It appears that I have been stuck on the topic of data centers lately when I would rather be writing about whatever happens to be in the news. Yet, for some reason, many of the stories I come across seem to circle back to data centers.
The articles often fall into one of two camps. The first rejects new data centers outright, citing concerns about power consumption, water usage, noise, and the impact on local communities. The second argues that we have no choice but to build more data centers if we do not want to fall behind in an increasingly digital world.
When I have written about data centers in the past, I have generally fallen into the group that believes we need to continue building them. However, unlike some of the reporting I see, I also understand the concerns being raised by those who oppose new developments. Rising utility costs, increased water consumption, noise, and other quality-of-life issues are legitimate concerns that deserve discussion.
What I have tried to focus on is not simply whether data centers should be built, but on the efforts being made to address those concerns.
With that said, one of the topics I keep coming back to is the push for small modular nuclear reactors. I discuss these reactors for a number of reasons.
For decades, we have neglected our electrical grid and power infrastructure. Concerns about the future capacity of our power grid are nothing new. Long before data centers became a major topic of discussion, experts were warning about the possibility of brownouts, increased demand, and the need for significant infrastructure upgrades.
At the same time, our demand for electricity continues to grow.
Electronic devices are everywhere.
Electric vehicles require charging infrastructure.
Data centers were already expanding to support cloud hosting, web hosting, banking, healthcare, government services, and large businesses long before AI became today’s buzzword.
Now, with the rapid push toward artificial intelligence, the pressure on our electrical infrastructure could increase much faster than many previously predicted.
Unfortunately, nuclear power generation has been on life support for years. Accidents in both the United States and around the world damaged public confidence, while the construction of traditional large-scale nuclear plants became known for massive budgets, lengthy construction timelines, and frequent cost overruns. As a result, very few new plants have been built.
This is where small modular reactors become interesting. As these reactors move from concept to reality, they may finally offer at least a partial solution to our growing need for electrical generation. They will not solve every problem, but they could become an important tool in meeting the increasing energy demands of our modern world.
Are these modular reactors ready for prime time and already being deployed across the country? Unfortunately, no. However, what is encouraging is that they appear to be getting much closer to becoming a reality.
AP news reports that a test reactor has reached a critical milestone.
WASHINGTON (AP) — The Energy Department says a small nuclear reactor under development at a national lab has reached a crucial milestone that could allow it to produce electricity within a few years.
The microreactor being developed by Antares Nuclear Inc. at the Idaho National Lab reached “criticality” on Thursday, Energy Secretary Chris Wright said. The milestone occurs when a nuclear reactor achieves a self-sustaining chain reaction capable of producing a steady release of energy.
As we used to say to our parents when we were kids, “Are we there yet? Are we there yet?”
When it comes to small modular reactors, the answer is still no.
However, we are much closer than we were even a few years ago. For the first time in a long time, it feels like we are moving beyond concepts, proposals, and promises. Actual reactors are being built, tested, and reaching important milestones.
Are we there yet? No.
But I think I can finally see the sign in the distance telling us that we are getting close.
One of the things I haven’t addressed, and that is on me for overlooking it, is the cost of these modular reactors. Since cost has to be part of any discussion about power generation, I decided to look into it.
The following comes from a 2025 article titled Faster, Cheaper, Smarter? The Promise and Pitfalls of Small Modular Reactors. While the article is over a year old, I believe the cost analysis is still worth considering today.
“The Long Road to Competitive Overnight Costs
Currently, projections suggest that the overnight costs of SMRs will be significantly higher than those of conventional nuclear power, which are already high. For instance, the International Energy Agency (IEA, 2025) estimates SMR overnight costs in the EU at around $10,000 per kW, compared to $6,600 per kW for traditional nuclear.
While SMR costs are projected to decline as the industry transitions from FOAK (first-of-a-kind) to NOAK (nth-of-a-kind) designs, even optimistic scenarios suggest it will take decades before SMRs reach cost parity with conventional nuclear power.
Under the IEA’s Stated Policies Scenario (STEPS), SMR capital costs in the West are projected to remain about one-third higher even by 2050. In the more optimistic Announced Pledges Scenario (APS), cost parity might be achieved by mid-century.”
