Find a Data CenterSearch 6,503 Data Centers

A Bold Bet on Nuclear-Powered Cloud InfrastructureIn a transformative move that may redefine the energy backbone of cloud computing, Amazon announced plans to invest over $20 billion in a nuclear-powered data center hub in northeastern Pennsylvania. The e-commerce and cloud giant, through its subsidiary Amazon Web Services (AWS), has acquired the Three Mile Island nuclear facility, aiming to retrofit and repurpose the once-infamous site into a state-of-the-art AI-ready hyperscale region powered by carbon-free nuclear energy.The announcement comes at a critical juncture when tech giants are seeking scalable, sustainable energy sources to power energy-hungry AI workloads. Nuclear energy—long criticized for its risks—is getting a major second look as nations and corporations scramble to decarbonize while expanding compute capacity.This article explores why Amazon’s nuclear strategy could be a masterstroke, what it means for data center sustainability, how the local and global markets may respond, and whether nuclear energy is the future of hyperscale computing.Why Amazon Is Turning to Nuclear EnergyAmazon’s decision to pursue nuclear power is driven by two converging pressures:1. Exponential Energy Demands from AIFrom large language models to generative AI platforms, Amazon’s compute needs are exploding. Training a single frontier model like Titan or Bedrock consumes megawatt-hours of energy per day. Maintaining global workloads at scale requires energy reliability that solar or wind alone cannot guarantee.2. Sustainability MandatesAmazon has committed to powering all operations with 100% renewable or carbon-free energy by 2030. But as renewable generation struggles with intermittency and land-use issues, nuclear energy offers:Zero carbon emissionsHigh capacity factor (above 90%)Stable baseload power supplySmall geographic footprintThe Three Mile Island Legacy and Amazon’s AcquisitionThe site of one of the worst nuclear accidents in U.S. history, Three Mile Island has long been a symbol of nuclear caution. Its partial meltdown in 1979 led to stricter U.S. nuclear regulation and a halt to many plant constructions.Fast forward to 2025, and Amazon’s acquisition of the site marks a symbolic turnaround: transforming a site of risk into a pillar of next-gen infrastructure.The Plan:Retrofitting existing nuclear infrastructure for hyperscale data center useBuilding 12 modular data center buildingsCo-locating small modular reactors (SMRs) to supplement existing nuclear outputEmploying 3,500 construction workers and 800 permanent rolesIt’s the first known case of a tech company directly owning a nuclear power generation asset.Technical Overview — Building a Nuclear-Backed Hyperscale CloudPower ProfileBase Load: 852 MW from Three Mile Island Unit 1 reactorSMRs (Planned): 2 x 300 MW units operational by 2028Total capacity: Over 1.4 GW, enough to power 1.2 million homesFacility DetailsModular data halls with liquid-cooled racks for AI chipsIntegration with AWS Nitro-based compute and storageQuantum and AI-ready clusters using custom Inferentia and Trainium chipsHardened physical and cybersecurity systems due to the nuclear settingLocal Economic Impact in PennsylvaniaThis initiative could revitalize Pennsylvania’s nuclear and industrial economy. Once a site of controversy, Three Mile Island is now poised to generate:$10B in direct economic activity3,500+ construction jobs over 5 years800 permanent technical and operational rolesNew STEM-focused university partnerships (Penn State, Carnegie Mellon)Amazon has committed to community reinvestment, promising grants for STEM education, infrastructure upgrades, and environmental restoration.AWS’s Broader Sustainability StrategyAmazon’s $20B nuclear investment is part of a broader cloud sustainability framework that includes:Over 500 solar and wind projects globallyInvestments in carbon capture and low-water coolingHydrogen energy pilots at edge data centersResearch into fusion energy via partnerships (e.g., Helion Energy)This move positions AWS as not only the largest cloud provider but potentially the cleanest at hyperscale.A Global Shift