy/china-unveils-worlds-1st-meltdown-proof-thorium-reactor)

Coal as an Interim Solution

To meet immediate energy demands, the Trump administration is exploring coal plant reopenings and new builds, as outlined in a ZeroHedge article. An April 2025 executive order aims to reverse coal leasing moratoriums, designating coal a “critical mineral” to boost its strategic role. Coal, supplying 15% of U.S. electricity in 2023, offers a reliable interim solution for AI data centers while nuclear capacity ramps up. The administration’s rollback of mercury and carbon regulations, combined with plans to reopen shuttered plants, could bridge the gap, though critics warn of environmental trade-offs. This strategy aligns with Interior Secretary Doug Burgum’s emphasis on “affordable and reliable” energy for data centers. Explore coal’s revival.

[](https://www.zerohedge.com/markets/trump-set-sign-exec-order-boosting-domestic-mining-and-coal)

Impact on the U.S. Energy Grid

The U.S. grid faces strain from AI data centers and electrification, with the North American Electric Reliability Corporation warning of shortage risks by 2035. Velvet-Wood’s uranium will bolster nuclear capacity, while thorium and coal could diversify supply. The DOI’s FAST-41 program, streamlining permits for critical minerals, and the Department of Energy’s site assessments for AI data centers on federal land enhance grid resilience. However, coal’s resurgence raises sustainability concerns, as noted by the Sierra Club, which criticizes rushed reviews for risking environmental harm. Balancing these energy sources will be critical to avoid blackouts and support AI growth. Learn about FAST-41.

[](https://www.doi.gov/pressreleases/trump-administration-adds-key-mining-projects-fast-41)[](https://www.zerohedge.com/energy/us-coal-plants-shutter-renewed-focus-nuclear)

National Security and Economic Benefits

The Velvet-Wood project addresses national security risks from foreign mineral dependence, as highlighted in Trump’s energy emergency declaration. China and Russia dominate uranium and vanadium markets, posing threats to U.S. defense and AI industries. Domestic production from Velvet-Wood, potentially paired with thorium development, strengthens supply chains for nuclear reactors and aerospace alloys. Economically, the project will create jobs and reduce import costs, with the DOE estimating billions in savings from secure mineral chains. Coal plant reopenings could further stimulate mining communities, though long-term reliance may face market and regulatory challenges. Read about mineral security.

[](https://www.zerohedge.com/commodities/how-ukraine-minerals-deal-supports-us-strategy-and-ukrainian-security)

Challenges and Sustainability Concerns

Expedited permitting for Velvet-Wood and coal initiatives risks environmental oversight, with the Sierra Club warning that 14-day reviews may miss hazards, potentially polluting air and water. Thorium reactors, while promising, face regulatory and fuel supply hurdles, with the DOE’s disposal of uranium-233 stocks drawing criticism from advocates like the Thorium Energy Alliance. Coal’s revival, though interim, could delay renewable transitions, conflicting with net-zero goals of tech giants like Google. Innovations like 3D-printed transformers and immersion cooling could mitigate some impacts by improving efficiency. Explore thorium challenges.

[](https://www.zerohedge.com/energy/nuclear-energy-world-awaits-trump)

Looking Ahead

The Velvet-Wood mine, alongside thorium exploration and coal’s interim role, positions the U.S. to meet AI data center demands while advancing energy independence. By 2030, domestic uranium and vanadium could power nuclear-driven AI facilities, with thorium offering a sustainable long-term option. Coal’s resurgence may provide short-term relief but risks environmental backlash. As projects like Stargate and Google’s Arizona campus scale, the U.S. must balance speed, sustainability, and security to lead in AI and energy. The Velvet-Wood project sets a precedent for rapid, responsible energy development, shaping a resilient future. Discover Google’s Arizona project.

— Reported based on ZeroHedge, U.S. Department of the Interior, and industry insights, May 2025

May 14, 2025

The AI revolution is driving transformative advancements in data center infrastructure, with projects like OpenAI’s Stargate and cutting-edge cooling solutions from Intel and Shell leading the charge. The Stargate project’s Phase 1 construction in Abilene, Texas, sets a new benchmark for hyperscale AI facilities, while Intel’s certification of Shell’s immersion cooling fluids marks an industry-first for energy-efficient data center operations. These developments address the escalating power and thermal demands of AI workloads, reshaping the future of computing. This article explores the Stargate project, Intel-Shell’s cooling innovation, and their broader implications for AI data centers. Read about Stargate’s Phase 1.

OpenAI’s Stargate Phase 1: A Hyperscale Milestone

The Stargate project, a $500 billion joint venture between OpenAI, Oracle, SoftBank, and MGX, is constructing its flagship data center in Abilene, Texas, with Phase 1 underway. This phase includes two buildings totaling 980,000 square feet and over 200 MW of power capacity, set to be energized in the first half of 2025. The facility will support GPU installations for OpenAI’s AI models, with plans to scale to 1 million GPUs by late 2025 or Q1 2026. A 360.5 MW natural gas plant and solar power with battery storage will ensure reliable energy, addressing grid constraints. Stargate’s expansion plans include additional U.S. sites and global locations, aiming to establish a network of sovereign AI infrastructure. Despite tariff-related delays slowing financing, the Abilene site’s progress signals robust momentum. Learn about Stargate’s power plans.

[](https://www.nextbigfuture.com/2025/05/openai-stargate-phase-1-construction-of-200-megawatts-and-980000-square-feet.html)[](https://www.datacenterdynamics.com/en/news/natural-gas-plant-planned-for-stargate-ai-data-center-campus-report/)

Intel and Shell’s Immersion Cooling Breakthrough

Intel, in collaboration with Shell, has achieved an industry-first by certifying Shell’s lubricant-based immersion cooling fluids for Xeon-based data centers, as announced on May 13, 2025. After a two-year trial, Intel validated Shell’s single-phase immersion cooling technology, which submerges servers in non-conductive fluid to dissipate heat more efficiently than air cooling. This solution, optimized for Intel’s Xeon processors, reduces energy consumption by up to 30% and supports high-density AI workloads. The certification, backed by partners like Supermicro and Submer, sets a new standard for sustainable data center cooling, critical for facilities like Stargate handling megawatt-scale racks. Read Intel’s announcement.

[](https://x.com/Techmeme/status/1922291748513931313)

Shell’s Cooling Fluids: A Global Standard

Shell’s immersion cooling fluids, certified by Intel, are now approved for global use in data centers, marking a significant milestone in thermal management. These fluids, developed through Shell’s lubricants division, offer superior heat transfer and energy efficiency, enabling denser server configurations without compromising performance. Posts on X highlight the technology’s potential, projecting that liquid cooling, including immersion, will account for 36% of the data center cooling market by 2028. Shell’s solution aligns with the industry’s shift toward sustainability, reducing carbon footprints and operational costs for AI-driven facilities. This innovation is particularly relevant for hyperscale projects like Stargate, where cooling demands are escalating. Explore Shell’s cooling fluids.

[](https://x.com/infra_digital/status/1922600129472651762)

Implications for AI Data Centers

The convergence of Stargate’s hyperscale ambitions and Intel-Shell’s cooling advancements addresses critical challenges in AI data centers: power and thermal management. AI workloads, with rack densities projected to reach 1–5 MW by 2030, demand robust infrastructure. Stargate’s 200 MW Phase 1 and planned 1.2 GW campus demonstrate the scale required, while immersion cooling enables denser, more efficient operations. Google’s 1 MW rack designs and Amazon’s nuclear investments further underscore the need for such innovations. These developments reduce reliance on strained grids, with Stargate’s natural gas and solar mix complementing Intel-Shell’s energy-saving cooling. Learn about Google’s rack plans.

Challenges and Opportunities

Despite their promise, these advancements face hurdles. Stargate’s financing is slowed by tariff uncertainties, as reported by Bloomberg, potentially delaying global expansion. Immersion cooling, while efficient, requires significant upfront investment and retrofitting for existing data centers, as noted in Data Centre Magazine. Scalability and material costs for cooling fluids also pose challenges. However, opportunities abound: Stargate’s Abilene site could create thousands of jobs, boosting local economies, while Intel-Shell’s technology positions them as leaders in sustainable cooling. The U.S. Department of Energy’s push for AI data centers on federal land could further accelerate such projects, leveraging domestic energy and minerals. Read about data center challenges.

[](https://www.bloomberg.com/news/articles/2025-05-12/softbank-stargate-venture-with-openai-hits-snags-on-tariff-fears)

Looking Ahead

The Stargate project and Intel-Shell’s immersion cooling breakthrough signal a transformative era for AI data centers. By mid-2025, Stargate’s Phase 1 will power advanced AI models, while Shell’s certified fluids will enable global adoption of energy-efficient cooling. These innovations align with industry trends toward sustainability and hyperscale growth, as seen in projects like xAI’s 1 GW Memphis data center. As AI demand surges, overcoming tariff, scalability, and retrofitting challenges will be critical. The U.S. is poised to lead in AI infrastructure, with Stargate and immersion cooling paving the way for a resilient, sustainable future. Explore grid bypass trends.

