Nuclear energy is having a moment. It seems as if every podcaster is confidently explaining why it's the obvious solution to our energy future. But after hearing similar confident proclamations about housing, an industry I know well, I've learned to be skeptical of expertise-by-enthusiasm.

So I reached out to a friend with a PhD in Engineering Systems from MIT who specializes in clean energy and has operational experience. His take? Nuclear has incredible fundamentals and policy tailwinds. However, the industry has a terrible track record of actually delivering on its promises when it comes to timeline and costs. Additionally, there are ~75 different reactor designs competing right now, and less than a handful will ultimately win. So in a nutshell: the market has potential but seems overhyped.

I decided to dig deeper and examine the underlying data. The first half provides an overview of energy fundamentals and the second half provides a deep-dive into nuclear power.

I encourage you to draw your own conclusions, but my quick takeaways are:

1. We need more power. The underlying assumption that energy consumption will increase over the next 30 years is logically sound. The exact amount (1000 TWh or 2500 TWh) and timing is anyone's guess.

2. We should take a portfolio approach. Gas historically was the quick/cheap stopgap, but with rising costs and long backlogs, renewables are becoming a strong competitor. Long-term, nuclear seems ideal, however the track record of build-outs is atrocious. We need a diversified energy mix - gas and renewables for the near-term, nuclear for the long-term - each playing a role in generating power in the most cost and scale efficient way.

3. Nuclear’s potential feels great, the industry feels frothy. Nuclear should grow as a percentage of total energy supply over the next 20 years, especially with recent regulatory tailwinds. However, there are far too many new entrants, and without having deep technical and regulatory expertise, picking a winner is extremely challenging. Also, for an industry that hasn't delivered on-time or on-cost in decades, investors are really betting startups can not only figure out a better solution, but also scale it.


Part I: Energy Fundamentals

Energy is measured in kilowatt-hours (kWh), or the total energy consumed or produced when 1 kW is sustained for 1 hour. The average US household consumes about 10,000 kWh annually (varies by climate, home size, HVAC load). For perspective:

This analysis focuses on terawatt-hours (TWh), where 1 TWh = 1 billion kWh.

Demand

In 2024, the US consumed ~4,100 TWh (terawatt-hours) for civilian electricity consumption, comprised of:

  • Residential: 38% (homes)
  • Commercial: 36% (offices, stores, data centers)
  • Industrial: 26% (manufacturing, mining)


This amount is not in itself concerning relative to historical levels, however, the concern comes from the projections being put out by the US Energy Information Administration (EIA), which shows a massive uptick from this baseline over the next 30 years. AI / data centers, industrial manufacturing, and electrification of transportation and heating, has increased projections for US electricity consumption from a relatively flat trend averaging ~4,000 TWh/yr from 2008-2021 to projected growth reaching ~5,000-6,600 TWh/yr by 2050. Not only does this increase the absolute amount of energy, it also increases the level of continuous power generation needed, emphasizing the policy debate around nuclear.

Supply

The supply to fulfill the ~4,100 TWh demand comes from several sources. Clean energy makes up about 40% (~1,640 TWh), with nuclear at 19% (~780 TWh), wind at 10% (~410 TWh), solar at 7% (~290 TWh), and the rest a combination of hydro, bioenergy, and other sources. Fossil fuels make up 60% (~2,460 TWh), with natural gas at 43% (~1,760 TWh), coal at 15% (~615 TWh), and other fossil fuels as the remainder. The best source for visualizing this is the EIA US energy flow website here (US Energy Information Administration):

From the trend data (also from EIA), natural gas production and renewable energy have been increasing sources of production, whereas coal has been in steep decline.

What has driven the sudden urgency around energy supply? There are two key factors which have catalyzed the discussion: (1) there has been a near-term demand surge from large commercial data center customers driven by AI, and (2) infrastructure constraints limit how quickly we can expand energy resources and have resulted in power bottlenecks and concerns around future supply.

Here's my assessment of the situation:

1. There is rapidly rising demand for energy, of which AI & subsequent data center build-out is a significant portion (especially in the near term), and this is colliding with barriers to timely resource expansion. This is putting pressure to find a solution that addresses this rising demand.

2. The EIA estimates we will need an additional ~2,000 TWh by 2050.

3. The solution will need to provide an energy source that is reliable, cost efficient, and considers the impact on climate change. 


Discussion of Future Energy Supply

Discussion 1: How confident should we be about needing another ~2,000 TWh? 

