Nscale Analysis 2025: How Liquid Cooling Moved from Niche to Necessity in AI Data Centers
Industry Adoption: Nscale’s Liquid Cooling Strategy from Niche Bet to Hyperscaler Prerequisite
Between 2021 and 2024, Nscale emerged as a specialized AI hyperscaler with a distinct, forward-looking thesis: the future of high-performance computing (HPC) would be defined by liquid cooling. During this foundational period, the company launched from stealth mode, establishing its vertically integrated model by building a GPU cloud service on AMD Instinct MI300X accelerators and acquiring Kontena, a modular data center specialist. This strategy was an early bet that traditional air cooling was fundamentally inadequate for the next wave of AI hardware. The company’s initial projects, like the N1 data center in Glomfjord, Norway, served as commercial proof points, integrating partner technologies like Lenovo’s Neptune™ liquid cooling to demonstrate a co-designed, full-stack solution. This period was characterized by building capabilities and validating the technical and economic viability of a liquid-cooling-first approach, positioning Nscale as a niche provider for demanding AI workloads.
The landscape shifted dramatically from January 2025 to the present day, marking an inflection point where Nscale’s niche strategy became a mainstream necessity for the world’s largest technology companies. This transition was catalyzed by the arrival of power-intensive hardware like NVIDIA’s Blackwell platform. Nscale moved from conceptual validation to massive-scale execution, underscored by a landmark, multi-billion dollar deal to deploy approximately 200,000 liquid-cooled GPUs for Microsoft. Further validation came from a joint venture with Aker and OpenAI for the “Stargate Norway” AI gigafactory, a 230MW project powered by renewable energy. The variety of applications expanded from general AI cloud services to specific, large-scale deployments for cornerstone tenants. This demonstrates that liquid cooling is no longer an alternative but a foundational requirement for deploying next-generation AI at scale. The new opportunity lies in becoming the go-to infrastructure partner for hyperscalers who cannot build or retrofit specialized liquid-cooled facilities fast enough, while the primary threat becomes the immense execution risk associated with delivering these gigawatt-scale projects on time and on budget.
Table: Nscale’s Capital Mobilization for Liquid-Cooled AI Infrastructure
| Funding Round / Investment | Time Frame | Details and Strategic Purpose | Source |
|---|---|---|---|
| Series B & Pre-Series C SAFE | Sep 2025 | Raised a landmark $1.1 billion in a Series B round backed by Aker, Nokia, and Nvidia, followed by an additional $433 million in a pre-Series C SAFE. The capital is for global expansion of its liquid-cooled AI infrastructure. | HPCwire |
| Stargate Norway Phase 1 Investment | Jul 2025 | Committed to an initial investment of approximately $1 billion (including ~$250M in equity) for the first 20MW phase of the “Stargate Norway” AI data center, designed with closed-loop, direct-to-chip liquid cooling. | PR Newswire |
| UK Data Centre Investment | Jan 2025 | Announced a £2 billion ($2.5 billion) investment plan over three years for the UK data center market, starting with a 50MW (scalable to 90MW) liquid-cooled site in Loughton, Essex. | GlobeNewswire |
| Series A Funding Round | Dec 2024 | Secured $155 million in a Series A round led by Sandton Capital Partners to expand its AI-ready data center footprint across Europe and North America and build out a 1.3 GW pipeline of sustainable sites. | Nscale |
| Acquisition of Kontena | Jul 2024 | Acquired Kontena, a specialist in high-density modular data centers and AI solutions, to enhance in-house capabilities for designing and deploying HPC and AI infrastructure with integrated liquid cooling. | Business Wire |
Table: Nscale’s Strategic Partnerships for AI Infrastructure Dominance
| Partner / Project | Time Frame | Details and Strategic Purpose | Source |
|---|---|---|---|
| Verne | Nov 2025 | Agreement to deploy 4,600 NVIDIA Blackwell Ultra GPUs at Verne’s Iceland campus. The 15MW project will use a hybrid system that is 85% liquid-cooled, leveraging Verne’s renewable energy infrastructure. | Data Center Dynamics |
