> ## Content Index > Fetch the complete content index at: https://www.growthsolutionskc.com/llms.txt > Use this file to discover other available public pages before exploring further. # Big Tech Goes Nuclear - URL: https://www.growthsolutionskc.com/big-tech-goes-nuclear/ - Published: 2026-09-22T14:00:07.000Z - Updated: 2026-09-22T14:00:08.000Z - Author: Matt Cucinotta - Tags: Tech & Innovation, #growthsolutionskc, #series-AI-energy ### Why the AI Race Is Bringing Nuclear Power Back Into Technology Strategy *Part 2 of the Growth Solutions KC AI Energy Series* --- For most of the modern technology era, electricity was something Big Tech purchased. Build the data center. Connect it to the grid. Sign renewable-energy agreements. Buy enough electricity to keep the servers running. Artificial intelligence is changing that relationship. As AI data centers become larger, more power-dense, and more important to the competitive position of companies such as Microsoft, Google, Amazon, and Meta, electricity is no longer simply an operating expense. **It is becoming part of technology strategy.** That helps explain one of the more surprising developments in the AI infrastructure boom: Big Tech is going nuclear. Microsoft is supporting the restart of a previously closed reactor in Pennsylvania. Google is helping commercialize advanced reactors. Amazon is investing in small modular reactor development. Meta has assembled agreements involving both existing nuclear plants and next-generation technologies. These companies are taking different approaches, but they are responding to the same problem. **AI requires enormous amounts of dependable electricity.** And nuclear power has something increasingly valuable in the AI era: **the ability to produce large quantities of electricity around the clock for decades.** --- ## What Matters: AI Needs More Than Electricity A data center cannot operate only when the sun is shining or the wind is blowing. Its servers, networking equipment, cooling systems, storage, and security infrastructure operate continuously. That does not mean renewables are unimportant. Wind and solar will remain major parts of the electricity mix, and battery storage continues to improve. But hyperscale AI infrastructure has placed a new premium on what the energy industry often calls **firm power**: electricity that can be counted on when it is needed. **Natural gas** can provide it. **Hydropower** can provide it where geography allows. **Nuclear** can provide it without direct carbon emissions from electricity generation. That combination—reliability, scale, longevity, and low operational carbon emissions—is why nuclear has moved back into the technology conversation. The AI race has not suddenly solved the decades-old debates surrounding nuclear power. It has changed the economics of the conversation. A technology company willing to commit to purchasing electricity for 20 years can make a power project look very different to the company trying to finance it. --- ![](https://storage.ghost.io/c/42/b4/42b4b6b8-b1f2-4aa3-8c91-88f72e71a120/content/images/2026/09/Four-paths.png) ## Microsoft: Restart What Already Exists Microsoft's nuclear strategy provides the clearest example of how AI-era electricity demand can change the economics of an existing asset. In 2024, Constellation announced a **20-year power purchase agreement with Microsoft** supporting the restart of Three Mile Island Unit 1, now renamed the Crane Clean Energy Center. The reactor is independent of Unit 2, the reactor involved in the 1979 accident. Unit 1 operated until 2019, when it closed for economic reasons. Constellation says the restored plant would return roughly **835 megawatts** of carbon-free generation to the grid and is expected to come online in 2028, subject to regulatory approvals. Microsoft will purchase electricity from the plant as part of its effort to match the power used by its regional data centers with carbon-free energy. The significance is larger than one power plant. Microsoft is not developing a reactor from scratch. It is helping create the economic conditions to bring an existing nuclear asset back into service. That may become an important part of the near-term nuclear story because restarting or extending existing plants can potentially add dependable capacity sooner than waiting for entirely new reactor technologies to mature. --- ## Google: Help Build the Next Generation Google is taking a different route. Rather than focusing only on existing nuclear plants, it entered an agreement with Kairos Power to support multiple advanced reactors. The original agreement provides a pathway for **up to 500 megawatts** of new nuclear capacity, beginning with an initial deployment targeted around 2030 and additional reactors through 2035. In 2025, that strategy became more concrete when Google, Kairos Power, and the Tennessee Valley Authority announced an arrangement for electricity from the planned Hermes 2 advanced reactor in Oak Ridge, Tennessee. TVA is expected to purchase power from the 50-megawatt plant, while Google will procure the associated clean-energy attributes for data centers in the region. This model is interesting because it distributes risk. The technology developer builds the reactor. The utility integrates the electricity into its system. The large technology customer helps create the demand and economic support that allow the project to move forward. That is more than an energy purchase. It is **market creation**. --- ## Amazon: Invest in the Technology Amazon has gone further into the development side. The company says it has invested more than $1 billion across nuclear projects and technologies and more than $500 million toward small modular reactor development. Its partnership with Energy Northwest and X-energy envisions an initial four-reactor project providing about 320 megawatts, with the potential to expand significantly over time. Amazon has also pursued nuclear strategies involving existing generation and explored additional SMR development with utilities. The broader lesson is important. The hyperscalers are no longer merely asking: **Where can we buy electricity?