Toward Nuclear-Powered Data CentersAmazon isn’t alone. Around the world, nuclear is re-emerging as a solution to the AI power dilemma:Microsoft + Helion: Fusion-powered pilot project by 2028France: Data centers in Paris powered by EDF’s nuclear gridChina: Dozens of nuclear-powered data zones for AI and quantum researchIndia and UAE: Modular nuclear reactors near new hyperscale campusesWith global data center electricity demand expected to reach 1,000 TWh by 2030, the pressure is on to balance compute expansion with climate responsibility.Addressing Nuclear’s Risks and Public ConcernsDespite its benefits, nuclear energy remains controversial. Amazon’s plans must overcome:Safety ConcernsUse of modern SMRs and AI monitoring for failsafe controlsU.S. Nuclear Regulatory Commission (NRC) oversightRemote operation zones and radiation buffersPublic PerceptionTransparency campaigns and town hallsEnvironmental impact assessmentsCommunity advisory councilsAmazon has also committed to carbon neutrality and zero radioactive waste leakage in compliance with international standards.The Future — Are Nuclear-Powered Clouds the Next Normal?As AI and cloud workloads surge, data centers are projected to consume up to 8% of global electricity by 2030. The transition to nuclear-backed compute hubs offers a powerful answer to:Grid reliability problemsESG pressures from institutional investorsSovereign compute strategies (esp. in defense, healthcare, AI)By taking direct ownership of energy infrastructure, Amazon is not just a tech company — it’s becoming a vertically integrated energy-and-compute enterprise.Amazon’s High-Stakes Gamble Could Redefine Cloud EnergyAmazon’s $20B investment in nuclear-powered data centers is both a milestone and a message. As the AI era dawns, it’s no longer enough to scale compute — it must scale sustainably.This is not just about gigawatts and GPUs. It’s about building the future of cloud infrastructure with a foundation in climate-safe energy. If successful, Amazon could lead the industry into a new model: one where the cloud runs not on coal or compromise, but on clean, consistent, carbon-free nuclear power.The world will be watching. And if Amazon’s nuclear bet pays off, it may forever change the energy calculus of the digital age.

The Dawn of a New AI Super-InfrastructureIn a landmark coalition that signals the next phase of global AI infrastructure development, Temasek Holdings, Microsoft, and BlackRock announced a joint $30 billion initiative to develop next-generation data center campuses across Asia, the Middle East, and Africa. This strategic collaboration—known internally as Project MGX—aims to accelerate the availability of AI infrastructure in regions historically underserved by hyperscale compute.At the core of this venture is a blend of state-backed capital (Temasek), global AI platform leadership (Microsoft), and financial muscle (BlackRock) to create a new infrastructure layer capable of supporting the next generation of artificial intelligence, machine learning, and quantum workloads.This blog dives into what makes this partnership unique, where the capital is going, and why the AI infrastructure map is about to be redrawn for the next decade.Meet the Players — A Triple Threat of Influence and ReachTemasek Holdings (Singapore)One of the largest sovereign wealth funds globally$300B+ portfolioStrong regional insight into Southeast Asia, Africa, and emerging marketsMicrosoftGlobal cloud leader with Azure regions in over 60 countriesAI powerhouse via OpenAI and in-house chips (Azure Maia, Cobalt)Increasing commitment to sovereign cloud and AI localizationBlackRock$10 trillion under managementInfrastructure investment powerhouseProvides long-term financing and risk structuringTogether, these three entities can fund, build, operate, and scale AI infrastructure—something few single companies can do on their own.Why This Deal Matters — The Infrastructure Gap in the Global SouthAI’s rise has been overwhelmingly concentrated in North America, Europe, and China. But this concentration has led to:Data sovereignty concerns in regions without local computeLatency issues in AI workloads for users