| Nokia | Nov 2025 | Partnership to enhance data center performance and automation. Nokia will provide technology, including switches in liquid-cooled variants, to deliver operational advantages for Nscale’s infrastructure. | Nokia |
| VAST Data | Oct 2025 | Selected VAST Data’s AI platform to provide storage solutions for its large-scale, liquid-cooled GPU deployments, supporting major clients like Microsoft and OpenAI. | Data Center Dynamics |
| Microsoft & Start Campus | Oct 2025 | Expanded agreement with Microsoft for ~200,000 GPUs. A portion involves deploying 12,600 NVIDIA Blackwell GPUs at Start Campus’s site in Portugal, which supports rack densities over 130kW using advanced liquid cooling. | SiliconANGLE |
| Aker and OpenAI | Jul 2025 | Formed a joint venture to establish “Stargate Norway,” a 230MW AI data center powered by renewable energy. The facility will feature closed-loop, direct-to-chip liquid cooling and deploy 100,000 GPUs. | Nscale |
| InfraPartners | Mar 2025 | Partnership to expand the Glomfjord data center in Norway, adding a new 30MW liquid-cooled facility to bring the site’s total capacity to 60MW. The expansion is set to go live in Q2 2025. | Nscale |
| Lenovo, Nokia, & AMD | Nov 2024 | Collaboration to provide a turnkey AI solution. Lenovo contributes its Neptune™ liquid cooling, Nokia its data fabric, and AMD its Instinct MI300X GPUs, forming the technological backbone of Nscale’s initial offering. | Lenovo |
Geography: Nscale’s Nordic-Centric Strategy Goes Global
Between 2021 and 2024, Nscale’s geographic footprint was strategically concentrated in the Nordics, specifically Norway. The selection of its initial 60MW site in Glomfjord was a calculated move to leverage the region’s abundance of low-cost, 100% renewable hydropower and naturally cool climate. This provided an ideal testbed to prove the operational and economic efficiency of its liquid-cooled, sustainable data center model. This early-stage, focused approach allowed the company to control variables and build a strong operational case study before pursuing broader expansion. The acquisition of Finland-based Kontena in July 2024 further solidified this Nordic focus, bringing in regional expertise in modular, high-density data center construction.
From 2025 onwards, Nscale’s geographic strategy has exploded from a regional focus to a multi-continental expansion, driven by massive anchor-tenant agreements. While Norway remains a strategic hub with the 230MW “Stargate Norway” project and the Glomfjord expansion, Nscale has aggressively entered new markets. Key projects now include a 15MW deployment in Iceland with Verne (leveraging geothermal and hydro), a major deployment in Portugal with Start Campus for Microsoft (using cold ocean water for cooling), and a planned £2 billion investment in the UK, starting with a 90MW-capable site in Loughton. This expansion also includes plans for new data centers in the United States as part of the expanded Microsoft deal. This geographic diversification shows that Nscale is following hyperscaler demand into regions that offer a trifecta of renewable energy, favorable climates, and supportive infrastructure, transforming its Nordic-first model into a global blueprint for sustainable AI.
Technology Maturity: Nscale’s Liquid Cooling Evolves from Integrated Feature to Core Enabler
In the 2021–2024 period, Nscale’s technology was in the early commercialization and integration phase. The company’s primary innovation was not inventing liquid cooling but vertically integrating it as a standard feature within a full-stack AI cloud platform. The launch of its GPU cloud service in May 2024, utilizing AMD MI300X accelerators and partner tech like Lenovo’s Neptune™ liquid cooling, was a key validation point. The acquisition of Kontena brought in modular “Edge data room” solutions and the “KONNECT ICE” containerized cooling unit, demonstrating a focus on creating scalable, replicable, and rapidly deployable infrastructure. The technology was commercially available but at a limited scale, with the Glomfjord data center acting as the main showcase. The focus was on proving that a co-designed, liquid-cooled stack could deliver superior performance and efficiency.