“Cost of Capital is Crucial
However, even with higher overnight costs, SMRs may still be cheaper when financing is considered. The overnight cost metric, though widely used, offers only a partial view, especially for nuclear projects, where financing costs can account for tens of percents of total expenditures due to lengthy construction periods.
In contrast, the shorter build times of SMRs can mitigate these interest-related expenses. Consider the following comparison: a conventional nuclear plant with an overnight cost of $6,600 per kW (IEA, 2024: 333) versus an SMR at $10,000 per kW (IEA, 2025: 47). Assuming a financing rate of 5%, and construction timelines of 15 years for the conventional plant and 5 years for the SMR, the total cost of the SMR will come out lower ($12,763 and $13,721 per kW). This is due to the reduced accumulation of interest over a shorter construction period.
Crucially, this cost advantage is achievable even before the anticipated price reductions from scaled-up SMR deployment begin to materialize. However, the sensitivity to lead times is difficult to overstate. If SMR construction timelines were to extend by just two years (from 5 to 7 years) the total cost would rise to $14,071 per kW, surpassing that of conventional nuclear.”
Honestly, this is the type of article I like to find and share. There is no sugarcoating here. The article discusses both the potential benefits and the challenges. It acknowledges that SMRs may cost more upfront, while also pointing out that shorter construction timelines could offset some of those costs through reduced financing expenses.
Whether the projections prove accurate remains to be seen. However, I appreciate seeing an analysis that discusses both the advantages and the drawbacks instead of treating the technology as either a miracle solution or a complete failure.
Personally, I don’t think we have the luxury of doing nothing. Our demand for electricity continues to grow, whether from homes, electric vehicles, businesses, or data centers.
But that is for each of us to decide.
Data centers are not the reason we need more power. They’re one more demand being added to a problem that already existed.
Now that I have discussed one possible way to address our need for additional power generation, let’s look at the other side of the equation.
The reality is that we need more electrical generation regardless of whether a single new data center is ever built. Our electrical grid has been under pressure for years. Homes consume more power than they once did. Businesses continue to expand their electrical needs. Electric vehicles require charging infrastructure. Existing data centers continue to grow to support cloud services, banking, healthcare, government systems, and countless other services we depend on every day.
Data centers and AI may be accelerating the conversation, but they are not the sole reason we need additional power generation.
At the same time, simply building more power plants is not enough. We also need to find ways to use energy more efficiently.
Honestly, I do not believe you can have one without the other.
That is why I find some of the recent developments coming from the technology industry so interesting. While one group is focused on producing more electricity, another is focused on reducing the amount of energy required to move and process data.
And that brings us to Nvidia’s push to replace copper connections with light.
When we talk about computers and data centers, one of the biggest challenges is heat. The computers themselves consume large amounts of electricity, and much of that energy eventually becomes heat that must be removed. The same thing happens with the countless copper connections used to move data throughout a data center. As electrical signals travel through copper wiring, resistance generates heat, wasting energy and increasing cooling requirements.
This is where Nvidia’s recent announcement caught my attention. Rather than relying solely on traditional copper connections, Nvidia is beginning to implement photonics technology that uses light to move data. By reducing the amount of electricity required to move information, the technology has the potential to reduce both power consumption and heat generation.
In simple terms, less wasted energy means less heat, and less heat means less cooling. Both can help lower the overall power requirements of future data centers.
What caught my attention wasn’t a specific technical specification. It was the simple fact that Nvidia announced the technology is now in production. For years we have heard about photonics, optical networking, and using light instead of copper. Now it appears the industry is beginning to move beyond the discussion stage and into actual deployment.
Over the number of posts I have made about data centers, AI, the energy infrastructure, and critical minerals required, I have tried to be honest about both the benefits and the problems that have been raised.
The one point I have tried to drive home is that our energy infrastructure has been neglected for far too long, and it has to be addressed now. AI and our continuing reliance on electronics in our everyday lives are just making this issue more pressing.
I have also noted that the genie is out of the bottle when it comes to AI and other advancements. There is no way we can put it back unless we want to lose our competitiveness in the world and also the quality of life that we have come to enjoy.
So, what I have tried to do is provide insights into the things that are being done to both address the concerns and acknowledge those concerns without ignoring the fact that we cannot turn back now.