outside core marketsLimited access to training infrastructure for startups and governmentsThe MGX alliance aims to resolve this imbalance by building 14 hyperscale campuses across key growth markets including:Jakarta (Indonesia)Nairobi (Kenya)Riyadh (Saudi Arabia)Mumbai (India)Ho Chi Minh City (Vietnam)Each region will get 200–500 MW facilities optimized for AI, co-located with renewable energy, and built using sovereign-compliant cloud standards.The MGX Blueprint — Data Centers Engineered for AIUnlike traditional hyperscale data centers, MGX facilities will be engineered from the ground up for generative AI, multi-modal LLMs, robotics, and quantum simulation. Key features include:1. Liquid-Cooled InfrastructureEvery rack supports 40kW+ densitiesDirect-to-chip cooling with AI-optimized flow regulation2. Modular Renewable IntegrationCo-located solar farms and geothermal installationsBattery-based backup, no diesel3. Azure-Ready AI FabricDeep integration with Microsoft’s Azure StackCustom-built clusters for GPT, DALL-E, Copilot, and Hugging Face integrations4. On-Site Semiconductor CapabilitiesMicro-foundries in Indonesia and Saudi Arabia for custom chip packagingJoint IP development for LLM inference acceleratorsWhy Now? Timing, Geopolitics, and the AI Gold RushThe world is in the midst of a global AI infrastructure shortage. According to industry estimates:The demand for AI servers will grow 7x between 2023 and 2028Over $1.6 trillion in investment is needed to meet global compute needsSovereign AI clouds will be mandated in over 40 countries by 2026Temasek, Microsoft, and BlackRock are stepping into a vacuum.Strategic Advantages:Geopolitical alignment with ASEAN, Gulf Cooperation Council, and African UnionSupply chain control through vertical integrationEdge advantage for smart cities, fintech, telecom, and defense sectorsThe Financials — How $30 Billion Is Being AllocatedThe deal breaks down into three distinct investment tranches:Core Infrastructure ($18B)Construction of 14 campusesReal estate acquisition and grid interconnectionRD + Talent ($6B)Training institutes in partnership with universitiesOpen-source model research hubsAI Capital Fund ($6B)Startup accelerator programs in 10 regionsDedicated GPU leasing platform (Project NeuronGrid)This is not just about buildings—MGX is laying the foundation for an entire AI economy in emerging markets.Regional Case Study – Jakarta’s Rise as Southeast Asia’s Compute HubJakarta is set to receive the first MGX data center, with 480 MW planned by 2026. Reasons for this choice:Explosive digital growth in Indonesia (350M+ internet users)Strong government alignment with sovereign AI strategiesProximity to subsea cables and green energy zonesThe campus will host:A regional LLM tuned for Bahasa and regional dialectsFree AI credits for startups and research institutions12,000 construction jobs and 2,000 permanent rolesThis model will be replicated across MGX zones.Strategic Implications for AI and Cloud1. Democratizing AIMGX’s sovereign zones enable universities, startups, and governments in the Global South to train and deploy models locally.2. Redefining Infrastructure PowerThis alliance displaces the "West-only" paradigm of AI dominance and signals a multipolar future for cloud infrastructure.3. Boosting Local Tech EconomiesMGX zones are designed not just to host data—but to catalyze entire innovation ecosystems, from education to export.4. Creating New Cloud SovereigntiesBy hosting AI within national borders, countries gain:Regulatory controlCultural model alignmentData privacy assuranceMGX Will Reshape AI Access in the 2020sTemasek, Microsoft, and BlackRock’s $30 billion MGX alliance is not just a real estate or tech play. It is a declaration that AI equity matters—that access to compute is as vital as access to internet was in the 1990s.With scalable infrastructure, sovereign compliance, and financial backing, MGX may well become the blueprint for how to build ethical, global AI platforms from the ground up.As 2025 unfolds, one thing is clear: the future of artificial intelligence is being written not just in Silicon Valley, but also in Jakarta, Nairobi, and Riyadh.

