# Amazon raises Spain AI data center commitment to $40 billion **TL;DR:** Amazon announced a **€33.7 billion (~$40 billion)** total investment in Spanish AI infrastructure at MWC Barcelona, adding **€18 billion** to its existing €15.7 billion commitment. Aragón becomes the centerpiece — hosting a new AI/ML server manufacturing and repair facility. The investment is projected to contribute **€31.7 billion to Spain's GDP** through 2035 and support **29,900 full-time equivalent jobs annually**. This is Amazon's single largest technology investment in any European country. It arrives as hyperscalers race to lock in European sovereign AI compute capacity ahead of regulatory tightening and regional demand acceleration. --- ## What you will learn 1. [The announcement: what Amazon committed and where](#the-announcement-what-amazon-committed-and-where) 2. [Aragón: why this region and what the facility does](#aragón-why-this-region-and-what-the-facility-does) 3. [The GDP and jobs math: €31.7B and 29,900 FTEs](#the-gdp-and-jobs-math-317b-and-29900-ftes) 4. [MWC Barcelona: why the timing matters](#mwc-barcelona-why-the-timing-matters) 5. [Europe's data center race: the hyperscaler scoreboard](#europes-data-center-race-the-hyperscaler-scoreboard) 6. [Spain's strategic positioning in European AI infrastructure](#spains-strategic-positioning-in-european-ai-infrastructure) 7. [Regulatory tailwinds and the EU AI Act factor](#regulatory-tailwinds-and-the-eu-ai-act-factor) 8. [What the €18B expansion actually buys](#what-the-18b-expansion-actually-buys) 9. [Risks: energy, water, and the Spain constraint problem](#risks-energy-water-and-the-spain-constraint-problem) 10. [What this means for European AI sovereignty](#what-this-means-for-european-ai-sovereignty) 11. [Frequently asked questions](#frequently-asked-questions) --- ## The announcement: what Amazon committed and where On March 3, 2026, at Mobile World Congress in Barcelona, Amazon announced it was raising its total AI infrastructure investment in Spain to **€33.7 billion** — approximately **$40 billion** at current exchange rates. The figure combines an existing **€15.7 billion** commitment with a new **€18 billion** addition announced at the event. The investment spans AWS cloud infrastructure, AI compute capacity, and a new Aragón-based manufacturing and repair facility for AI/ML servers. Amazon described the total commitment as its **largest technology investment in a single European country** in the company's history — a statement that reflects both the absolute scale and the competitive signaling intent of the announcement. At MWC, the venue itself mattered. Mobile World Congress is the world's largest telecommunications and technology trade show, attracting senior government officials, enterprise buyers, and international press. Making the announcement there gave Amazon a captive audience of European decision-makers at the precise moment when European AI policy, infrastructure sovereignty, and cloud procurement are all live political and commercial questions. The Spanish government's response was immediate and publicly enthusiastic. Spain has been actively competing to position itself as a preferred destination for hyperscaler infrastructure investment, offering a combination of EU membership, stable regulatory environment, and geographic advantages — southern European connectivity, access to African and Latin American cable routes, and renewable energy capacity that northern European alternatives often cannot match. The investment is structured as a multi-year commitment through 2035, which means the headline number is cumulative rather than immediate. But the commitment at this scale, made publicly at a major international forum, is not easily reversed — and the Aragón manufacturing facility announcement represents capital expenditure with a physical footprint that goes beyond cloud infrastructure spending. --- ## Aragón: why this region and what the facility does Aragón's selection as the site for Amazon's new AI/ML server manufacturing and repair facility is not coincidental. The region offers a combination of land availability, renewable energy infrastructure, skilled technical labor, and regional government incentive structures that made it competitive against alternatives in central Europe and Iberia. The new facility will handle two distinct functions: manufacturing of AI and machine learning server hardware, and repair and maintenance of existing server infrastructure. Both functions serve Amazon's broader goal of building vertically integrated AI infrastructure capacity in Europe. Instead of shipping hardware from Asia or North America for European data center deployments, the Aragón facility allows Amazon to source, test, and service hardware regionally — reducing supply chain lead times, tariff exposure, and logistical complexity for its European AI buildout. Aragón has been positioning itself as a data center hub for several years. The region already hosts data center capacity from multiple operators, and its government has invested in the power grid connections and road infrastructure that large data center campuses require. The Amazon facility adds a manufacturing dimension that goes beyond what most European data center investments include — it is not simply a facility that houses