The period from 2025 to today marks the transition of Nscale’s liquid cooling technology from a commercial offering to a scaled, mission-critical enabler. The technology is no longer just an option but the fundamental prerequisite for deploying next-generation hardware like the NVIDIA GB200 NVL72 and Blackwell Ultra GPUs, which Nscale now offers. These systems are engineered to function as single massive GPUs, and their thermal output makes traditional air cooling non-viable. Projects like the Verne deployment in Iceland (85% liquid-cooled), the Start Campus facility in Portugal (supporting >130kW racks), and the “Stargate Norway” gigafactory all explicitly call for closed-loop, direct-to-chip liquid cooling. This shift from deploying thousands of GPUs to deploying hundreds of thousands for clients like Microsoft and OpenAI validates that Nscale’s specialized technology has achieved scaled commercial maturity and is now defining the new standard for AI data center design.
Table: SWOT Analysis of Nscale’s Liquid Cooling Strategy
| SWOT Category | 2021 – 2024 | 2025 – Today | What Changed / Resolved / Validated |
|---|---|---|---|
| Strengths | Vertically integrated model focused on liquid cooling. Strategic acquisition of Kontena for modular deployment. Foundational partnerships with tech leaders (Lenovo, AMD, Nokia). | Massive funding over $1.5B (Series B, SAFE). Landmark anchor contracts with hyperscalers (Microsoft, OpenAI). Greenfield designs optimized for >130kW rack densities and NVIDIA Blackwell GPUs. | The market validated Nscale’s liquid-cooling-first thesis. The model shifted from a strategic bet to an industry prerequisite, attracting premier partners and massive capital, proving its competitive edge over retrofits. |
| Weaknesses | Capital-intensive model with only Series A funding ($155M). Limited operational track record, having just launched from stealth. Dependency on partners for core technology (e.g., Lenovo’s Neptune). | Immense execution risk on multiple large-scale, capital-intensive projects. High dependency on a few key customers (Microsoft, OpenAI) for revenue. Reliance on NVIDIA for next-gen GPU supply. | While funding de-risked the capital-intensive model, the scale of new projects introduces significant execution risk. The customer base has consolidated around a few massive but critical partners. |
| Opportunities | Growing demand for AI compute exceeding capabilities of air-cooled data centers. First-mover advantage in building purpose-built, sustainable AI infrastructure in energy-rich regions (Norway). | Explosive demand for generative AI requiring next-gen, liquid-cooled hardware (NVIDIA Blackwell). Hyperscalers need specialized partners like Nscale to meet infrastructure goals. ESG focus drives demand for sustainable, efficient data centers. | The market opportunity fully materialized. The insufficiency of air cooling became an acute bottleneck, and Nscale was perfectly positioned with a ready-made, scalable solution that major AI players desperately needed. |
| Threats | Competition from legacy hyperscalers (AWS, Google) potentially retrofitting their own facilities. Supply chain constraints on high-performance GPUs. | Competitors rushing to build their own greenfield liquid-cooled facilities. Potential for construction delays or energy procurement issues on gigawatt-scale projects. Geopolitical risks impacting global supply chains. | The threat shifted from established players adapting to new competitors mimicking Nscale’s successful model. The primary threat is no longer strategic but operational: the ability to execute faster than the competition. |
Forward-Looking Insights: Nscale’s Year Ahead Is All About Execution
The data from 2025 paints a clear picture: Nscale has successfully transitioned from a promising startup to a critical pillar of the global AI infrastructure build-out. The year ahead will be defined not by strategy, but by execution. The market has validated the company’s liquid-cooling-first approach through over $1.5 billion in funding and cornerstone contracts with Microsoft and OpenAI. Now, the focus shifts to delivering on these monumental promises.