Over the past couple of years the EIA has been revising up the amount of energy they think we will need (see figure 5), and right now it stands at closer to 6,600 TWh by 2050. The EIA has high confidence that:

  • We will see continued growth above the historical ~4,000 TWh baseline
  • AI & Data centers will be the largest growth driver in the short-term (through 2030)
  • Growth will be concentrated in commercial and industrial sectors


They are less certain about:

  • Timing: Whether we hit 5,000 TWh by 2030, 2035, or 2040+
  • Peak demand: Whether we eventually reach 6,000+ TWh as EIA projects
  • Growth sustainability: Whether infrastructure and grid constraints will limit growth



As discussed in this report, AI & subsequent data center build-out accounts for ~25% of the higher energy consumption. While not the sole driver, it is the fastest growing segment and presents an urgent near-term bottleneck. According to the EIA, “Computing accounted for an estimated 8% of commercial sector electricity consumption in 2024 and will grow to 20% by 2050.” While near-term resource constraints are coming from AI, the long-term impact is unknown and depends on how the technology evolves. As the report notes, the internet boom from 2000-2005 caused data center electricity usage to double, but then the industry focused on driving efficiency resulting in slower growth thereafter and almost no growth from 2010-2018. So depending on how computationally efficient AI is and how data center technology develops, this projection could be highly variable.

Additionally, there is an argument that the existing US power system is underutilized, and with more efficient load flexibility we could squeeze more power out of our current system. The argument, highlighted in this report, states that we can support new loads as long as customers can scale back usage during some peak hours of the year. Basically, get more efficient with the energy supply we do have. 

Finally, a quick look into historical projections suggests that energy consumption figures have historically deviated from projections by ~9%, with there being less risk in near-term projections and more risk the further out the projections go. However, even if the EIA’s projections are off by a factor of 2x from their historical rate, we’d still need at least ~1,000+ TWh by 2050. 

Answer 1: There is clearly a “power grab” driving short-term demand, and it seems logical that long-term demand will, for the reasons stated above, be higher than the ~4,000 TWh of the past 25 years. The amount? Could be 5,000 TWh or 6,600+ TWh, depending on computational demands and how efficiently we can load balance. The timing is also uncertain. However, the data shows a clear uptick indicating we're headed towards more, not less, power consumption. 

Discussion 2: What are our options for supplying this higher demand?

Argument for Natural Gas: Natural gas has been the largest energy supply, historically fast to add capacity, and has existing infrastructure. Since gas produces lower emissions than coal, increasing gas supply is serving as a stopgap. But the recent surge in data center demand has led to supply constraints, resulting in higher cost and longer timelines to build gas power plants. On costs, according to the CEO of NextEra Energy, “We built our last gas-fired facility in 2022, at $785/kW. If we wanted to build that same gas-fired combined cycle unit today…$2,400/kW,” he said. “The cost of gas-fired generation has gone up three-fold.” On timing, if you place a new order today, you're likely to get your units in 2030. Finally, on fuel availability, several regions (e.g. Arizona) are critically short, so some data centers are being told they need to wait for new gas pipeline buildout. While gas seemed like an easy stopgap, cost and timing pressures are driving people to look elsewhere. Finally, the focus on climate change and emissions does make relying solely on gas long-term out of favor with some big-tech companies who are driving recent data center demand.

Argument for Renewables (Wind / Solar): Solar costs have plummeted 85% since 2010, and wind has fallen 70%, making them cost competitive with gas. Renewables projects can be built in 2-4 years versus 10-15 years for nuclear (again, similar timeline to gas with the current backlogs). Given the immediate urgency of AI demand growth, renewables can scale much faster with an easier regulatory overhead than nuclear. Finally, battery costs have fallen 90% since 2010. Grid-scale storage is becoming economically viable, addressing renewables' historical intermittency problem (discussed in more detail below). 

Argument for Nuclear:  Nuclear plants run 24/7 at ~90% capacity factor (meaning they produce 90% of their maximum possible output), while solar averages ~25% and wind ~35% due to weather dependence. For data centers that need constant power, this consistency is appealing. A nuclear plant generates massive amounts of power on a small footprint. To replace a 1,000 MW nuclear plant, you'd need roughly 3,000 MW of solar (covering ~13,000 acres) plus enormous battery storage, versus nuclear's ~1,000 acres. Nuclear provides "firm" power that helps maintain grid frequency and voltage. Renewables require sophisticated grid management and the ability to rely on renewables during peak periods is limited, requiring a portfolio approach. We will discuss this below, but the biggest argument against nuclear is the long build timelines and high costs.