Climate Policy Meets ColocationData centers power the cloud, AI, and digital life — but they also consume enormous amounts of energy. In 2025, the European Union has drawn a clear line: data center sustainability is no longer optional. With newly enforced regulations and sustainability reporting mandates under the revised Energy Efficiency Directive (EED), the EU is becoming the first major economic bloc to regulate data center emissions at scale.This blog explores how the EU’s regulatory crackdown is changing the industry, what colocation and cloud providers must do to comply, and why this may set the tone for global data center sustainability policy for the rest of the decade.What’s the EU Doing — and Why Now?The Regulatory Trigger: Energy Efficiency Directive (EED)As part of the European Green Deal and the EU's Fit for 55 package, the Energy Efficiency Directive (EED) has been revised to include mandatory data center reporting on sustainability metrics.Effective from September 15, 2024, all data centers with an IT power demand of 500 kW or more must report:Total energy consumptionPower Usage Effectiveness (PUE)Water usageRenewable energy shareWaste heat reuseData center location and floor areaThis data is submitted to a central EU database and will be used to benchmark, rate, and enforce improvements over time.Why It’s HappeningData centers account for ~3% of the EU’s electricity consumption — a figure that’s rising fast due to AI, cloud, and edge deployments.Without regulation, this growth threatens the EU’s goal to be climate neutral by 2050.Policymakers are concerned about:Grid strainWater usage during heatwavesThe carbon footprint of hyperscale campusesWhat This Means for Data Center Operators in Europe1. Compliance Reporting Is MandatoryOperators must now track and disclose environmental performance. Failure to comply could result in:FinesDelayed permitsPublic ESG rating penalties2. Infrastructure Design Must ShiftTo meet aggressive targets, operators are adopting:Liquid and immersion cooling to reduce power and water usageOn-site solar and wind generationHeat recovery systems that feed district heating networks3. Data Centers Must Prove Efficiency, Not Just Offer ItMarketing a facility as “green” isn’t enough. The EU’s model is based on verifiable metrics, requiring:Metered energy consumptionTransparency in data handlingAnnual public reportingHow Hyperscalers Are ReactingGoogleExpanded carbon-free energy (CFE) tracking in Ireland and FinlandInvesting in geothermal and low-temperature heat recovery projectsMicrosoftDeveloping AI tools to forecast energy needs and optimize power usageTargeting 100% renewable energy matching, 24/7, in its European facilitiesAWSBuilding solar and wind farms in Spain, Italy, and GermanyDeploying new low-water cooling systems in Frankfurt and StockholmThese providers are using the EU as a test bed for global sustainability architecture — and influencing what’s next elsewhere.Implications for Colocation Providers1. Competitive Differentiation Will Be Regulatory-DrivenColo providers who can demonstrate compliance with:Low PUE (1.2)High renewable energy use (85%)Certified carbon offset plans...will have a clear edge in RFPs from sustainability-focused enterprises.2. Green SLAs and Certifications Will Become StandardExpect growing demand for:ISO 14001 and ISO 50001 certificationsEnergy Star or CEEDA ratingsFacility-level sustainability SLAs3. ESG Reporting Becomes a ServiceMore colocation operators will bundle:Carbon footprint dashboardsClient-specific sustainability metricsGreen procurement consultingThis moves ESG from a back-end compliance effort to a front-line selling point.Global Ripple EffectsThe EU’s model is already inspiring other governments:United StatesStates like California, Oregon, and Virginia are drafting sustainability mandates for large-scale digital infrastructure.SEC rules on ESG disclosures will force publicly traded colo and cloud firms to report carbon intensity.SingaporeRecently lifted a moratorium on new data centers, but only under strict carbon and energy conditions.United Arab EmiratesGreen building codes now apply to digital infrastructure, with incentives for renewables and thermal design.The EU’s leadership creates a policy blueprint that other regions will likely emulate or adapt.What Buyers and Tenants Should KnowIf you colocate infrastructure in the EU:Ask for proof of compliance with EED reportingRequest full environmental metrics in proposalsUnderstand whether your provider’s energy mix aligns with your ESG goalsIf you operate globally:Expect similar frameworks to emerge in North America and APACChoose partners who are proactive — not reactive — in sustainabilityFrom Greenwashing to Green AccountabilityThe EU isn’t just encouraging sustainability — it’s enforcing it. The new regulations mark the beginning of a global transition from green ambition to green accountability in digital infrastructure.For data center operators, this means faster innovation in cooling, energy sourcing, and environmental transparency. For buyers, it means putting ESG metrics next to uptime and price. And for the industry, it means acknowledging that the future of the cloud will depend on how we power it — and how responsibly we report it.