— Reported based on NextBigFuture, Intel, Shell, and industry insights, May 2025

May 14, 2025

The U.S. Department of the Interior’s expedited permitting of the Velvet-Wood mine in Utah, announced on May 11, 2025, marks a pivotal step in addressing the national energy emergency declared by President Donald J. Trump. By accelerating environmental reviews to 14 days under new emergency procedures, the Bureau of Land Management (BLM) aims to unlock critical minerals vital for power systems and AI data centers. This move, part of a broader push for energy dominance, has significant implications for the U.S. energy grid, national security, and the booming AI sector. This article explores the Velvet-Wood project, its role in supporting AI data centers, and its broader impact on the U.S. energy landscape. Read the DOI press release.

The Velvet-Wood Mine and Critical Minerals

The Velvet-Wood mine is poised to extract critical minerals like lithium, cobalt, and rare earth elements, essential for advanced power systems, batteries, and electronics. The Department of the Interior’s emergency permitting, authorized under the National Environmental Policy Act, Endangered Species Act, and National Historic Preservation Act, reduces a typically years-long process to just 14 days. This aligns with President Trump’s January 20, 2025, National Energy Emergency declaration, which prioritizes domestic energy and mineral production to counter reliance on foreign adversaries. The BLM’s oversight ensures responsible extraction, supporting technologies like transformers and grid infrastructure critical for AI data centers. Learn about BLM’s role.

Implications for AI Data Centers

AI data centers, with rack densities projected to reach 1–5 MW by 2030, are driving unprecedented energy demand, potentially consuming 9% of U.S. electricity by 2030. Critical minerals from Velvet-Wood are vital for manufacturing high-efficiency transformers and energy storage systems to support these facilities. For instance, Google’s 1 MW rack designs and Amazon’s nuclear-powered data centers rely on advanced power components that use rare earths and lithium. By securing a domestic mineral supply, the U.S. can reduce supply chain vulnerabilities, ensuring AI infrastructure scales without reliance on geopolitical competitors like China, which dominates global mineral markets. Explore Google’s 1 MW rack plans.

Strengthening the U.S. Energy Grid

The U.S. energy grid faces strain from AI data centers and electrification trends, with power infrastructure bottlenecks delaying projects. Velvet-Wood’s minerals will support the production of transformers and grid-scale batteries, enhancing grid reliability. The Department of Energy’s assessment of 16 DOE-managed sites for AI data center co-location, as noted in posts on X, underscores the need for robust power systems. Fast-tracking mineral extraction aligns with the Trump administration’s energy dominance agenda, reducing permitting delays that can stretch 7–10 years for mines, compared to 2–5 years in Australia and Canada. This ensures the grid can handle AI-driven loads while supporting renewable and nuclear integration. Read about AI data center power demands.

National Security and Economic Benefits

The Velvet-Wood project addresses national security concerns outlined in the National Energy Emergency declaration, which cites reliance on foreign minerals as a threat to economic stability and defense readiness. The U.S. imports many critical minerals despite abundant domestic reserves, leaving industries like AI and defense vulnerable to supply chain disruptions. By streamlining permitting, the Interior Department aims to bolster domestic production, creating jobs and reducing costs for manufacturers. A Deloitte report estimates that secure mineral supply chains could save U.S. industries billions annually, with Velvet-Wood contributing to this economic resilience. Learn about critical mineral strategies.

[](https://www.doi.gov/pressreleases/trump-administration-adds-key-mining-projects-fast-41)

Challenges and Sustainability Concerns

Expedited permitting raises concerns about environmental and community impacts. Critics argue that compressing reviews to 14 days risks overlooking ecological or cultural consequences, as seen in debates over the Biden-era Western Solar Plan. The BLM’s commitment to responsible extraction, backed by emergency procedures, aims to balance speed with stewardship, but stakeholder engagement remains critical. Additionally, scaling mineral processing to meet AI data center demands requires investment in refining infrastructure, as raw minerals alone cannot address supply chain gaps. Innovations like 3D-printed transformers, which reduce material waste, could complement Velvet-Wood’s output. Explore sustainable land management.

[](https://www.doi.gov/pressreleases/interior-department-finalizes-framework-future-solar-development-public-lands)

Industry-Wide Context and Trends

The Velvet-Wood project aligns with industry trends toward domestic energy and mineral independence. Amazon’s $500 million investment in small modular reactors (SMRs) and Microsoft’s Three Mile Island restart reflect the need for reliable power for AI data centers, which Velvet-Wood’s minerals will support through advanced power components. The Department of the Interior’s FAST-41 program, which streamlines permitting for critical mineral projects, further accelerates this trend, reducing delays that deter investment. These efforts position the U.S. to compete with nations like Canada, where faster permitting attracts global capital. Read about Amazon’s SMR investment.

[](https://www.doi.gov/pressreleases/trump-administration-adds-key-mining-projects-fast-41)

Looking Ahead

The expedited permitting of Velvet-Wood sets a precedent for rapid energy and mineral development, critical for powering AI data centers and strengthening the U.S. grid. By 2030, domestic mineral production could reduce reliance on foreign supplies, enhancing national security and economic stability. However, balancing speed with sustainability will be key to maintaining public trust. As projects like Google’s Arizona data center and the DOE’s site assessments advance, Velvet-Wood’s minerals will play a pivotal role in building a resilient, AI-ready energy infrastructure. The U.S. is poised to lead in AI and energy dominance, provided it navigates these challenges effectively. Discover Google’s Arizona project.

— Reported based on U.S. Department of the Interior, Data Centre Magazine, and industry insights, May 2025

May 13, 2025

Additive manufacturing, or 3D printing, is transforming the production of transformer components, offering unprecedented efficiency, reduced material waste, and design flexibility. As outlined in a ScienceDirect study, 3D printing enables the creation of complex electromechanical devices like transformers with enhanced performance and sustainability. This technology is poised to redefine power system fabrication, but challenges such as material durability and scalability must be addressed. This article explores how 3D printing is revolutionizing transformer manufacturing, its benefits, hurdles, and potential to reshape supply chains. Read the ScienceDirect study on 3D printing transformers.

3D Printing in Transformer Fabrication

Traditional transformer manufacturing relies on subtractive methods, which generate significant material waste and limit design complexity. 3D printing, by contrast, builds components layer by layer, enabling intricate geometries that optimize performance. For instance, additive manufacturing allows for the production of transformer cores with reduced energy losses through precise material deposition. A CANWIN report highlights how 3D printing minimizes waste and energy use, producing lighter, more efficient components. This is critical for transformers, which are essential for power distribution in data centres, renewable energy systems, and urban grids. Explore CANWIN’s sustainable practices.

[](https://www.canwindg.com/a-news-sustainable-practices-in-transformer-manufacturing-environmental-impact-and-solutions)

Improved Efficiency and Reduced Waste

3D printing enhances transformer efficiency by enabling designs that reduce core losses and improve thermal management. For example, high-performance polymers like polyetheretherketone (PEEK) and polyphenylene sulfone (PPSU), tested in a 2024 study, show excellent dielectric strength and compatibility with insulating oils, making them ideal for transformer components. These materials, printed via fused filament fabrication (FFF), allow for customized parts that enhance electrical performance. Additionally, additive manufacturing cuts material waste by up to 90% compared to subtractive methods, as noted in a Wikipedia entry on 3D printing, aligning with sustainability goals in power system production. Read about 3D-printed polymers for transformers.

[](https://link.springer.com/article/10.1007/s40964-024-00614-z)[](https://en.wikipedia.org/wiki/3D_printing)

Material Durability Challenges

While 3D printing offers design advantages, material durability remains a hurdle. Transformer components must withstand harsh conditions, including high voltages, extreme temperatures, and exposure to insulating oils. The 2024 study found that while PEEK and PPSU perform well, materials like Biofila crack under mineral oil exposure, and others like polyvinylidene fluoride (PVDF) lack sufficient dielectric strength. Developing durable, high-performance materials for 3D printing is critical, as is ensuring their long-term stability in operational transformers. Advances in material science, such as bio-based polymers and composites, are being explored to address these issues. Learn about material challenges in AM.

[](https://link.springer.com/article/10.1007/s40964-024-00614-z)

Scalability and Production Hurdles

Scalability is another challenge for 3D-printed transformers. Current additive manufacturing systems excel at prototyping and low-volume production but struggle with the high-throughput demands of industrial transformer manufacturing. A Deloitte report notes that 3D printing’s speed and material limitations hinder mass production, requiring significant investment in larger, faster printers. Additionally, the high energy consumption of processes like laser-based printing, as highlighted in a ScienceDirect review, raises concerns about cost and environmental impact. Overcoming these barriers will require innovations in print head technology and multi-material printing systems. Discover AM supply chain impacts.