Three key signals will be critical to watch. First, the operational launch of the 30MW Glomfjord expansion in Q2 2025 will be the first major test of Nscale’s accelerated deployment capabilities. Second, progress on the “Stargate Norway” project, with its ambitious goal of deploying 100,000 NVIDIA GPUs by the end of 2026, will serve as a barometer for the company’s ability to manage gigawatt-scale construction. Finally, the phased deployment of 4,600 Blackwell GPUs in Iceland will provide a real-world proof point for its hybrid cooling model and its ability to integrate into partner-operated facilities. Nscale has created a new standard where advanced liquid cooling is the foundation of AI infrastructure; its challenge now is to build that future on time and at an unprecedented scale. Any delays will create an opening for fast-following competitors, but successful and timely execution will solidify its position as an indispensable leader in the AI arms race.
Frequently Asked Questions
What is Nscale’s core business strategy?
Nscale is a specialized AI hyperscaler focused on building and operating vertically integrated, liquid-cooled data centers. Its strategy is to provide purpose-built infrastructure for next-generation AI hardware that traditional air-cooled facilities cannot support. This has made them an essential partner for hyperscalers like Microsoft and OpenAI who need to deploy power-intensive GPUs at a massive scale.
Why has liquid cooling become a necessity for AI data centers?
Liquid cooling is now a foundational requirement because the latest generation of AI accelerators, like NVIDIA’s Blackwell platform, are so power-intensive and generate immense heat. Traditional air cooling is fundamentally inadequate for these systems, which can create rack densities exceeding 130kW. As a result, advanced methods like direct-to-chip liquid cooling are the only viable way to operate this hardware efficiently and at scale.
Who are Nscale’s main customers and partners?
Nscale’s primary customers are major technology companies and hyperscalers. The article highlights a landmark deal to deploy approximately 200,000 GPUs for Microsoft and a joint venture with Aker and OpenAI for the “Stargate Norway” AI gigafactory. Its key technology partners include NVIDIA and AMD for GPUs, Lenovo for its Neptune™ liquid cooling systems, Nokia for data center networking, and VAST Data for storage solutions.
How is Nscale funding its large-scale infrastructure projects?
Nscale has mobilized significant capital through several funding rounds. Key investments listed in the article include a $155 million Series A in late 2024, followed by a landmark $1.1 billion Series B and a $433 million SAFE agreement in September 2025. This capital, raised from investors like Aker, Nokia, and Nvidia, is designated for the global expansion of its liquid-cooled AI infrastructure, including major projects in the UK, Norway, and for clients like Microsoft.
Where is Nscale building its data centers?
Nscale began with a Nordic-centric strategy, establishing its first sites in Norway to take advantage of renewable energy and the cool climate. Since 2025, driven by major contracts, the company has expanded globally. Key projects are now located in Iceland, Portugal, and the United Kingdom, with further expansion planned in the United States. Its strategy involves selecting locations that offer a combination of renewable energy, supportive infrastructure, and favorable climates for cooling.
Experience In-Depth, Real-Time Analysis
For just $200/year (not $200/hour). Stop wasting time with alternatives:
- Consultancies take weeks and cost thousands.
- ChatGPT and Perplexity lack depth.
- Googling wastes hours with scattered results.
Enki delivers fresh, evidence-based insights covering your market, your customers, and your competitors.
Trusted by Fortune 500 teams. Market-specific intelligence.
Explore Your Market →One-week free trial. Cancel anytime.
Related Articles
If you found this article helpful, you might also enjoy these related articles that dive deeper into similar topics and provide further insights.
- E-Methanol Market Analysis: Growth, Confidence, and Market Reality(2023-2025)
- Carbon Engineering & DAC Market Trends 2025: Analysis
- Climeworks 2025: DAC Market Analysis & Future Outlook
- Climeworks- From Breakout Growth to Operational Crossroads
- Battery Storage Market Analysis: Growth, Confidence, and Market Reality(2023-2025)
Erhan Eren
Ready to uncover market signals like these in your own clean tech niche?
Let Enki Research Assistant do the heavy lifting.
Whether you’re tracking hydrogen, fuel cells, CCUS, or next-gen batteries—Enki delivers tailored insights from global project data, fast.
Email erhan@enkiai.com for your one-week trial.