A Landmark Move in AI InfrastructureIn a bold and strategic move, Crusoe Energy Systems announced a massive $400 million deal to acquire cutting-edge graphics processing units (GPUs) from AMD. The announcement sent ripples through the AI infrastructure community, not just for its size, but for its implications. Crusoe, a company once primarily known for flaring mitigation and sustainable computing, is now taking center stage in the arms race for AI processing power.The move comes amid a global scramble for compute resources, as demand for artificial intelligence, machine learning, and high-performance computing (HPC) continues to surge. With competitors like NVIDIA dominating the AI chip space, Crusoe’s pivot to AMD’s Instinct MI300X accelerators signals both a strategic diversification and an aggressive growth plan.This article explores the many dimensions of this $400 million AMD deal — from the technology underpinning the purchase to the market impacts, the future of AI data centers, and Crusoe's ambitious roadmap.The Background — Why Compute Is the New OilOver the last five years, the phrase “compute is the new oil” has gained traction in the tech world. Just as oil powered the industrial revolution, computing power now fuels the digital revolution. AI models like OpenAI’s GPT-4o and Anthropic’s Claude 3 demand enormous computational resources. Training a frontier model requires thousands of GPUs running for weeks on end.This has resulted in unprecedented pressure on the semiconductor industry. Companies from Meta to Microsoft are pouring billions into infrastructure — not just in building data centers but also in securing the hardware needed to run large-scale AI workloads.Enter Crusoe. Originally focused on reducing natural gas flaring by converting stranded energy into compute power, the company has carved out a niche by pairing sustainable energy with digital workloads. Its early experiments with crypto mining have since evolved into a full-fledged AI compute strategy, and this AMD deal solidifies its evolution.Who Is Crusoe Energy? From Methane to Machine LearningCrusoe’s origin story is unlike any other AI infrastructure company. Founded in 2018, the company’s initial mission was environmental — to reduce methane emissions from oil and gas operations by deploying mobile data centers that consume otherwise wasted energy.But Crusoe quickly realized that stranded energy could become a competitive advantage in the compute market. It pivoted from just flaring reduction to offering cloud computing and high-performance infrastructure services.Its "Digital Flare Mitigation" (DFM) units — containerized data centers deployed in remote oilfields — represented a fusion of environmental technology and cloud computing. Now, it’s leveraging this unconventional infrastructure model to leapfrog into the AI era.The AMD purchase marks Crusoe's shift from niche player to potential hyperscaler contender.Inside the Deal — Why AMD, and Why Now?The heart of the deal lies in AMD’s Instinct MI300X accelerators, a new class of GPUs purpose-built for generative AI and HPC applications. Unlike earlier models, the MI300X offers:192GB of HBM3 memory (50% more than NVIDIA’s H100)Advanced ROCm software supportSuperior FP8 performance for generative AI trainingCompetitive pricing and scalabilityCrusoe's $400M order includes thousands of MI300X units, set to be deployed across its U.S. and international data centers by Q1 2026.Strategic Advantages for Crusoe:Supply Chain Diversification: NVIDIA chips are notoriously hard to procure. AMD offers a viable and powerful alternative.Cost Efficiency: AMD’s accelerators offer a better price-to-performance ratio in certain workloads.Energy Integration: MI300X units are being customized for Crusoe’s unique edge deployments, which rely on variable energy sources like flared gas and renewables.Scalability for AI-as-a-Service: With this deal, Crusoe is expected to expand its AI