[](https://www2.deloitte.com/us/en/insights/focus/3d-opportunity/additive-manufacturing-3d-printing-supply-chain-transformation.html)[](https://www.sciencedirect.com/science/article/pii/S2589234725000636)

Reshaping Supply Chains

3D printing has the potential to revolutionize transformer supply chains by enabling on-demand, localized production. A Maine Pointe analysis suggests that additive manufacturing can decentralize production, reducing reliance on complex global supply chains and cutting lead times for critical components. For example, utilities could print replacement parts on-site, minimizing downtime and inventory costs. This is particularly valuable for data centres, where transformer failures can disrupt AI operations. However, scaling this model requires digital integration, such as AI-driven design optimization, and robust quality control to ensure component reliability. Read about 3D printing’s supply chain impact.

[](https://www.mainepointe.com/blog/five-ways-3d-printing-will-impact-the-global-supply-chain)

Industry-Wide Adoption and Trends

The adoption of 3D printing for transformers is part of a broader trend in power system innovation. Companies like Siemens are exploring additive manufacturing for gas turbine components, which share similar material and durability requirements, as noted in an Additive Manufacturing Media report. In nuclear energy, MakerVerse highlights how 3D printing is used for reactor components, suggesting parallels for transformer applications. These cross-industry advancements, coupled with regulatory support for sustainable manufacturing, are accelerating 3D printing’s integration into power systems. Explore AM in power generation.

[](https://www.additivemanufacturing.media/articles/in-power-generation-the-changes-from-3d-printing-will-be-system-wide)[](https://www.makerverse.com/resources/powering-the-future-how-3d-printing-is-reshaping-nuclear-innovation/)

Looking Ahead

The future of 3D-printed transformers is promising, with the potential to enhance efficiency, sustainability, and supply chain resilience. By 2030, advancements in durable materials and scalable printing systems could enable mass production of transformer components, reducing costs and environmental impact. Projects like Google’s 1 MW rack designs, which demand efficient power systems, underscore the need for innovative transformers. As the industry overcomes durability and scalability challenges, 3D printing could become a cornerstone of power system fabrication, powering the AI and renewable energy revolutions. Learn about Google’s 1 MW racks.

— Reported based on ScienceDirect, CANWIN, and industry insights, May 2025

May 12, 2025

Google is pushing the boundaries of data centre design with plans for 1-megawatt (MW) IT racks, leveraging the electric vehicle (EV) supply chain to revolutionize power delivery and cooling infrastructure, as detailed in a May 2025 report by The Register. This ambitious move, unveiled at the Open Compute Project (OCP) EMEA summit, addresses the soaring energy demands of AI workloads. By adopting ±400 VDC power systems and advanced cooling distribution units (CDUs), Google aims to support hyperscale AI processing while navigating significant engineering challenges. This article explores Google’s strategy, the implications of 1 MW racks, and complementary industry developments. Read the full Register article.

Google’s 1 MW Rack Vision

At OCP EMEA 2025, Google outlined its plan to design data centre racks capable of supporting 1 MW of IT hardware, a dramatic leap from the 30–100 kW racks common in 2025. This shift is driven by AI’s energy-intensive workloads, particularly for training large language models and generative AI applications. By exploiting the EV supply chain, Google is adopting ±400 VDC power delivery systems, which offer higher efficiency and compatibility with mass-produced EV components. This approach reduces costs and accelerates deployment, but it requires rethinking power infrastructure to handle unprecedented energy densities. Explore data centre power trends.

[](https://www.theregister.com/2025/05/01/google_details_plans_for_1/)

Rack Density and Future Power Requirements

The move to 1 MW racks reflects a broader industry trend toward extreme rack density, with projections of 1–5 MW per rack in the coming years. Posts on X note that rack power has surged from 10–20 kW in the 2010s to 30–100 kW in 2025, with Google leading the charge toward megawatt-scale designs. Such densities demands advanced cooling solutions, as traditional air cooling is insufficient. Google’s next-generation CDUs, designed for liquid cooling, aim to manage the heat generated by 1 MW racks, potentially requiring megawatts of cooling capacity per rack. This aligns with industry forecasts that AI data centres could consume as much power as small cities by 2030. Learn about AI data centre power demands.

Cooling Infrastructure Challenges

Cooling 1 MW racks poses significant engineering hurdles. Liquid cooling, already critical for 100 kW racks, becomes non-negotiable at megawatt scales. Google’s CDUs are designed to handle high thermal loads, but scaling to 1 MW requires innovations in coolant flow, heat exchanger efficiency, and system reliability. Industry analyses suggest that direct-to-chip and immersion cooling will dominate, with companies like Nvidia and AMD integrating liquid cooling into GPU designs. The risk of thermal runaway or cooling failures at such power levels also necessitates advanced fire suppression systems, moving beyond oxygen-deprivation methods to address high-energy electrical risks. Discover liquid cooling advancements.

[](https://forums.theregister.com/forum/all/2025/05/01/google_details_plans_for_1/)

Power Delivery and EV Supply Chain Integration

Google’s adoption of ±400 VDC power draws inspiration from EV battery systems, which operate at similar voltages. This allows Google to tap into mass-produced components like inverters and power modules, reducing costs and lead times. However, delivering 1 MW per rack requires high-voltage cabling, advanced power distribution units, and grid-scale energy sources. Google’s concurrent investments in nuclear power, including a 1.8 GW deal with Elementl Power for three advanced reactor sites, aim to secure the reliable, carbon-free energy needed for these racks. This strategy mitigates grid constraints, as data centres are projected to consume 9% of U.S. electricity by 2030. Read about Google’s nuclear investments.

[](https://www.theregister.com/2025/05/07/google_signs_another_nuclear_deal)

Industry-Wide Implications

Google’s 1 MW rack plans set a precedent for the industry, with competitors like Amazon and Microsoft also scaling infrastructure for AI. Amazon’s $100 billion 2025 investment in AI data centres and Microsoft’s Three Mile Island restart reflect similar ambitions. However, the shift to megawatt racks raises safety and regulatory concerns. Forums on The Register highlight parallels to the 19th-century railway boom, warning of “explosive” risks if engineering standards lag. The industry must balance innovation with robust safety protocols to prevent incidents at high-energy facilities. Explore Microsoft’s nuclear deal.

[](https://forums.theregister.com/forum/all/2025/05/01/google_details_plans_for_1/)

Challenges and Risks

Implementing 1 MW racks involves significant challenges. Power infrastructure must handle extreme current loads, requiring costly upgrades to transformers and cabling. Cooling systems face scalability limits, with potential bottlenecks in coolant supply or heat dissipation. Regulatory hurdles, including fire safety and grid interconnection standards, could delay deployments. Additionally, the speculative nature of hyperscale buildouts risks overcapacity if AI demand doesn’t match projections, as noted by Moody’s Ratings. Strategic planning and modular designs will be critical to mitigate these risks. Learn about hyperscale risks.

Looking Ahead

Google’s 1 MW rack plans signal a transformative shift in data centre design, driven by AI’s relentless growth. By leveraging EV supply chains, nuclear power, and advanced cooling, Google aims to sustain its leadership in AI infrastructure. As rack densities approaches 1–5 MW, the industry must innovate rapidly to address power, cooling,marker cooling, and safety challenges. Projects like Google’s Mesa, Arizona, data centre and nuclear partnerships will shape the future of sustainable AI computing. The success of these megawatt-scale racks could redefine how data centres power the AI revolution. Discover Google’s Arizona data centre.

— Reported based on The Register, Data Centre Magazine, and industry insights, May 2025

May 12, 2025

The global AI boom is driving an unprecedented surge in hyperscale data centre development, with Moody’s Ratings forecasting a 20% annual capacity increase after 2028. However, this rapid expansion comes with significant technical, financial, and sustainability challenges, as outlined in Data Centre Magazine’s report, “The AI Data Centre Potential: The Urgent Need for Balance.” From escalating rack densities to geopolitical trade tensions, the industry must navigate a complex landscape to sustain growth. This article explores the opportunities, risks, and future power demands shaping AI data centres. Read the full Data Centre Magazine report.

[](https://datacentremagazine.com/hyperscale/the-ai-data-centre-potential-the-urgent-need-for-balance)

AI-Driven Hyperscale Expansion

The rise of AI, particularly generative AI, is transforming data centre infrastructure. Hyperscalers like Meta, OpenAI, and others are building massive campuses, such as the 5.6GW Wonder Valley project in Alberta and Meta’s 2GW Louisiana site, to centralize gigawatts of computing power. These facilities, designed to handle AI’s intensive workloads, are projected to add multiple gigawatts by 2029. However, Moody’s warns that much of this capacity is speculative, built in anticipation of future needs, risking overbuild if demand doesn’t align. Explore recent hyperscale projects.

[](https://datacentremagazine.com/hyperscale/the-ai-data-centre-potential-the-urgent-need-for-balance)

Rack Density and Future Power Requirements

AI data centres are pushing rack density to new heights, with current designs reaching 130 kilowatts per rack and projections of 1–5 megawatts per rack in the coming years. Posts on X highlight this trend, noting that average rack density has surged from 10–20 kW in the 2010s to 30–100 kW in 2025, with Google designing for 1MW racks. These extreme power demands require advanced cooling systems, such as liquid cooling, and robust power infrastructure, potentially consuming as much electricity as a medium-sized city. Meeting 1–5MW per rack will necessitate innovative energy solutions, including nuclear power and energy storage systems, to avoid grid strain. Learn about power constraints.