servers, but one that produces and services the hardware that other facilities run. The renewable energy picture is a significant factor. Spain generates roughly **60% of its electricity from renewable sources**, with wind and solar accounting for the majority. Aragón specifically has exceptional solar irradiance and wind resources, giving it structural cost advantages in energy procurement for power-intensive operations. Amazon has published corporate commitments to 100% renewable energy for its global operations — locating a major facility in a high-renewable region reduces the cost and complexity of meeting those commitments. For Spain, the Aragón facility represents something beyond a standard data center lease arrangement. Manufacturing and repair operations create skilled jobs with longer tenure and deeper local supply chain integration than pure data center operations. A data center can be scaled down or turned off relatively quickly; a manufacturing facility has tooling, workforce, and supplier relationships that create genuine economic roots. --- ## The GDP and jobs math: €31.7B and 29,900 FTEs Amazon's announcement included economic impact projections that deserve both amplification and scrutiny: a projected **€31.7 billion contribution to Spain's GDP** through 2035, and **29,900 full-time equivalent jobs** supported annually. These figures are Amazon's own projections, not independent analyses — a distinction worth keeping in mind when evaluating them. Large capital investment announcements routinely include economic impact estimates that make favorable assumptions about multiplier effects, supply chain linkages, and induced economic activity. The actual economic impact will depend heavily on implementation details: how much of the supply chain is locally sourced versus imported, what the wage levels are for the direct jobs created, and how much of the GDP contribution is genuinely additive versus displacement of other economic activity. With that caveat noted, the order of magnitude is consistent with what comparable infrastructure investments generate. Data center and cloud infrastructure spending tends to have strong local multiplier effects because the facilities require continuous inputs — power, cooling, maintenance, security, catering, facilities management — that are inherently local. The manufacturing component of the Aragón facility adds further local procurement linkages that a pure data center operation would not. The **29,900 FTE figure** is an annual ongoing number, not a construction-phase peak. Construction of facilities at this scale typically employs several times more people during the build phase than during steady-state operations, so the total employment impact over the investment period is higher than the annual figure suggests. For context, Amazon's entire Spanish workforce in 2024 was estimated at roughly 25,000 — this investment is projected to support more annual FTE equivalents (counting indirect and induced employment) than Amazon's direct headcount in the country. The **€31.7 billion GDP figure** through 2035 represents roughly **€3.2 billion per year** on average — approximately **0.2% of Spain's current annual GDP**. That is a material contribution for a single corporate investment, particularly one concentrated in a single region. --- ## MWC Barcelona: why the timing matters Amazon's choice to announce at MWC Barcelona on March 3, 2026, rather than in a press release or at AWS re:Invent, was a calculated decision about audience and message. MWC is where European telecommunications executives, government ministers, enterprise technology buyers, and infrastructure investors gather in one place for one week. Making a €33.7 billion commitment announcement at this venue means Amazon is speaking simultaneously to: potential enterprise customers who make cloud procurement decisions, government officials who shape regulatory and tax environments, infrastructure partners who supply data center components, and the press that amplifies to a European business audience. The announcement also lands at a specific geopolitical moment. The European Union is actively developing AI policy under the EU AI Act, negotiating data localization requirements, and experiencing growing political pressure to ensure that European AI infrastructure is not entirely dependent on U.S. hyperscalers. An announcement of this scale — from the dominant global cloud provider — serves as a demonstration that Amazon intends to be embedded deeply in European AI infrastructure rather than serving it from a distance. The MWC timing also puts Amazon in conversation with its competitors. Microsoft, Google, and Meta have all made major European data center announcements in the past eighteen months. By making a headline-grabbing commitment at the region's most visible tech event, Amazon signals that it is not conceding the European AI infrastructure market to rivals who have been more publicly aggressive in the region. For Spain specifically, hosting the announcement at a conference physically located in Barcelona — Spain's most internationally prominent business city — is a domestic political win for the Spanish government, which has actively courted this investment and can now point to MWC as the stage where it came to fruition. --- ## Europe's data