Answer 2: Figuring out the right mix of supply isn’t easy and you can argue for any of these paths. Gas isn’t the obvious stopgap it once was, and while transitory, the rise in costs and timelines for gas power plants is driving people to look elsewhere. There are also long-term climate goals that weigh-in on the decision. This has caused renewables to seem more attractive, given their cost, time to build and easier regulatory process. However, the counter argument for renewables is that there are intermittency issues, meaning customers want 24/7 supply for 24/7 demand. This, plus power density, has caused people to gravitate towards nuclear. However, the counter to nuclear is: it is high risk given the industry hasn’t shipped on-time or on-cost in decades. So then we end up talking in circles :)

There is one point worth unpacking, which is the improvement of renewables at continual power generation, which means providing a reliable flow of energy 24/7 to match demand as it fluctuates through times of the day and seasons. Many will cite that gas excels at load following, meaning it can ramp up and down quickly. Whereas renewables, namely solar at night and wind day-to-day, are cited as having intermittency concerns. Nuclear is seen as steady and able to run a constant output for months. With tech companies spinning up massive datacenters and wanting a reliable source of energy to go with it, nuclear is seen as a power dense (a lot of power on a smaller footprint), high capacity (electrical output), and reliable solution. 

The question should likely be framed as: given peak load amounts, what is the cheapest and most reliable way to supply power? Today, the cost of lithium-ion batteries have declined by ~90%, and duration has extended from 1-2 hours to 4-8 hours being the standard (with some companies targeting 12+ hours). While challenges remain for longer-duration seasonal storage, there has been enough progress that storage for renewables can now compete with gas during peak demand hours. 

So it seems that folks are jumping to nuclear power because it gives them the assurance of continual power rather than risking any downtime, despite renewables making tremendous strides on this front. There is a secondary reason…queue the next section!

Regulatory Environment

I think a lot of the recent hype around nuclear power comes from increased regulatory support. Back in May 2025, Trump signed a series of executive orders aimed at fast tracking nuclear build-outs to help provide energy needs to win the AI race. In addition, Trump’s One Big Beautiful Bill that was signed on July 4th significantly alters clean energy tax credits established by the Inflation Reduction Act, favoring nuclear energy and reducing incentives for solar and wind projects. Whether or not you agree with these policies, it is undeniable that the regulatory environment for nuclear power has become more favorable in the last 4 months. From an investor perspective, this feels like a rare opportunity where there could be massive market demand in a favorable regulatory environment. 

My Two Cents

We likely should (and need to) do it all and see which one wins out on reliability, price, and scale. Betting on just nuclear seems like a massive gamble to make. Instead, let's increase natural gas production as a bridge, build out wind/solar and continue to benefit from falling costs, better storage and quicker deployment, and build nuclear for the longer-term goal of cleaner, more reliable energy supply. I can see the long-term argument for nuclear, however I do think the US needs a plan for the next decade which focuses more heavily on gas, solar, and wind.


Part II: Nuclear Deep-Dive

So now that we’ve gotten out of that massive rabbit hole, time to jump down the next: nuclear.

History

Nuclear is an established technology. In 1938, German chemists Otto Hahn and Fritz Strassmann demonstrated nuclear fission, by bombarding uranium with neutrons thus splitting up its atoms into lighter elements (like barium), releasing a tremendous amount of energy. In 1942, Enrico Fermi achieved the first controlled nuclear chain reaction. The Manhattan Project in the United States built the first atomic bombs, tested in 1945, demonstrating the destructive power of uncontrolled fission chain reactions. After the war, attention shifted to harnessing controlled fission for useful energy. In 1951, the first experimental electricity from a nuclear reactor was produced in Idaho, and by the 1970s nuclear energy was providing a significant share of electricity in many countries. However, two high profile accidents, Three Mile Island (1979) and Chernobyl (1986) stoked public fear, causing construction costs to escalate with increasing regulation, and pausing most nuclear build outs. Fukushima (2011) exacerbated fears and prompted many countries to phase out nuclear.