[](https://x.com/storagereview/status/1919455463348777432)[](https://x.com/GWTJAL/status/1921234053220512055)

Supply Chain and Trade Challenges

Geopolitical tensions, particularly U.S. tariffs on materials like structural steel and rare earth minerals, are disrupting data centre supply chains. Moody’s notes that these tariffs increase costs and delay projects, as developers reassess budgets and timelines. Rising prices for critical electrical components further complicate bringing new capacity online, potentially squeezing profit margins for hyperscalers. The report emphasizes that abrupt policy changes create uncertainty, making cost forecasting difficult. Read about supply chain impacts.

[](https://datacentremagazine.com/hyperscale/the-ai-data-centre-potential-the-urgent-need-for-balance)

Sustainability and Construction Risks

AI data centres, often built in remote locations for access to reliable, low-cost power, face unique construction challenges. Maintaining a skilled workforce and managing high transport costs in these areas can delay projects. Additionally, the energy-intensive nature of AI workloads raises sustainability concerns. Moody’s suggests that hyperscalers must prioritize cleaner energy sources, such as nuclear or renewables, to align with net-zero goals. Innovations like liquid cooling and modular construction are critical to improving efficiency. Discover nuclear power solutions.

[](https://datacentremagazine.com/hyperscale/the-ai-data-centre-potential-the-urgent-need-for-balance)

Financial and Strategic Risks

The speculative nature of hyperscale investments poses long-term credit risks for developers and investors. Moody’s highlights that while cash-rich hyperscalers like Google and Amazon can absorb financial pressures, smaller operators may struggle. The report warns of potential capital reallocation as hyperscalers “right-size” capacity to match actual demand. Strategic planning, including site selection and power procurement, will be crucial to avoid overbuild and ensure profitability. Explore strategic data centre planning.

[](https://datacentremagazine.com/hyperscale/the-ai-data-centre-potential-the-urgent-need-for-balance)

Looking Ahead

The AI data centre boom presents immense opportunities but demands careful balance. With capacity projected to grow 20% annually post-2028, hyperscalers must address rack density challenges, secure sustainable power, and navigate supply chain volatility. Projects like Wonder Valley and Louisiana’s 2GW campus signal confidence in AI’s future, but the industry must avoid overbuilding by aligning infrastructure with real-world needs. As AI continues to reshape technology, data centres will remain the backbone of this transformation, provided they adapt to evolving demands. Read about grid bypass trends.

[](https://datacentremagazine.com/hyperscale/the-ai-data-centre-potential-the-urgent-need-for-balance)

— Reported based on Data Centre Magazine, Moody’s Ratings, and industry insights, May 2025

May 12, 2025

The United States is witnessing an unprecedented surge in data center construction, fueled by the escalating demand for artificial intelligence (AI) and cloud computing infrastructure. While DataCenterKnowledge’s May 2025 roundup highlighted significant projects like Meta’s $1 billion Wisconsin investment and NTT’s Arizona expansion, numerous other developments across the country are shaping the industry’s future. This article explores new U.S. data center projects not covered in the aforementioned report, emphasizing their scale, innovation, and impact on the AI-driven digital economy. Read the DataCenterKnowledge May 2025 roundup.

Homer City’s Massive 4.5 GW Campus

In Pennsylvania, the Homer City Generating Station, once the state’s largest coal-burning power plant, is being transformed into a 4.5-gigawatt data center campus tailored for AI and high-performance computing. Homer City Redevelopment and Kiewit Power Constructors signed a $10 billion deal to develop this 3,200-acre facility, which will leverage natural gas to ensure reliable power. The project, expected to break ground in 2025, aims to capitalize on the site’s existing infrastructure to meet the energy demands of hyperscale AI workloads. Learn about the Homer City project.

Amazon’s $2 Billion Ohio Expansion

Amazon is advancing its cloud computing footprint with a $2 billion data center campus in Sunbury, Ohio, part of a broader $10 billion investment in the state. The 450,000-square-foot facility, set to occupy 200 acres in an industrial park, is scheduled to begin construction in January 2028, with completion targeted for December 2034. This project complements Amazon’s earlier $7.8 billion Ohio investments, reinforcing central Ohio’s status as a data center hub. Explore Amazon’s Ohio plans.

Google’s $600 Million Arizona Data Center

Google broke ground in July 2023 on a 750,000-square-foot data center in Mesa, Arizona, with the first phase expected to be operational by July 2025. This $600 million facility will support Google’s core services, including Gmail and Google Cloud, and ties into the establishment of a new Phoenix cloud region. The project underscores Google’s aggressive expansion to meet AI and cloud infrastructure demands, leveraging Arizona’s growing status as a data center hotspot. Read about Google’s Mesa project.

Aligned Data Centers’ Texas Campus

Aligned Data Centers is expanding in the Dallas-Fort Worth area with a new 27-acre campus in Mansfield, Texas, named DFW-03. Set to go live in Q4 2025, this facility will support AI, cloud, and enterprise workloads with an on-site substation to ensure power reliability. The project responds to the region’s booming demand for data center capacity, driven by Texas’ business-friendly environment and robust energy infrastructure. Discover Aligned’s Texas expansion.

Northern Data’s 120 MW Georgia Project

Northern Data Group is developing a 120-megawatt high-performance computing data center in Maysville, Georgia, approximately 70 miles northeast of Atlanta. The 63-acre campus, expected to be fully operational by Q1 2027, will cater to AI-driven workloads. This project highlights Georgia’s emergence as a key data center market, bolstered by its access to power and proximity to Atlanta’s tech ecosystem. Learn about Northern Data’s Georgia campus.

Teton Digital’s 100 MW North Dakota Development

Teton Digital has received approval for a 100-megawatt data center in North Dakota, a state increasingly attractive for its abundant energy resources and cold_FWD:1 cold climate, ideal for cooling efficiency. The project, greenlit in early 2025, aims to support AI and high-performance computing with a focus on sustainable energy integration. This development positions North Dakota as an emerging player in the U.S. data center landscape. Read about Teton Digital’s North Dakota project.

Challenges and Opportunities

These projects face challenges such as power infrastructure bottlenecks, regulatory hurdles, and land scarcity in traditional data center hubs like Northern Virginia. However, innovations like nuclear power integration—seen in Amazon’s and Microsoft’s nuclear deals—and sustainable cooling solutions are creating opportunities. The U.S. Department of Energy’s plan to fast-track data center development on federal land, leveraging nuclear reactors, could further accelerate growth. Explore grid bypass and nuclear trends.

The economic impact is significant, with projects generating thousands of jobs and billions in local investment. For instance, Google’s Indiana project is expected to bolster Fort Wayne’s economy, while Amazon’s Ohio campus will create long-term employment opportunities. Strategic site selection and partnerships with utilities are critical to overcoming power constraints. See JLL’s 2025 Data Center Outlook.

Looking Ahead

The U.S. data center market is on track to add 10 gigawatts of new capacity in 2025, driven by AI and cloud demand. Projects like Homer City’s 4.5 GW campus and Google’s Arizona facility highlight the scale of investment, while emerging markets like North Dakota and Georgia offer new growth opportunities. As tech giants and developers navigate power and regulatory challenges, nuclear energy and federal land initiatives could redefine the industry’s trajectory. Discover major 2025 data center projects.

— Reported based on DataCenterKnowledge, Construction Dive, and industry insights, May 2025

May 11, 2025

Google is intensifying its push into nuclear energy to meet the soaring power demands of its AI-driven data centers, announcing a strategic partnership with Elementl Power to develop three advanced nuclear sites and a prior deal with Kairos Power for small modular reactors (SMRs). These initiatives aim to deliver over 2.3 gigawatts of clean, reliable energy by 2035, aligning with Google’s net-zero carbon goals. This article explores Google’s nuclear investments, the technology behind SMRs, and the implications for the AI-powered future. Read about Google’s Elementl partnership.

Google’s Partnership with Elementl Power

Google has committed early-stage capital to Elementl Power, a South Carolina-based nuclear developer, to prepare three U.S. sites for advanced nuclear projects, each capable of generating at least 600 megawatts. This 1.8-gigawatt initiative supports Elementl’s goal of deploying 10 gigawatts by 2035. Google retains the option to purchase power once the sites are operational, ensuring a steady supply for its data centers. The deal, announced in May 2025, underscores Google’s role in financing clean energy to fuel AI innovation. Learn more about the Elementl deal.

[](https://www.datacenterdynamics.com/en/news/google-partners-with-elementl-to-advance-development-of-three-advanced-nuclear-projects-in-us/)

Kairos Power and SMR Innovation

In October 2024, Google signed a landmark agreement with Kairos Power to purchase power from a fleet of SMRs, targeting 500 megawatts by 2035, with the first deployment expected by 2030. Kairos’ fluoride salt-cooled, high-temperature reactors use a molten-salt cooling system and ceramic pebble-type fuel, offering a compact, efficient design. These SMRs promise faster deployment and proximity to data centers, addressing the AI sector’s need for 24/7 carbon-free power. Explore Google’s Kairos partnership.