cloud platform, targeting enterprises and research institutions needing on-demand, eco-friendly compute.The AI Arms Race and Compute ScarcityThe AMD deal is best understood in the context of a larger industry crisis: compute scarcity. Demand for GPUs has far outpaced supply, leading to:Lengthy waitlists (sometimes 6–12 months)Strained relationships between cloud providers and chip vendorsExorbitant leasing costs on the secondary marketCrusoe is making a proactive bet — by securing AMD chips in bulk, it ensures priority access and sidesteps market volatility.Companies like CoreWeave, Lambda Labs, and Voltage Park have made similar moves, locking down inventory and building AI-dedicated data centers. Crusoe is now playing in the same league, but with a green-tech twist.Impacts on the AI EcosystemShort-Term EffectsIncreased Competition: Crusoe’s entry could pressure existing players, especially as it undercuts traditional cloud pricing.Acceleration of AMD's Market Share: As more hyperscalers warm to AMD, it could erode NVIDIA’s dominance in AI compute.Sustainable Compute as Differentiator: Crusoe’s environmentally-conscious model may appeal to ESG-minded enterprises.Long-Term PotentialOpen Source + AMD = Democratization: AMD’s ROCm platform aligns with open-source initiatives, potentially broadening developer adoption.Partnership Opportunities: Expect Crusoe to form alliances with model developers like Mistral, EleutherAI, and startups lacking their own infrastructure.Data Center Expansion and Infrastructure PlansCrusoe is reportedly planning to expand its data center footprint across Texas, North Dakota, and parts of Africa and Latin America — areas rich in stranded or underutilized energy.Next-Gen Data CentersModular, containerized deploymentsRenewable and flare-powered hybrid energy modelsLiquid cooling for high-density GPU clustersAI-first networking architectureGreen Compute as a Service (GCaaS)Crusoe plans to roll out a public-facing AI compute platform — a hybrid of Google Cloud and Lambda — but powered by waste energy. It's a bold play to combine decarbonization with commercial AI.Risks and ChallengesNo tech move is without risk. Crusoe will need to overcome:Software Stack Parity: ROCm still lags behind NVIDIA’s CUDA in some developer workflows.Ecosystem Lock-In: Many ML models are optimized for CUDA; porting can be difficult.Operational Complexity: Deploying and maintaining containerized, remote GPU clusters is logistically intense.Market Education: Convincing developers and enterprises to choose AMD over NVIDIA still requires outreach and tooling.But Crusoe is betting that the convergence of AI demand, energy waste, and sustainability pressures will tilt the market in its favor.What This Means for the Future of Data CentersCrusoe’s AMD deal is more than just a chip purchase — it’s a vision of the future:AI infrastructure won't just live in Ashburn or Silicon Valley. It’ll emerge in oilfields, deserts, and remote regions powered by solar or stranded gas.GPUs are the new utility. Just as electricity powered the 20th century, compute will define the 21st.Sustainable AI is no longer niche. It’s a core market requirement, especially as environmental regulations tighten.Crusoe is at the intersection of three megatrends: compute scarcity, climate responsibility, and AI acceleration.Crusoe’s AscentCrusoe’s $400M investment in AMD GPUs is both visionary and pragmatic. It reflects the tectonic shifts happening across AI infrastructure — where access to energy, chips, and innovative thinking define success.In a time when demand for AI compute is at an all-time high, Crusoe isn’t just buying hardware. It’s buying a seat at the table of the future of infrastructure.Whether the bet pays off remains to be seen. But one thing is certain: Crusoe is no longer an outsider. With AMD at its side and sustainability as its backbone, it’s now a force to be reckoned with in the global AI data center arms race.