center race: the hyperscaler scoreboard The Amazon Spain announcement is one move in a broader hyperscaler race to lock in European AI compute capacity. The scoreboard as of early 2026 is aggressive across all major players. **Microsoft** committed **€4.3 billion** to cloud and AI infrastructure in France in early 2025, followed by similar announcements in Germany, the Netherlands, and the UK. Microsoft's total European investment across 2025–2026 commitments exceeded **€20 billion** by the end of 2025. **Google** announced a **€1 billion** expansion of its Finnish data center operations in 2025, plus new investments in Ireland, Belgium, and the Netherlands. Google's total European cloud infrastructure commitment for the 2025–2026 period reached approximately **$15 billion**. **Meta** committed **$800 million** to a new data center in Spain — specifically in Talavera de la Reina — in 2025, its first owned data center in Spain and part of a broader European infrastructure expansion. **Oracle** announced a **$6.5 billion** investment in Japanese AI infrastructure in early 2026, signaling that the same dynamics driving European investment are playing out globally. Against this backdrop, Amazon's **€33.7 billion** total Spain commitment is the single largest national commitment any hyperscaler has made in Europe. It is not just a data point in the race — it is an attempt to redefine the scale at which the race is being run. By making a commitment that is roughly **twice the next-largest comparable announcement**, Amazon is setting a new benchmark that competitors must respond to or acknowledge they cannot match. The race has a strategic logic beyond market share. European enterprises — banks, manufacturers, healthcare systems, governments — are increasingly required or politically motivated to process their most sensitive data within European borders. Building European AI compute capacity is the prerequisite for serving this market segment. The hyperscaler that has the most European capacity when these procurement decisions accelerate will win the contracts. --- ## Spain's strategic positioning in European AI infrastructure Spain's emergence as a priority destination for hyperscaler AI infrastructure investment is not accidental. The country has built a specific value proposition over the past several years that addresses the requirements of large-scale data center operations in ways that northern European alternatives often cannot. **Renewable energy capacity.** Spain generated approximately **60% renewable electricity** in 2025, with exceptional solar and wind resources across the country. For hyperscalers with corporate renewable energy commitments, Spain offers high-quality Power Purchase Agreement (PPA) opportunities at scale — something that data-center-dense markets like Ireland and the Netherlands are increasingly struggling to provide as their renewable capacity is oversubscribed. **Cable connectivity.** Spain's geographic position — at the corner of Europe where Atlantic and Mediterranean routes converge — makes it a natural termination point for transatlantic and Africa-facing submarine cables. The MAREA cable (Microsoft and Meta), the Amitié cable (Facebook, Microsoft), and the 2Africa cable all connect to or through Spanish infrastructure. This gives Spanish data centers connectivity advantages for content serving to Latin America and Africa that no other European country can match. **Land and cost.** Inland Spain, including Aragón, offers significantly lower land and construction costs than the Netherlands, Germany, or the UK. For facilities at Amazon's planned scale, cost per megawatt of deployed capacity is a meaningful number. **Regulatory stability.** Spain's position within the EU provides the legal framework that enterprise customers require for data processing, while the country's own regulatory approach to data centers has been more permissive than some northern European alternatives on specific issues like water use and local grid connection. The combination creates a differentiated offer. Spain is not the cheapest European option or the most connected, but it is the best intersection of renewable energy quality, cable connectivity, land availability, and EU regulatory standing of any Southern European country — and in some dimensions, of any European country. --- ## Regulatory tailwinds and the EU AI Act factor The EU AI Act, which came into force in August 2024 with phased compliance deadlines through 2027, is reshaping where and how AI workloads are processed in Europe. The Act's requirements around high-risk AI systems — which include AI used in employment, credit, education, law enforcement, and critical infrastructure — specify documentation, audit, and governance requirements that are substantially easier to satisfy when the underlying compute infrastructure is within EU jurisdiction. For Amazon Web Services, this creates a strategic imperative: European enterprises subject to the AI Act need EU-based compute for their highest-stakes AI workloads. AWS can serve this need with its existing European regions, but the volume of EU AI Act-compliant workloads is projected to grow substantially as compliance deadlines arrive and enterprises move from planning to deployment. The capacity to serve that demand requires investment