[](https://www.theregister.com/2025/05/07/google_signs_another_nuclear_deal/)

Why Nuclear for AI Data Centers?

The AI boom has driven data center power consumption to unprecedented levels, with a 2024 McKinsey report noting that modern data centers require 200 megawatts, up from 30 megawatts a decade ago. Nuclear power, with its consistent, zero-carbon output, is ideal for meeting these demands. Google’s investments align with its goal to mitigate a 48% emissions increase over five years, largely due to AI expansion. Nuclear offers a reliable alternative to fossil fuels, supporting Google’s 2030 carbon-free energy target. Read about nuclear’s role in data centers.

[](https://www.powermag.com/google-will-fund-three-nuclear-projects-with-elementl-power/)[](https://www.theregister.com/2025/05/07/google_signs_another_nuclear_deal/)

Industry-Wide Nuclear Trend

Google joins other tech giants like Amazon, which invested $500 million in X-energy’s SMRs, and Microsoft, which is restarting Three Mile Island, in embracing nuclear energy. Elementl’s “technology-agnostic” approach allows flexibility in choosing reactor designs, potentially leveraging Kairos or other SMR providers. These partnerships reflect a broader industry shift toward nuclear to power AI, with U.S. data center electricity demand projected to rise 130% by 2030. Discover tech’s nuclear investments.

[](https://www.foxbusiness.com/technology/google-signs-deal-nuclear-developer-trio-advanced-energy-projects)

Challenges and Opportunities

Advanced nuclear projects face significant hurdles, including regulatory delays, high costs, and public skepticism about safety. No commercial SMRs are operational in the U.S., and past projects, like NuScale’s, have faced cancellations due to cost overruns. However, Google’s funding for permitting, grid connections, and contractor hiring accelerates Elementl’s timeline. The partnership’s focus on utility collaboration and site selection could streamline development, setting a model for future projects. Learn about SMR challenges.

[](https://techcrunch.com/2025/05/09/google-inks-deal-to-develop-1-8-gw-of-advanced-nuclear-power/)

Looking Ahead

Google’s nuclear strategy positions it as a leader in clean energy innovation, with Elementl’s 1.8 gigawatts and Kairos’ 500 megawatts forming a robust pipeline for AI data center power. By 2035, these projects could transform the energy landscape, delivering over 2.3 gigawatts of carbon-free electricity. As AI demand grows, Google’s investments signal confidence in nuclear’s role in a sustainable, tech-driven future. Explore Google’s nuclear vision.

[](https://blog.google/feed/google-and-elementl-nuclear-energy-site-development/)

— Reported based on World Nuclear News, The Register, and industry insights, May 2025

May 11, 2025

As artificial intelligence (AI) drives unprecedented energy demands, tech giants are bypassing traditional power grids to secure reliable, carbon-free energy for their data centers. Microsoft’s deal to restart Three Mile Island’s Unit 1 and Amazon’s nuclear-powered data center acquisitions highlight a growing trend of direct nuclear energy procurement. This article explores how data centers are leveraging nuclear power to meet AI’s energy needs, the challenges of grid reliability, and the implications for the energy landscape. Read more on grid bypass trends.

Microsoft’s Three Mile Island Revival

Microsoft has signed a 20-year agreement with Constellation Energy to restart Three Mile Island’s Unit 1, a nuclear reactor shuttered in 2019 for economic reasons, to power its AI data centers. Renamed the Crane Clean Energy Center, the 837-megawatt facility is expected to come online by 2028, supplying carbon-free energy to Microsoft’s data centers across Pennsylvania, Chicago, Virginia, and Ohio. This deal, the first U.S. nuclear reactor restart, aims to support Microsoft’s goal of becoming carbon negative by 2030. Learn about the Microsoft-Constellation deal.

[](https://www.datacenterdynamics.com/en/news/three-mile-island-nuclear-power-plant-to-return-as-microsoft-signs-20-year-835mw-ai-data-center-ppa/)

Bypassing the Grid: A Growing Trend

Data centers are increasingly seeking direct power sources to avoid grid constraints. Amazon’s $650 million purchase of a data center campus next to the Susquehanna nuclear plant in Pennsylvania allows it to draw up to 960 megawatts directly from the facility, bypassing the grid. However, the Federal Energy Regulatory Commission (FERC) blocked Amazon’s request for additional direct power, citing risks to grid reliability and consumer costs. This trend reflects the immense energy needs of AI, with data centers projected to consume 9% of U.S. electricity by 2030. Explore Amazon’s nuclear data center acquisition.

[](https://thebulletin.org/2024/12/ai-goes-nuclear/)

Why Nuclear Power?

Nuclear energy’s appeal lies in its reliability and carbon-free output, critical for AI data centers requiring constant power. Unlike solar or wind, nuclear plants provide stable baseload energy, making them ideal for tech companies with aggressive sustainability goals. Microsoft’s Three Mile Island deal and Amazon’s Susquehanna project underscore nuclear’s role in meeting these demands. Industry experts note that nuclear restarts, like Three Mile Island and Michigan’s Palisades plant, are faster and cheaper than building new reactors. Read about nuclear’s role in AI power.

[](https://www.powermag.com/microsoft-would-restart-three-mile-island-nuclear-plant-to-power-ai/)

Challenges and Regulatory Hurdles

Bypassing the grid raises concerns about equity and grid stability. FERC’s rejection of Amazon’s direct power request highlights fears that tech giants’ energy demands could raise costs for other consumers or strain the grid. Microsoft’s Three Mile Island restart requires significant upgrades—turbines, generators, and cooling systems—and Nuclear Regulatory Commission (NRC) approval, with a review process expected to conclude by 2027. Critics also point to nuclear’s high costs and unresolved waste storage issues. Discover nuclear energy challenges.

[](https://www.reuters.com/markets/deals/constellation-inks-power-supply-deal-with-microsoft-2024-09-20/)

Economic and Environmental Impacts

The Three Mile Island restart is projected to create 3,400 jobs and contribute $16 billion to Pennsylvania’s GDP, alongside $3 billion in taxes, according to Constellation. Nuclear power supports tech companies’ decarbonization goals, but critics argue that monopolizing clean energy could limit access for other sectors. Environmental advocates stress the need for sustainable AI practices beyond nuclear reliance, citing the long-term challenges of radioactive waste. Learn about the economic impacts.

[](https://www.washingtonpost.com/business/2024/09/20/microsoft-three-mile-island-nuclear-constellation/)

Looking Ahead

The shift toward nuclear-powered data centers signals a new era for energy procurement. As AI continues to drive power demands, tech giants like Microsoft, Amazon, and others are investing in nuclear restarts and small modular reactors (SMRs) to secure reliable energy. While these efforts align with climate goals, balancing grid reliability, consumer costs, and environmental concerns will be critical. The success of projects like Three Mile Island could redefine how data centers power the AI revolution. Explore the future of nuclear-powered data centers.

[](https://www.technologyreview.com/2024/09/26/1104516/three-mile-island-microsoft/)

— Reported based on DataCenterKnowledge, The Register, and industry insights, May 2025

May 10, 2025

Amazon is diving headfirst into nuclear energy, announcing a $500 million investment in small modular reactors (SMRs) to fuel its growing data center empire, as reported by The Register. With AI-driven energy demands soaring, the tech giant is partnering with X-energy, Energy Northwest, and Dominion Energy to develop SMR projects aiming for 5 gigawatts of carbon-free power by 2039. This article explores Amazon’s nuclear ambitions, the promise of SMRs, and the challenges ahead. Read the full Register article.

Amazon’s Nuclear Push

Amazon’s investment includes a $500 million Series-C funding round for X-energy, a leader in SMR technology, to accelerate reactor development. The company is also supporting three new nuclear projects, including a feasibility study with Energy Northwest in Washington state and an SMR project near Dominion Energy’s North Anna nuclear station in Virginia. These initiatives aim to deliver reliable, carbon-free energy to meet the surging power needs of AWS data centers. Learn more about Amazon’s nuclear strategy.

The Promise of Small Modular Reactors

SMRs are compact nuclear reactors designed for faster construction and deployment closer to power grids compared to traditional plants. Amazon highlights their potential for “faster build times” and scalability, with X-energy’s Xe-100 reactors targeting 320 megawatts initially, expandable to 960 megawatts—enough to power over 770,000 homes. AWS CEO Matt Garman emphasizes nuclear’s role in achieving Amazon’s net-zero carbon goal by 2040. Explore X-energy’s SMR technology.

Addressing AI Data Center Energy Demands

The AI boom is pushing data center energy consumption to new heights, with estimates suggesting data centers could account for 9% of U.S. electricity by 2030. Amazon’s SMR projects, alongside its earlier $650 million acquisition of a nuclear-powered data center in Pennsylvania, reflect a strategic shift toward reliable, 24/7 clean energy sources to support AI workloads. Read about data center energy trends.

