Appearance
Google Invests $15B+ in Finland Nuclear AI Data Centers
In a landmark announcement reshaping the European AI infrastructure landscape, Google has committed more than $15 billion to expand its data center footprint in Finland, paired with an unprecedented 22-year nuclear power supply agreement with Finnish energy company Fortum. The deal, unveiled in September 2026, marks one of the largest corporate nuclear power contracts in history and underscores the escalating energy demands of artificial intelligence training and inference workloads. Google's investment in Finland represents a strategic bet on Nordic nuclear reliability as AI compute hunger continues to double roughly every six months. The partnership with Fortum's Loviisa nuclear power plant will deliver baseload carbon-free electricity to power Google's expanding fleet of AI-optimized data centers across the country.
Background
- Google first established a data center presence in Finland in 2009 with its Hamina facility, leveraging the country's cold climate and abundant renewable energy
- The global AI boom has pushed data center power demand to record levels, with major tech firms seeking stable, carbon-free baseload power beyond what wind and solar alone can provide
- Finland's Loviisa plant, operated by Fortum, consists of two VVER-440 pressurized water reactors with a combined capacity of approximately 1,000 megawatts, both recently granted lifetime extensions through the 2050s
Key facts
| Item | Detail |
|---|---|
| Total Google investment | $15 billion+ in Finnish AI infrastructure |
| Nuclear contract duration | 22 years with Fortum |
| Power plant | Fortum Loviisa Nuclear Power Plant |
| Plant location | Loviisa, Finland (Gulf of Finland coast) |
| Reactor capacity | ~1,000 MW combined (two VVER-440 units) |
| Existing Google facilities | Hamina data center (operational since 2011) |
| New facilities planned | Multiple AI-optimized data center campuses |
| Carbon impact | 100% nuclear baseload, zero operational emissions |
| Jobs created | Estimated 3,000+ construction and 800+ permanent roles |
Highlights
The 22-Year Nuclear Power Purchase Agreement
The 22-year power purchase agreement between Google and Fortum represents a paradigm shift in how hyperscale tech companies approach AI energy procurement. Unlike typical corporate renewable energy deals that run for 10 to 15 years, this contract spans more than two decades, locking in predictable pricing for carbon-free baseload power at a scale sufficient to power multiple AI-optimized data center campuses. The deal allocates a substantial portion of Loviisa's output directly to Google's Finnish operations, providing the round-the-clock reliability that AI training clusters require. Unlike intermittent wind and solar resources, nuclear power delivers steady output independent of weather conditions or time of day, making it uniquely suited for the continuous high-density compute demands of large language model training.
Why Finland and Why Nuclear?
Finland has emerged as a prime destination for AI data center investment thanks to its cold climate, robust grid infrastructure, political stability, and increasingly competitive nuclear energy landscape. The country currently operates four commercial nuclear reactors with two more under construction at the Olkiluoto site and plans for additional next-generation reactors in the 2030s. For Google, the combination of Nordic cooling advantages and nuclear baseload power creates a compelling total cost of ownership profile for AI infrastructure. Cooling represents 30 to 40 percent of data center energy overhead in warmer climates, a cost dramatically reduced in Finland's subarctic environment. The Fortum deal also aligns with Google's 24/7 carbon-free energy goal, which targets every data center running on carbon-free power at every hour of every day by 2030.
Industry positioning & impact
Google's massive Finnish investment and nuclear deal send reverberations across the technology, energy, and geopolitical landscapes, positioning the company as a leader in the race to secure AI compute capacity in Europe. The move comes amid intensifying competition among hyperscalers -- including Microsoft, Amazon, and Meta -- all scrambling to lock in energy supplies for their AI infrastructure. By securing 22 years of nuclear baseload, Google gains a multi-decade competitive advantage in energy pricing and reliability that wind and solar PPAs alone cannot match.
The deal also signals a broader industry pivot toward nuclear power as a critical enabler of AI scaling. Industry analysts estimate that a single large AI training cluster can consume as much electricity as a mid-sized city, and traditional renewable energy paired with battery storage cannot economically provide the 24/7 high-density power these clusters demand. As AI models grow from trillions to quadrillions of parameters, the power requirements are projected to increase tenfold or more by the end of the decade.
For Finland, the investment cements its position as Europe's leading AI infrastructure hub, attracting billions in foreign direct investment and creating thousands of high-skilled jobs. The Fortum partnership also provides stable long-term revenue for the Finnish utility, supporting continued reactor operations and potentially financing next-generation SMR (small modular reactor) deployments. European policymakers are watching the deal closely as a potential model for balancing AI growth with climate goals, though it also raises questions about whether Europe's nuclear fleet can expand quickly enough to meet projected demand.