** Increasingly they are asking: **What new electricity resources can our capital help bring into existence?** That is a significant change in the relationship between technology and energy. --- ## Meta: Build a Portfolio Meta illustrates another approach: diversification. In early 2026, the company announced agreements involving Vistra, TerraPower, and Oklo, building on an earlier agreement with Constellation. Meta says the collection of projects could support **up to 6.6 gigawatts of new and existing nuclear energy by 2035**. The strategy spans operating nuclear plants as well as advanced reactor developers rather than betting entirely on one technology. That portfolio approach makes sense in an uncertain environment. Existing nuclear plants offer proven technology but may require investment, relicensing, or life-extension work. Advanced reactors may offer greater flexibility and new deployment models, but they carry more development risk Instead of choosing one path, Meta is effectively spreading its bets across several. --- ## Why Big Tech Matters to Nuclear ![](https://storage.ghost.io/c/42/b4/42b4b6b8-b1f2-4aa3-8c91-88f72e71a120/content/images/2026/09/Why-nuclear-fits.png) For decades, one of nuclear power's biggest challenges has been economic. Large plants require enormous upfront investment. Projects can take years to permit and construct. Delays are expensive. Investors and utilities therefore face a basic question: **Who will buy the electricity for long enough to justify the investment?** Hyperscalers can help answer that question. - They have enormous balance sheets. - They expect long-term electricity demand. - They can sign contracts measured in decades rather than years. - They have a large incentive: losing access to sufficient computing capacity can threaten businesses worth hundreds of billions of dollars. That makes technology companies unusually powerful **anchor customers** for energy projects. Their capital cannot eliminate engineering or regulatory risk. But it can make projects financeable that might otherwise struggle to move forward. --- ## What Nuclear Does Not Solve This is where perspective matters. Big Tech's nuclear push does **not** mean nuclear power will solve the immediate AI electricity shortage. The Crane restart is targeting 2028. Google's initial advanced reactor is expected around 2030. Amazon's first large SMR deployments are aimed at the early 2030s. Advanced reactor projects still must navigate licensing, financing, construction, supply chains, workforce needs, fuel availability, and first-of-a-kind costs. Some projects will be delayed. Some technologies may fail commercially. Some announced capacity may never be built. And the grid still needs natural gas, renewables, storage, transmission, efficiency improvements, and other generation while those projects develop. Nuclear therefore sits in a different part of the timeline. > Natural gas and existing grid resources may help bridge the immediate gap. Nuclear is increasingly a bet on the durable electricity system the AI economy may need for decades. --- ## The Bigger Shift: From Tech Company to Infrastructure Partner ![](https://storage.ghost.io/c/42/b4/42b4b6b8-b1f2-4aa3-8c91-88f72e71a120/content/images/2026/09/buyer-to-Partner-1.png) The most important lesson may not be nuclear at all. It is what nuclear reveals about the changing nature of technology companies. AI is forcing hyperscalers deeper into the physical economy. They already design chips. They build enormous data centers. They invest in networking and cooling. Now they are helping finance power generation. That does not make Microsoft, Google, Amazon, or Meta utilities. But it does mean the line between **technology strategy and infrastructure strategy is getting thinner**. Power procurement affects where data centers can be built. Generation timelines affect when computing capacity comes online. Energy reliability affects how confidently companies can deploy billions of dollars of AI hardware. **The electricity system is becoming part of the competitive landscape.** --- ## What People Should Watch Rather than assuming every nuclear announcement represents a coming renaissance, watch what actually gets built. - Do reactor restarts meet their schedules? - Do advanced reactors move through licensing and construction? - Do costs fall after the first projects? - Do utilities sign additional agreements? - Does the nuclear fuel supply chain expand? And most importantly: **Do these projects begin delivering real electricity to the grid?** Announcements tell us where companies want to go. Operating power plants tell us whether the strategy worked. --- ## What Happens Next Big Tech's renewed interest in nuclear power tells us something important about the AI infrastructure buildout. The companies closest to the AI demand curve increasingly believe reliable electricity will be valuable enough to justify billions in investment and commitments lasting decades. That is a major signal. But the nuclear story contains two very different pieces. One is relatively familiar: **Keep existing reactors operating. Restart useful assets where the economics and safety case support it.** The other is far more uncertain: **Build an entirely new generation of smaller, modular reactors that could someday sit much closer to the infrastructure they power.** That second idea deserves its own examination. Because SMRs may look almost tailor-made for the AI era. The question is whether they can move from engineering promise to commercial reality fast enough to matter. That is where we go next. --- — Matt Cucinotta | Growth Solutions KC | Inspire · Inform · Ignite --- > *Source and methodology note: This article uses the Growth Solutions KC AI Energy Research Dossier as a research framework and verifies major project details against primary-source announcements from Constellation Energy, Google, Amazon, and Meta. Nuclear projects discussed remain subject to regulatory approvals, construction, financing, fuel supply, project execution, and changing commercial timelines. The article is intended to explain emerging infrastructure trends rather than advocate for any particular energy technology.*