Industry-Wide Nuclear Momentum

Amazon isn’t alone in its nuclear pivot. Google has partnered with Kairos Power for SMRs, Microsoft signed a deal to restart Three Mile Island, and Oracle secured permits for SMR-powered data centers. These moves signal a growing industry consensus that nuclear energy, particularly SMRs, is critical for meeting AI-driven power demands while maintaining sustainability goals. Discover Big Tech’s nuclear investments.

Challenges and Criticisms

Despite the enthusiasm, SMRs face hurdles. No commercial SMRs are operational in the U.S., with only NuScale’s design approved by the Nuclear Regulatory Commission. Critics argue SMRs are costly and produce radioactive waste, while projects like NuScale’s Idaho initiative have collapsed due to financial issues. Regulatory delays and public perception also pose risks. Learn about SMR challenges.

Looking Ahead

Amazon’s SMR investments mark a bold step toward redefining data center power infrastructure. With projects targeting the early 2030s and a goal of 5 gigawatts by 2039, the company is betting on nuclear to bridge the gap between AI growth and sustainability. As regulatory and technical challenges loom, Amazon’s partnerships with X-energy and utilities could set a precedent for the industry. Explore Amazon’s 5GW nuclear goal.

— Reported based on The Register and industry insights, May 2025

May 10, 2025

Despite recent adjustments by tech giants Amazon and Microsoft, the global AI data center market continues its robust expansion, driven by the escalating demand for artificial intelligence (AI) infrastructure. According to a Wells Fargo research note, both companies have paused or slowed certain data center projects, yet the broader industry remains resilient. This article delves into the dynamics of this growth, Microsoft’s strategic shifts, and the enduring momentum in the AI data center sector. Read the full NextBigFuture article.

Microsoft’s Strategic Adjustments

Microsoft has made headlines with reports of canceling 2GW of non-binding Letters of Intent (LOIs) for data center leases. However, this represents only a fraction of their activity, as the company holds approximately 5GW of pre-leased capacity under binding contracts set to commence operations between 2025 and 2028. In mid-2024, Microsoft was in talks with nearly every major vendor for additional capacity but has since frozen new leasing activity. Instead, the company is ramping up self-build efforts, acquiring tens of thousands of acres globally and securing gigawatts of power for future sites. Learn more about Microsoft’s AI investments.

Amazon’s Pause on Colocation

Amazon Web Services (AWS) has also adjusted its data center strategy, pausing discussions on some colocated data center projects, according to industry sources cited by Wells Fargo. Despite this, Amazon’s overall capital expenditure remains substantial, with plans to invest $100 billion in 2025, a significant increase from $83 billion in 2024, primarily to bolster AI infrastructure. This reflects Amazon’s commitment to maintaining leadership in the cloud and AI markets. Explore Amazon’s 2025 spending plans.

Industry-Wide Growth and Resilience

The broader AI data center market is far from slowing down. Industry analyses, such as those from SemiAnalysis, indicate that despite Microsoft’s pause on new leases, their self-build initiatives and existing contracts ensure continued growth. Other major players, including Google ($75 billion) and Meta ($65 billion), are also increasing capital expenditures, contributing to a projected $320 billion in AI infrastructure spending by the top four companies in 2025. This underscores the sector’s resilience amid strategic recalibrations. Read about the AI data center boom.

Opportunities for Equipment Suppliers

The ongoing data center expansion creates significant opportunities for equipment suppliers like Vertiv, which benefit from the demand for cooling, power management, and other critical infrastructure components. Even with Microsoft’s 1.5GW self-build pause, their global construction and existing contracts drive substantial activity for suppliers. This trend extends to other providers scaling up to meet AI-driven compute demands. Discover opportunities in data center supply chains.

Economic and Strategic Implications

The adjustments by Amazon and Microsoft highlight a shift toward aligning capital expenditures with economic demand, as noted in industry commentary. While public demand for AI services is high, economic demand—what users are willing to pay for—requires careful calibration. This strategic reallocation, rather than a retreat, ensures sustainable growth. The involvement of other players like OpenAI and Oracle in projects like the $500 billion Stargate initiative further signals long-term confidence in AI infrastructure. Learn about the Stargate project.

Looking Ahead

The AI data center market is poised for continued growth, with 2025 marking a pivotal year for infrastructure scaling. As companies like Microsoft and Amazon refine their strategies, the industry’s focus on self-builds, energy efficiency, and sustainable power solutions will shape its trajectory. With billions in investments and innovations driving the sector, AI data centers remain a cornerstone of the global tech landscape. Explore the $2 trillion AI data center market.

— Reported based on NextBigFuture and industry insights, May 2025

May 10, 2025

The data center industry is poised for unprecedented growth in 2025, driven by the insatiable demand for artificial intelligence (AI) and high-performance computing (HPC). According to JLL’s 2025 Global Data Center Outlook, the sector is expected to expand at a phenomenal pace, with a projected compound annual growth rate (CAGR) of 15-20% through 2027. This article explores the key trends, challenges, and opportunities shaping the future of data centers, with a focus on AI innovation, power infrastructure, and sustainable energy solutions. Read the full JLL report.

The AI Revolution and Data Center Demand

At the heart of the data center boom lies the rapid advancement in semiconductor technology, particularly graphics processing units (GPUs). Tasks that once took 32 hours can now be completed in just one second, enabling AI models to train on increasingly large datasets. This acceleration is enhancing the value of the AI ecosystem, fueling innovation, and driving demand for data center capacity. Tech companies, the largest occupiers of data center space, are leveraging this power to meet their aggressive net-zero targets while scaling their operations. Learn more about GPU advancements.

The rise of generative AI is also transforming data center design. Higher rack density, liquid cooling adoption, and energy efficiency improvements are becoming critical to support AI workloads. As the digital economy and data storage needs grow, data centers are evolving to meet these demands, with providers exploring new markets to overcome limitations in power and land availability. Explore liquid cooling trends.

Power Infrastructure Challenges

Despite the sector’s growth, power infrastructure bottlenecks remain a significant hurdle. Power scarcity and extended timelines for building transmission lines are major impediments to data center development, particularly as the industry expands into new geographies. Utilities are addressing this by becoming more selective in approving Purchase Power Agreements (PPAs), using thorough intake forms and application fees to prioritize well-funded projects. While this helps allocate resources efficiently, it does not resolve the long lead times for infrastructure development. Read about power constraints.

These challenges are compounded by the strain on traditional power grids, which struggle to keep up with the energy demands of AI and HPC applications. As a result, data center operators are seeking innovative solutions to ensure reliable and scalable power supplies. Learn about energy demands in data centers.

Nuclear Power: A Sustainable Solution

Nuclear power is emerging as a preferred solution to meet the energy demands of data centers. Offering a clean and reliable alternative to traditional grids, nuclear energy aligns with the sustainability goals of tech companies while addressing power scarcity. The enthusiasm for nuclear power is growing, particularly for AI-driven data centers, as it supports both high energy requirements and net-zero ambitions. This shift toward nuclear energy is expected to play a pivotal role in shaping the future of data center development. Discover nuclear power for data centers.

Strategic Opportunities for Growth

Amid these challenges, opportunities abound for data center investors, developers, and operators. JLL’s report highlights the importance of strategic site selection, energy procurement, and innovative design to overcome supply constraints. Flexible, long-term leases are becoming the norm, allowing users to adapt to evolving technology needs over 5 to 10 years. Additionally, emerging markets are gaining attention as operators seek locations with available power and land to support expansion. Explore JLL’s data center services.

Collaborating with real estate experts, such as JLL’s research team, can provide valuable market insights and strategic advice to navigate this dynamic landscape. From colocation site selection to facility management, end-to-end solutions are essential for optimizing data center strategies. Access more JLL insights.

Conclusion

The data center industry is at a critical juncture, with AI and HPC driving unprecedented demand. While power infrastructure challenges persist, solutions like nuclear energy and selective PPA approvals are paving the way for sustainable growth. By embracing innovation and strategic planning, stakeholders can seize the opportunities presented by this rapidly evolving sector. For more insights, download JLL’s 2025 Global Data Center Outlook report or connect with their research team to shape your data center strategy. Contact JLL’s research team.

— Reported based on JLL’s 2025 Global Data Center Outlook, May 2025

May 5, 2025

While modular data centers, pioneered by companies like Sun Microsystems and Dell in the mid-2000s, transformed IT infrastructure, a new wave of innovation is reshaping power distribution: modular substations. These pre-fabricated, flexible units offer the same functions as traditional substations but with enhanced efficiency, scalability, and rapid deployment. Miller Industries is at the forefront of this revolution, delivering cutting-edge solutions for modern energy needs. This article explores the innovations, benefits, and market trends driving the modular substation boom.

The Rise of Modular Substations

Modular substations are gaining traction as a game-changer in power infrastructure, offering faster installation and greater flexibility than conventional substations. According to a 360iResearch report, the modular substation market is projected to grow by USD 28.84 billion at a CAGR of 7.25% by 2030, driven by the need for scalable and efficient power solutions. From urban grids to remote industrial sites, these units are addressing the global demand for reliable electricity with minimal downtime.