Related reading
For more on the global AI infrastructure arms race, see our analysis of hyperscaler data center expansion trends and the rising demand for specialized AI compute hardware. Our previous coverage of Microsoft's nuclear energy partnerships explores how competing tech giants are approaching the energy challenge differently. Readers interested in the European regulatory angle should review our deep dive into the EU's AI Act and its implications for data center infrastructure investments across the bloc.
References
Google's official announcement of the Finland investment and Fortum partnership was published on the Google Cloud blog, detailing the scope of infrastructure expansion and sustainability commitments. Fortum's press release on the Loviisa power agreement provides technical specifications and contract terms. The International Energy Agency's recent report on data center electricity demand offers global context on the AI-driven power consumption surge. World Nuclear Association data on Finland's reactor fleet and lifetime extension programs provides independent verification of Loviisa's operational status and capacity.
Buying advice & audience
For enterprise AI teams evaluating cloud infrastructure providers, the Google-Fortum nuclear deal is a significant differentiator worth factoring into long-term procurement decisions. Organizations running sustained high-compute AI workloads -- such as large language model fine-tuning, computer vision training at scale, or generative AI inference fleets -- should consider the implications of baseload nuclear power on both cost predictability and sustainability reporting. If your business requires consistent, high-throughput GPU capacity for mission-critical AI applications, Google's Finnish data center footprint with dedicated nuclear power offers a level of reliability and carbon credibility that few providers can match.
IT decision makers should also evaluate how this infrastructure investment translates into regional service availability and latency for European customers. Companies subject to EU data residency requirements may find additional value in locating AI workloads within Finland's robust regulatory environment. When comparing cloud AI offerings, ask prospective vendors about their long-term power procurement strategies, 24/7 carbon-free energy timelines, and ability to scale GPU capacity without energy-related bottlenecks. The organizations that position themselves now with providers that have secured decades of stable, carbon-free power will be best positioned to scale AI capabilities cost-effectively through the 2030s and beyond.
FAQ
Why is Google investing in nuclear power for AI data centers?
Google is investing in nuclear power because AI training and inference workloads require massive amounts of 24/7 electricity that intermittent renewable sources like wind and solar cannot reliably provide on their own. A single large AI training cluster can consume hundreds of megawatts continuously. Nuclear power delivers stable, baseload carbon-free electricity that matches the always-on demands of AI compute infrastructure, helping Google meet both its scalability needs and its 24/7 carbon-free energy goals.
How does the 22-year contract compare to typical energy deals?
The 22-year Fortum contract is unusually long compared to standard corporate renewable energy power purchase agreements, which typically range from 10 to 15 years. The extended duration reflects Google's confidence that AI compute demand will continue growing for decades and that nuclear power will remain a cost-effective, carbon-free baseload resource. For Fortum, the long-term commitment provides revenue certainty that supports continued reactor operations, maintenance, and potentially future expansion of Finland's nuclear fleet.
What makes Finland an attractive location for AI data centers?
Finland offers several unique advantages for AI data center deployment: a cold climate that significantly reduces cooling costs, a robust and reliable electrical grid, abundant nuclear and hydroelectric power, political stability, strong data privacy protections under EU law, and geographic proximity to both European and Asian markets. The country's proactive approach to nuclear energy expansion, including lifetime extensions for existing reactors and plans for next-generation small modular reactors, further strengthens its position as a leading AI infrastructure hub.
How much power do AI data centers actually consume?
AI data centers consume dramatically more power than traditional cloud data centers due to the density of GPU and TPU accelerators required. A typical AI-optimized data center campus can draw 200 to 500 megawatts, with some of the largest planned facilities targeting 1,000 megawatts or more. By comparison, a conventional enterprise data center might use 20 to 50 megawatts. The International Energy Agency projects that AI data centers could account for 4 to 8 percent of global electricity demand by 2030, up from less than 2 percent today.
Will other tech companies follow Google's nuclear strategy?
Multiple hyperscale technology companies are already exploring or pursuing nuclear power agreements for AI infrastructure. Microsoft has announced partnerships with nuclear operators and invested in SMR technology, while Amazon and Meta have been linked to nuclear procurement discussions. The economics and reliability of nuclear power are increasingly compelling as AI workloads scale, and industry analysts expect more nuclear deals from major tech firms over the next two to three years as the competition for AI compute capacity intensifies.