Ecosystem

Today's global nuclear fleet comprises 440 reactors in 32 countries generating 2,600 TWh annually, or roughly 10% of world electricity. The United States leads with 94 reactors producing ~780 TWh, but China is rapidly catching up with 56 reactors and aggressive expansion plans. There has been a lot of public chatter about China outpacing the US on nuclear, and this is because they currently have ~25 reactors under construction (and has spoken about plans for 40-50) while the US only has 2 under construction (Vogtle Units 3 & 4 in Georgia). In addition, China builds their reactors at an average speed of 7 years compared to the US which can take 10-15 years. Finally, China’s state bank will finance nuclear build out at ~70% of reactor costs and a low interest rate (~1-2%), resulting in Chinese construction costs of $2,500-3,000 per kilowatt vs. US construction costs of $6,000-10,000 per kilowatt (note: I’m quite skeptical of all numbers coming out of China so take these with a grain of salt). So when you hear China is “outpacing” the US on nuclear, it refers to the momentum of deployment: the US is ahead but walking, while China is behind but sprinting.

Building new reactors involves a complex ecosystem of players. At the top are the reactor owners, either single entities or joint ventures that hold the equity in the project. These owners secure financing through a mix of corporate bonds and federal funding. The owners then hire reactor vendors (companies like Westinghouse, GE-Hitachi) who design the nuclear reactors, own the intellectual property and patents, and manufacture the specialized nuclear components. Finally, the reactor vendors partner with construction firms (like Bechtel, Fluor, CB&I, or Kiewit) that handle the physical construction, building the plants, installing equipment, and managing all the civil engineering work.

Most owners operate their own plants, but there are exceptions. Constellation Energy is an owner and operator, as is Duke energy. Energy Harbor contracts with Constellation to operate its Beaver Valley plant in Pennsylvania, but operates Davis-Besse and Perry itself. A lot of this comes down to regulation requirements and economies of scale.

The nuclear reactor market is dominated by a handful of companies that own the intellectual property for reactor designs. Currently, only Westinghouse's AP1000 is being built in the US (at Georgia's Vogtle plant). The AP1000 is a Generation III+ reactor, which represents evolutionary improvements to traditional water-cooled designs with enhanced safety features like passive cooling that works without external power. Internationally, other Gen III+ designs are being built include Framatome's EPR, Russia's VVER-1200, China's Hualong One, and Korea's APR1400, and all are advanced versions of conventional pressurized water reactors that have dominated nuclear power since the 1960s. 

For the future, nuclear is pursuing two parallel paths for fission: 

Small Modular Reactors (SMRs) - These are reactors that shrink Gen III+ technology for faster deployment. While their output is lower compared to conventional plants (300 MW vs gigawatts), the idea is these reactors can be deployed more quickly using factory production, standardized designs, and thus a streamlined regulatory process. China already has SMRs operational, and while the US has some planned none have received permits yet. 

Generation IV - These reactors that use entirely different coolants and physics, promising higher temperatures for industrial heat, the ability to consume nuclear waste as fuel, and inherent safety where meltdowns are physically impossible. The US has broken ground on a planned Gen IV reactor in Wyoming (Terrapower, a Natrium reactor,  SMR at ~345 MWe with liquid sodium coolant) but actual construction hasn’t begun and is waiting on their construction permit from the NRC. China has already operated the world's first commercial Gen IV reactor, which went live in December 2023.

From a regulatory perspective, there is a two-step approval process. Step 1 is design approval, which can take 3-5 years. The process is roughly:

  • Submit your reactor design to the Nuclear Regulatory Commission (NRC)
  • NRC reviews safety systems, engineering, and potential accident scenarios
  • If approved, you get a "Standard Design Approval" - basically a stamp saying "this design is theoretically safe"


Step 2 is construction & operating license, which can take 2-5 years. This process is roughly:

  • Submit application for a specific site where you want to build
  • NRC reviews site-specific factors: geology, population density, emergency planning
  • Environmental impact assessment
  • Public hearings and comment periods
  • If approved, you can build and eventually operate


The timeline can be a total of 5-10 years, and often the approval process takes more time and is more expensive than building the reactor itself, causing many startups to burn through significant capital in the process.

History of Builds

The biggest criticism of nuclear is that it has struggled to execute on-time and on-cost, causing skepticism around whether legacy incumbents or new startups will be able to change this in the future. While most will point to China as the obvious counterfactual on what is possible, I’m skeptical on their costs and timelines and question what a well executed steady-state buildout could look like. Per the below chatgpt chart of the last 5 reactor builds, our track record is atrocious. 

Only 3 of the last 5 planned reactors were actually completed, average overrun on cost was 250%+, and the average delay was over 6 years. The crazy thing is: we should be better at this. We know the science, the issue has been on execution and regulatory hurdles. So I can fully see why new entrants think they can have better results, especially with SMRs and a favorable regulatory environment, but it is hard to look past this track record. Outside of the regulatory tailwinds, are all of these startups really that much better at execution than GE or Westinghouse?