Key Innovations Driving Adoption

Innovations in modular substation design are accelerating their adoption. Esfro Solutions’ Powerskid exemplifies the plug-and-play approach, enabling rapid deployment with minimal on-site work. Similarly, Flex Air’s 10,000 kVA (5.6MW) modular substation reduces labor costs through off-site manufacturing, minimizing disruptions during installation. These advancements, including compact designs and advanced monitoring systems, make modular substations ideal for dynamic energy demands in industries like utilities and renewables.

Sustainability and Scalability Benefits

Modular substations are uniquely suited to support sustainable energy systems. As noted by DataHorizzon Research, their flexibility accommodates the variability of renewable energy sources like solar and wind, enabling smoother grid integration. Their scalability allows operators to expand capacity without extensive retrofitting, making them a cost-effective choice for urban settings and growing energy demands. By reducing construction waste and energy losses, modular substations align with global sustainability goals.

Market Trends and Growth Projections

The modular substation market is shaped by trends like cost-efficiency, rapid deployment, and enhanced reliability. Global Market Insights highlights their long-term benefits, such as reduced operational downtime and improved safety, which are driving adoption in power utilities and industrial applications. With a projected market size of USD 47.69 billion by 2032 at a CAGR of 7.7%, the industry is poised for robust growth, fueled by urbanization and electrification initiatives.

Challenges and Opportunities

Despite their advantages, modular substations face challenges, including higher initial costs compared to traditional setups. However, their quick installation and flexibility offer significant opportunities, particularly in urban areas and temporary setups like construction sites or disaster recovery. As outlined in industry analyses by Mordor Intelligence, the Asia-Pacific region, with its rapid urbanization, is expected to lead market growth, creating opportunities for innovative designs tailored to dense environments.

Looking Ahead

The future of modular substations is bright, with their role in global electrification and sustainability becoming increasingly vital. As investments in renewable energy and smart grids accelerate, modular substations will provide the backbone for resilient, efficient power systems. Future Market Insights projects the market to reach USD 48 billion by 2034, underscoring their potential to transform energy infrastructure. By addressing challenges and embracing innovation, the industry can power a sustainable, electrified future.

— Reported based on industry insights and market research, May 2025

May 3, 2025

The data center industry is undergoing a transformative shift, driven by the demands of artificial intelligence (AI) and the need for sustainable, high-performance infrastructure. In this dynamic landscape, Avtron Power Solutions is making waves with its innovative load bank solutions, as highlighted in an upcoming episode of Planet TV Studios’ New Frontiers, set to air in Q3 2025. This article explores Avtron’s impact, the growing load bank rental market, and the critical role of liquid cooling in addressing data center challenges.

Avtron Power Solutions Spotlighted on New Frontiers

Planet TV Studios’ New Frontiers series, known for showcasing industry pioneers, will feature Avtron Power Solutions in its Q3 2025 episode titled “New Frontier in Data Centers.” The episode will delve into how Avtron, a Cleveland-based global leader in load bank technology, is redefining power testing for critical infrastructure. With over 70 years of expertise, Avtron’s advanced load banks ensure the reliability of backup power systems in data centers, hospitals, and beyond. The episode highlights groundbreaking products like the LC-20 liquid-cooled load bank, a 500kW unit tailored for data centers adopting liquid cooling to manage heat and boost energy efficiency, and the SIGMA Unity software, which streamlines load bank testing for unparalleled control.

Load Bank Rental Market Set for Exponential Growth

The load bank rental market is poised for significant expansion from 2025 to 2034, driven by the rapid growth of data centers and the need for reliable power testing. Load banks are essential for validating backup power systems, simulating real-world stresses to ensure generators and UPS systems perform under pressure. As data centers scale to meet AI-driven demand, the flexibility of rental load banks offers a cost-effective solution for commissioning, maintenance, and expansion. Industry forecasts predict robust growth in this sector, fueled by the increasing complexity of data center infrastructure and the adoption of advanced testing solutions like Avtron’s liquid-cooled load banks.

Liquid Cooling: A Necessity Facing Challenges

A ScienceDirect article underscores the growing necessity of liquid cooling in data centers, driven by the heat generated by high-density AI workloads. Liquid cooling, including direct-to-chip and immersion systems, offers superior energy efficiency compared to traditional air cooling, addressing the thermal challenges of modern servers. However, challenges such as liquid quality, maintenance complexity, and structural considerations—like reinforced flooring for heavy immersion cooling baths—persist. Despite these hurdles, liquid cooling is becoming a default in new data center designs, with retrofits gaining traction in existing facilities to support AI-driven workloads.

Water-Cooled Load Bank Testing: The Future of Power Validation

Water-cooled load bank testing is emerging as a game-changer for data center commissioning and maintenance. Avtron’s LC-20 liquid-cooled load bank, launched in October 2024, exemplifies this trend. Designed for modern data centers, the LC-20 offers a 500kW capacity with a fine 5kW resolution, corrosion-resistant stainless steel components, and the ability to network with up to 200 load banks. This technology aligns with the shift toward liquid-cooled data centers, providing efficient, reliable testing for high-power environments. Similarly, Aggreko’s 500 kW liquid-cooled load bank supports precise testing of cooling systems, ensuring optimal server performance. Other providers, like ComRent and Simplex, offer tailored liquid-cooled solutions for data center testing, enhancing infrastructure resilience. As AI workloads push rack density beyond 150 kW, water-cooled load banks will play a critical role in ensuring infrastructure resilience. Note: ComRent is no longer in business and has been purchased by Sunbelt.

Balancing Innovation and Sustainability

The rise of AI and liquid cooling introduces both opportunities and complexities for data center operators. Innovations like Avtron’s SIGMA Unity software and liquid-cooled load banks enhance testing precision and efficiency, but the industry must also address sustainability. Liquid cooling reduces energy consumption compared to air cooling, yet the environmental impact of water usage and cooling fluid disposal requires careful management. Avtron’s commitment to UL and CE standards ensures its solutions meet stringent quality and safety requirements, supporting data centers in their quest for sustainable, high-performance operations.

Looking Ahead

As featured in New Frontiers, Avtron Power Solutions is at the forefront of the data center revolution, delivering cutting-edge load bank technology to meet the demands of AI-driven infrastructure. The exponential growth of the load bank rental market and the adoption of liquid cooling signal a future where efficiency and reliability are paramount. By overcoming the challenges of liquid cooling and embracing water-cooled load bank testing, the industry can build resilient, sustainable data centers capable of powering the digital age. The Q3 2025 episode of New Frontiers will offer a compelling look at how Avtron is shaping this future.

— Reported based on industry insights and Avtron Power Solutions announcements, May 2025

May 2, 2025

The demand for data center infrastructure is surging, driven by the explosive growth of artificial intelligence (AI). A recent Colliers report projects a 160% increase in data center demand, with $57 billion invested in the sector in 2024 alone. As this “supercycle” reshapes the industry, data centers face the dual challenge of powering AI’s immense computational needs while prioritizing sustainability.

A $57 Billion Supercycle

The rapid rise of AI has triggered unprecedented investment in data centers, with 2024 marking a historic $57 billion in global spending. According to Colliers, AI-driven workloads are expected to increase data center demand by 160%, pushing the industry to expand capacity at an extraordinary pace. This supercycle is not just about scale—it’s about reengineering infrastructure to handle the unique requirements of AI, from high-density computing to advanced cooling systems.

The Race for AI Leadership

Tech giants are fiercely competing to dominate the AI market, and data centers are the backbone of this race. The need for optimized infrastructure has led to innovations in server design, power distribution, and cooling technologies. However, these advancements come with operational complexities, including managing larger data center footprints, diverse load mixes, and the integration of cutting-edge systems like water-cooled load and capacitive load banks for commissioning new facilities, as discussed in industry analyses.

Sustainability Challenges

As data centers grow to meet AI’s energy demands, sustainability has become a critical concern. AI workloads are notoriously power-hungry, requiring operators to explore diverse energy sources, including renewables like solar and wind, alongside advanced energy storage solutions. The push for sustainability is driving innovation, but it also complicates infrastructure planning as operators strive to balance performance with environmental responsibility, according to initiatives like the Uptime Institute.

Water Cooling and Load Banks

The rise of AI has increased the demand for water-cooled load systems and capacitive load banks during data center commissioning. Water cooling is emerging as a sustainable solution to manage the intense heat generated by AI-driven servers, offering greater efficiency than traditional air-cooling methods, as noted in research on liquid cooling. Meanwhile, capacitive load banks, such as those provided by Avtron Power Solutions, are critical for testing infrastructure under the high-power conditions required by modern data centers, ensuring reliability and performance.

Navigating Operational Complexity

Today’s data centers are more complex than ever, with operators managing larger facilities and diverse workloads. The integration of AI-driven tasks alongside traditional computing requires flexible infrastructure capable of handling varied power and cooling needs. Additionally, the shift toward diverse energy sources, such as renewables and battery storage, adds another layer of complexity, challenging operators to maintain efficiency and reliability, as outlined in resources from Schneider Electric.