now. The **€33.7 billion** Amazon commitment can be read, in part, as a bet that EU AI Act-driven demand for European AI compute will be substantial and durable. The company is not investing at this scale based on current European AI cloud revenue — it is investing based on projected demand from the compliance-driven migration of AI workloads into EU jurisdiction over the 2025–2030 period. The Spanish government has also demonstrated regulatory pragmatism that matters for infrastructure investment. Spain's Ministry of Digital Transformation has been actively engaged in facilitating major tech investments, including streamlining environmental approval processes for data centers in priority regions. Aragón's regional government has additional incentive structures for major industrial investments that reduce the effective cost of the facility Amazon is building. --- ## What the €18B expansion actually buys The new **€18 billion** addition to Amazon's Spain commitment is not a single project. It represents a multi-year buildout across several capability areas that together constitute a step-change in Amazon's European AI infrastructure footprint. **Additional data center capacity.** The bulk of large-scale cloud investment goes to physical data center buildings, power infrastructure, cooling systems, and the network connectivity that ties them together. At current industry costs — roughly **$10–15 million per megawatt** for hyperscale data center capacity — €18 billion funds between **1,000 and 1,500 megawatts** of new AI compute capacity. That is a significant fraction of Spain's entire current data center capacity being added by a single operator. **AI/ML server manufacturing facility in Aragón.** This is the component of the announcement that goes beyond standard data center investment. A manufacturing and repair facility for AI servers represents vertical integration — bringing hardware production closer to deployment. The facility reduces Amazon's dependence on Asian manufacturing supply chains for European AI infrastructure and gives the company more control over hardware customization, testing cycles, and repair turnaround times. **Network infrastructure.** Expanding data center capacity without commensurate network investment would create bottlenecks. The €18 billion commitment will include investment in backbone connectivity between Spanish AWS regions, edge infrastructure that routes Spanish and regional traffic efficiently, and likely contributions to new submarine cable capacity. **Power infrastructure.** Large data center expansions require dedicated power agreements and often direct investment in grid connection infrastructure. Amazon has historically invested in wind and solar capacity through PPAs and direct ownership. The Spanish commitment likely includes renewable energy procurement at scale that will flow into Aragón and other Spanish regions' clean energy buildout. The Aragón manufacturing facility is the strategically novel element. Cloud investment announcements are common. Announcements of owned manufacturing capacity for AI hardware in Europe are not — this positions Amazon to control a supply chain component that its competitors are typically procuring through Asian contract manufacturers. --- ## Risks: energy, water, and the Spain constraint problem A commitment of this scale is not risk-free, and the risks in Spain are specific and well-documented. **Energy grid constraints.** Spain's renewable energy capacity is abundant, but its transmission grid — particularly in inland regions like Aragón — was not designed to handle the load profiles that hyperscale data center clusters generate. Grid connection timelines for major industrial facilities in Spain have historically run **18–36 months** from application to energization. If Amazon's buildout timelines assume faster connections, delays will compress the return profile of the investment. **Water use in water-stressed regions.** Data center cooling consumes significant water, and Aragón, like much of inland Spain, faces periodic drought conditions. The Ebro River basin — Aragón's primary water source — is under increasing pressure from agricultural, industrial, and municipal demand. Local communities and environmental regulators have raised concerns about large data center water use in water-stressed Spanish regions. Amazon will need to demonstrate either water-efficient cooling approaches or credible water sourcing strategies that do not draw down stressed regional supplies. **Construction cost and timeline risk.** At €33.7 billion total, even a 10% cost overrun represents **€3.4 billion** in additional capital. Global construction cost inflation for data center projects — driven by structural steel prices, electrical component shortages, and specialized labor scarcity — has been running above general inflation rates. The multi-year timeline of the commitment provides some exposure to cost conditions that are currently elevated and uncertain. **Political risk.