Fission vs Fusion

The above discussion is all based on fission reactors, where the new technology being built is SMRs and the Gen IV reactor. SMRs provide a way to standardize and thus speed up nuclear build outs, and Gen IV is the latest reactor technology that makes meltdowns impossible. 

However, there is one other massive technological innovation new entrants are working on: fusion. 

Nuclear fission (the legacy technology) splits heavy atoms like uranium-235. When a neutron strikes a uranium nucleus, it becomes unstable and breaks apart into smaller fragments, releasing 2-3 additional neutrons and roughly 200 million electron volts of energy. This creates a self-sustaining chain reaction that, when controlled, generates steady heat for electricity production. Current reactor designs carefully manage this process. Pressurized Water Reactors, which comprise 85% of the global fleet, circulate water at 155 bar pressure through the reactor core. At this pressure, water remains liquid even at 300°C, absorbing neutrons to control the reaction rate while carrying away heat. Control rods made of neutron-absorbing materials like boron can be inserted to slow or stop the reaction entirely.

Nuclear fusion (the new technology) combines light atoms under extreme conditions. The most promising approach fuses deuterium and tritium (heavy hydrogen isotopes) to create helium, releasing 17.6 million electron volts per reaction. While each individual fusion reaction releases less energy than fission, fusion produces 3-4 times more energy per unit mass of fuel. The challenge lies in creating conditions that exist naturally only in stellar cores: temperatures of 150 million°C or ten times hotter than the sun's center. 

Fusion offers compelling advantages. Deuterium exists abundantly in seawater, providing virtually limitless supply. Tritium must be "bred" from lithium using neutrons from fusion reactions, but lithium reserves could power fusion plants for thousands of years. A 1,000-megawatt fusion plant would need only 250 kilograms of fuel annually compared to 24,495 kilograms for fission plants (~100x the fuel).

The safety profiles differ dramatically. Fission requires active control systems to prevent runaway chain reactions that could lead to meltdowns. Historical accidents occurred when these systems failed or operators made errors. Fusion reactions stop immediately if conditions change; the precise temperature and pressure for fusion requires constant energy input. Disturb the system, and fusion simply ceases.

Waste represents another crucial distinction. Fission produces high-level radioactive waste containing transuranics with half-lives of thousands to millions of years, requiring geological disposal facilities. Fusion generates no long-lived radioactive waste. 

To be clear, nobody has produced a net energy gain from a fusion reaction (energy to run the facility is greater than the output), and even if this is possible, fusion would then need to be scaled. So basically, fusion is still very far off from being commercially viable.

Competitive Landscape

The number of players in the nuclear space is significant. To make this section more digestible, I’m going to focus on the most “known” names, and break them down into: (1) legacy players, (2) new fission startups developing SMRs, and (3) fusion startups.

1. Legacy Players: The traditional nuclear industry is dominated by a handful of major corporations that design reactors and manufacture nuclear components. Westinghouse Electric Company currently owns the only reactor design being built in the US (AP1000 at Georgia's Vogtle plant). The company is developing the AP300 SMR, a 300-MW reactor based on its proven AP1000 design, targeting deployment in the early 2030s. GE Hitachi Nuclear Energy partners with TerraPower on the Natrium reactor design and manufactures nuclear components. The company has decades of experience but has struggled with cost overruns on recent projects.

These legacy players are optimized for a different era. They focus on massive, custom design reactors rather than new technologies. Regulatory moats and existing manufacturing scale matters, and these two companies will likely continue to dominate the large scale reactor space. There is also the possibility they acquire / consolidate with a new entrant to gain talent and technology.