Looking Ahead

The AI-driven data center boom is reshaping the industry, pushing the boundaries of innovation while highlighting the need for sustainable practices. As investments continue to flow, the focus must remain on building infrastructure that can power the AI revolution without compromising environmental goals. By addressing challenges like energy diversity, cooling efficiency, and load management, the industry can pave the way for a future where AI and sustainability coexist, as projected in Deloitte’s industry outlook.

— Reported based on industry data and Colliers research, May 2025

May 1, 2025

The United States faces a growing energy deficit, driven by surging demand from data centers and the decommissioning of coal plants, while China ramps up its coal-fired power capacity. President Donald Trump’s declaration of a national energy emergency aims to address this gap by boosting domestic energy production, particularly coal, leveraging America’s vast reserves.

A 45GW Energy Deficit

The rapid expansion of data centers, fueled by AI and cloud computing, is projected to require an additional 45 gigawatts (GW) of power in the U.S. This surge in demand is straining the nation’s grid, with the largest grid operator, PJM Interconnection, warning of potential shortfalls due to planned power plant retirements (PJM Interconnection Reports). The Energy Information Administration (EIA) notes that coal production in 2023 was less than half of 2008 levels, exacerbating the challenge (EIA Coal Data).

China’s Coal Surge

In contrast, China has significantly increased its coal power capacity, commissioning approximately 94.5 GW of new coal plants in 2024, the highest in a decade (Reuters). This expansion, driven by energy security needs, contrasts sharply with global trends toward renewables and gas. China’s low electricity costs and robust infrastructure give it a competitive edge in meeting the energy demands of AI and other technologies (Global Energy Monitor).

U.S. Coal Decommissioning

In 2024, the U.S. decommissioned over 3 GW of coal-fired power capacity, part of a broader trend where 8 GW is slated for retirement in 2025, representing nearly 5% of the 2024 coal fleet (EIA Coal Data). This reduction, driven by cheaper natural gas and renewables, has left the U.S. with about 200 coal plants generating 16% of its electricity. The closure of plants like the Homer City Power Plant in Pennsylvania, now being redeveloped into a data center campus, highlights the shift away from coal.

America’s $9 Trillion Coal Reserves

Despite the decline, the U.S. holds an estimated $9 trillion worth of coal reserves ready to be mined. Industry advocates argue that coal can be repurposed with cleaner technologies, such as Core Natural Resources’ efforts to create synthetic materials for lithium-ion batteries and carbon foam for aerospace applications (Core Natural Resources). These innovations aim to reduce reliance on foreign supply chains, particularly from China.

Trump’s National Energy Emergency

On January 20, 2025, President Trump declared a “national energy emergency” under the National Emergencies Act, emphasizing domestic fossil fuel production with his “drill, baby, drill” mantra. The White House announcement highlighted the need to fast-track energy infrastructure to meet rising demand (White House Executive Order). Executive orders signed in April 2025 aim to extend the life of coal plants and allow mining on federal land, reversing environmental regulations to keep older, dirtier plants operational (Washington Post).

Trump’s actions, however, face challenges. Experts note that coal’s decline is primarily due to market forces—cheap natural gas and renewables—rather than regulations alone. The New York Times reports that these economic realities may limit the impact of Trump’s policies (New York Times). Critics also warn of environmental trade-offs, as coal combustion emits more carbon and pollutants than other fuels (Scientific American).

— Reported based on industry data and White House announcements, May 2025

April 30, 2025

At the Hill and Valley Forum in Washington, D.C. (Hill and Valley Forum), Nvidia CEO Jensen Huang delivered a stark warning about the intensifying AI supremacy race between the U.S. and China, while highlighting AI’s transformative potential to revolutionize industries and create a surge in trade jobs.

A Fierce, Infinite AI Race

In an interview with CNBC’s Squawk on the Street (CNBC Interview), Huang emphasized the closeness of the competition. “China is right behind us. We’re very, very close,” he said, framing the rivalry as a long-term endeavor. “This is a long-term, an infinite race. In the world of life, there’s no two-minute drill or end of the quarter.” His remarks underscore the high stakes of AI development, with both nations investing heavily in the technology that promises to reshape global economies (Reuters).

AI’s Industrial Transformation

During a conversation with Jacob Helberg, Under Secretary of State for Economic Growth, Energy, and the Environment nominee, Huang outlined AI’s impact on manufacturing. He predicted a shift toward autonomous systems across industries. “Every company that makes things today—whether lawnmowers or construction machinery—will shift from manual to autonomous or semi-autonomous systems,” he said. This transition will require “AI factories” to produce the software that powers these systems, alongside traditional factories building physical products (Forbes).

A Net-Positive for Jobs

Huang acknowledged that AI will disrupt the labor market but argued its overall impact will be positive. “New jobs will be created, some jobs will be lost, every job will be changed,” he said, pointing to San Francisco’s AI-driven economic revival as evidence. The shift from human-coded software on CPUs to machine-learning-generated code on GPUs is creating new roles in data curation and AI safety. “All of that technology is being invented right now, and it creates tons of jobs,” Huang noted (Wall Street Journal).

The Rise of AI Factories

The most significant opportunity, according to Huang, lies in the construction of “AI factories”—massive facilities that convert electricity into computational intelligence. These factories represent a new industrial frontier, with a single one-gigawatt facility costing $60 billion, comparable to Boeing’s annual revenue (Bloomberg). Huang described plans for even larger 7-10 gigawatt factories, which will drive demand for skilled labor. “You need carpenters, steelworkers, masons, mechanical and electrical engineers, plumbers, and IT specialists,” he said, estimating a three-year construction cycle per factory (MIT Technology Review).

Huang’s vision paints a future where AI not only fuels technological innovation but also revitalizes the skilled trades, creating a new wave of economic opportunities.

Insights from the Hill and Valley Forum

Watch Jensen Huang discuss the global AI race, AI factories, and their impact on industries in this clip from the Hill and Valley Forum:

— Reported based on remarks by Jensen Huang at the Hill and Valley Forum, April 2025

April 29, 2025

Buckle up, because Kevin O’Leary, aka Mr. Wonderful, is diving headfirst into the A.I. revolution, and he’s got his sights set on North Dakota with a jaw-dropping 45 GW of power to fuel the future! Here’s why his plan is as bold as his Shark Tank zingers—and why you should be paying attention (Valley News Live).

A.I. and Data Centers: The Gold Rush of the 2020s

If you’re 25 and hungry for opportunity, Mr. Wonderful’s dropping wisdom like it’s a Shark Tank pitch: A.I. implementation and data center development are where the money’s at. Small businesses are tripping over themselves to adopt A.I. but don’t know where to start. That’s your cue—swoop in, solve their problems, and cash in. Meanwhile, data centers are the backbone of this tech tsunami. With hyperscalers like Tesla, Microsoft, and Google gobbling up capacity faster than you can say “cloud computing,” real estate and tech are colliding in the most lucrative way possible. As O’Leary puts it, “This is where the future’s heading. Don’t miss it” (Data Center Frontier).

Alberta’s Got Gas, But North Dakota’s Got Guts

While O’Leary’s raising $70 billion to build the world’s lowest-cost, highest-efficiency data center in Alberta—powered by abundant natural gas and a premier who’s all-in (Calgary Herald, TechRepublic)—North Dakota’s stealing the spotlight. Why? It’s got the land, the power, and the leadership to make A.I. and data centers thrive. With 45 GW on the table, North Dakota’s ready to power the A.I. revolution like nobody’s business. O’Leary’s already investing here, and he’s shouting it from the rooftops (or at least on AM 1100 The Flag): “Data centers, A.I., agtech, drones—North Dakota’s wide open for business!” (Kevin O’Leary Official, UND Today).

China’s Out, North Dakota’s In

Mr. Wonderful’s not mincing words: China’s been “stealing from American entrepreneurs for years,” and it’s time they face the music. Meanwhile, North Dakota’s rolling out the red carpet for innovation. Forget the Great Wall—think Great Plains, where A.I., data centers, and agtech are about to make waves. O’Leary’s betting big, and he’s got the receipts to prove it (Reuters).

Watch Mr. Wonderful Break It Down

Don’t just take our word for it—check out O’Leary laying out the vision himself in this video from AM 1100 The Flag. It’s got all the energy of a Shark Tank deal about to close:

Why North Dakota? Why Now?

Power Galore: 45 GW of juice to keep A.I. and data centers humming.
Land for Days: Wide-open spaces perfect for sprawling tech campuses.
Leadership with Vision: North Dakota’s ready to play ball and win the A.I. race.
Mr. Wonderful’s Stamp of Approval: If O’Leary’s investing, you know it’s legit (Kevin O’Leary Official).

So, whether you’re a 25-year-old hustler ready to help small businesses conquer A.I. or an investor eyeing the next big thing, North Dakota’s calling. Mr. Wonderful’s already there, powering up 45 GW of pure opportunity. Don’t get left in the dark!