** Spain's political environment — coalition government, regional tensions, and active debate about the terms under which foreign infrastructure investment is welcomed — creates execution risk. A commitment made under one government's incentive framework can become more complicated if the political composition of the Spanish or Aragón government changes during the implementation period. None of these risks are dealbreakers at the scale Amazon is operating. The company has navigated similar constraints in every major infrastructure buildout globally. But they are real factors that will affect the pace and cost of realization. --- ## What this means for European AI sovereignty The Amazon Spain commitment is the largest single event yet in the reshaping of European AI infrastructure — and it has implications that go beyond one company's capital allocation. **Concentration of AI compute is accelerating.** As Amazon, Microsoft, Google, and Meta each make multi-billion-euro commitments to European data center capacity, the European AI infrastructure landscape is consolidating around a small number of hyperscale operators. This concentration has benefits: hyperscalers bring world-class operational expertise, capital efficiency, and security practices. It has risks: European enterprises become deeply dependent on the infrastructure terms, pricing, and policy decisions of U.S. companies. **The sovereignty gap persists.** Despite the scale of investment, European-owned AI infrastructure — hyperscale compute capacity owned and operated by European companies under European control — remains a small fraction of total capacity. Initiatives like GAIA-X, which aimed to develop a European cloud infrastructure ecosystem, have not produced hyperscale-competitive operators. The Amazon investment makes Europe's AI compute capability more powerful but not more sovereign. **Spain wins the location competition, at least for now.** For European enterprises evaluating where to place regulated AI workloads, the concentration of Amazon's commitment in Spain — combined with Microsoft and Meta investments in the country — makes Spanish data centers an increasingly credible primary option rather than a secondary market. The network effects of infrastructure concentration tend to be self-reinforcing: more capacity attracts more customers attracts more investment. **The investment baseline has moved.** A single hyperscaler committing €33.7 billion to a single European country sets a new floor for what "serious" European AI infrastructure investment looks like. Competitors who want to credibly claim European AI infrastructure leadership will need to respond at comparable scale — which means the total European AI infrastructure buildout trajectory has just been revised upward. For enterprises planning AI infrastructure strategy in Europe, the Amazon Spain commitment is a significant data point. It confirms that high-quality, EU-compliant, hyperscale AI compute will be available in Southern Europe at scale through 2035 — and that Amazon is betting heavily that European demand will justify the investment. --- ## Frequently asked questions ### What is Amazon's total investment in Spain? Amazon has committed a total of **€33.7 billion (~$40 billion)** to AI and cloud infrastructure in Spain. This consists of an existing **€15.7 billion** commitment and a new **€18 billion** addition announced at MWC Barcelona on March 3, 2026. ### What is being built in Aragón specifically? Aragón is the site of a new **AI/ML server manufacturing and repair facility** — a hardware production and servicing operation that will support Amazon's European data center deployments. This goes beyond a standard data center; it is a manufacturing facility that produces and services the server hardware that data centers run. ### How many jobs does the investment create? Amazon projects the investment will support **29,900 full-time equivalent jobs annually** across direct, indirect, and induced employment. The figure covers jobs at Amazon facilities, across the supply chain, and in the broader economy through spending by Amazon employees. Construction-phase employment during the buildout period will be higher than this steady-state annual figure. ### What is the projected economic impact? Amazon projects the investment will contribute **€31.7 billion to Spain's GDP** through 2035 — an average of approximately **€3.2 billion per year**. These are Amazon's own projections based on economic multiplier models; actual impact will depend on implementation details including local supply chain sourcing and wage levels. ### Why Spain specifically? Spain offers a combination of high renewable energy capacity (~60% of electricity from renewables), favorable submarine cable connectivity for transatlantic and Africa-facing routes, lower land and construction costs versus northern European alternatives, and a regulatory environment that has been actively accommodating to large infrastructure investments. The EU membership provides the legal framework enterprise customers require for regulated AI workloads. ### Is this Amazon's largest European investment? Yes. Amazon described the €33.7 billion total Spain commitment as its **largest technology investment in a single European country** in the company's history. ### When will the investment be completed? The investment is structured as a multi-year commitment through **2035**. Individual components — data center expansions, the Aragón facility, network infrastructure — will be built on separate timelines within that window. No single completion date applies to the full commitment.