2. New SMR Startups: The SMR sector has attracted billions in funding from both government and private investors, with a promise of faster deployment, lower costs, and enhanced safety compared to traditional nuclear plants. TerraPower has raised over $1.4B for its Natrium sodium-cooled reactor. Its founder / chairman is Bill Gates, bringing in tremendous capital and credibility. TerraPower broke ground in Wyoming in June 2024, has submitted an NRC construction permit application, and is targeting 2030 operations. Their track record of actually breaking ground and the DOE covering up to half of the Wyoming plant costs, makes TerraPower one of the most promising of the SMR startups. X-Energy has raised close to $800M for its Xe-100 high-temperature gas-cooled reactor (80 MW per unit, scalable to 960 MW). X-Energy is backed by Amazon’s climate pledge fund, and is building a facility in Tennessee, partnering with Amazon for 5 large deployments by 2039, and working on a grid-scale advanced nuclear plant with Dow in Texas. X-Energy’s CEO is a former deputy secretary of energy, who knows how to navigate the nuclear regulatory maze. Additionally, their TRISO fuel (TRi-structural ISOtropic particle fuel) is a technological advancement that is “walk away” safe. They seem to have the right regulatory approach, proven technology, and real customers. Kairos Power has raised over $600M, with large backers in the DOE and Google. They have NRC approval for construction of two test reactors in Tennessee and are targeting their first commercial reactor online by 2030. Google is an early customer and signed up for a 500 MW Master Plan Development Agreement with them. They also use TRISO fuel and have broken ground. NuScale Power is public and has a SMR reactor that scales from 77 MW per module up to 924 MW. They are the only company with NRC design approval for SMRs, however they suffered a massive setback in 2023, when they cancelled a project in Idaho because of higher than estimated build costs. Their stock has seemingly run up recently on AI hype but their underlying fundamentals are still challenging. Oklo is also public and is backed by Sam Altman. Their Aurora microreactor (15-75 MW) uses recycled nuclear waste as fuel, and they have a DOE site use permit for Idaho National Laboratory, targeting late 2027/early 2028 operations. Unlike other nuclear companies that license technology, Oklo will own and operate the plants under long-term power purchase agreements. Their regulatory track record is mixed, having received a key environmental permit from DOE for its Idaho site and DOE approval of its fuel fabrication facility design. However, the NRC rejected Oklo's first license application in 2022 citing "significant information gaps." They're reapplying in 2025 and targeting 2027 operations, but acknowledge regulatory risk.

3. Fusion Startups: Fusion has not been proven at scale despite billions being funded into these startups. If fusion is figured out at commercial scale, it would be a feat of humanity, and that company would be massive - however the timing is unknown and likely very long. Commonwealth Fusion Systems has raised over $2B with the main backers being Bill Gates, Google, Temasek and Khosla. Their SPARC reactor utilizes high-temperature superconducting magnets and is under construction in Massachusetts. They are targeting net energy generation by 2027 and their first commercial plant in Virginia in the 2030s (very aggressive). Google signed a massive 200 MW power purchase agreement with them in June 2025 providing strong technology validation. Their timelines have been ambitious but they seem to be the darling of the fusion startups. Helion Energy has raised over $1B with the biggest backers being Sam Altman, SoftBank, Reid Hoffman, and Peter Thiel's Mithril Capital. Their reactor, Polaris, is operational and has a power purchase agreement with Microsoft for 2028 delivery. There seems to be mixed feedback on Helion, with several experts questioning their technology. 

Conclusion

Nuclear feels to me like a new take on an old industry that will play out over decades, not years. Fission startups must navigate complex NRC licensing (typically 5-10 years), prove economic viability against cheaper natural gas and renewables, and overcome an insanely poor historical track record.

Fusion startups face even steeper technical hurdles, with most experts predicting commercial viability won't arrive before 2050, despite company claims of 2030s deployment.

As I reflect on all the data, two things aren’t clear to me:

1. That nuclear companies can execute. The track record on nuclear is hard to overlook. While renewables have made systematic improvement on cost and storage, nuclear has gone sideways. I like to think humans are smart and hungry enough to overcome this, but the data so far doesn’t make the answer obvious.

2. If someone can execute, who will it be? The nuclear industry feels somewhat similar to how the autonomous vehicle (AV) industry felt to me back in 2015. At that time I worked in venture, and it felt like every week we were hearing a new AV pitch with those founders explaining how their sensor technology and ML models were superior. With the benefit of a decade of hindsight, it is clear that Waymo and Tesla will win - two companies with access to deep pools of capital and the ability to take a long-term outlook. A quick GPT query suggests there have been ~50 fusion startups and ~100 fission startups founded in the past 10 years. I am not smart enough to know whose technology will win on the nuclear side. Even if I knew this, I don’t think technology is the only factor. The winner needs to execute at a level no other nuclear company has been able to.

I do think there is enough momentum, and the technology is sound enough, that nuclear should be able to grow its share of energy consumption over the next 20 years. However, I am skeptical on execution, and even more skeptical when it comes to investing in the space. With this many startups, and this many new reactor designs, unless you are an expert in the field, willing to build patiently over 20 years, with billions to back the right idea, I think the risk is far too high. For the average retail investor, like me, my money is sitting with a nuclear ETF and out of the private sector. For those who make a living investing, I think you really need to understand the technology and team you are betting on.