The best AI investment might be in energy tech

Over the past five years, the venture capital industry has poured over half a trillion dollars into artificial intelligence. Investors raced to gain access to the latest models, the best talent, and the greatest computing power. But now it turns out they may have their focus completely wrong. According to a Sightline Climate report, as much as 50% of announced data center projects may be delayed, and the main culprit is not a lack of capital or technology — it's a lack of electrical power.

This observation changes how we should think about the AI ecosystem. Over recent years, the discussion has focused on the race between OpenAI, Anthropic, Google, and other giants in creating increasingly advanced models. But behind the scenes, an equally important — and perhaps more important — battle was taking place over access to clean, stable, and cheap power. Without energy, there are no data centers. Without data centers, there is no model training. Without models, there is no AI revolution. It's a simple equation that many investors overlooked.

For venture capitalists and strategic investors, this means that the most promising growth opportunities may no longer lie in the next chatbot or language model, but in energy technologies that will power the entire AI infrastructure for the next decade.

Energy as the new infrastructure crisis

The scale of the problem is hard to overstate. A modern data center consumes energy comparable to a small city. Google's computing center in Singapore, one of the largest in the world, draws as much power as 200,000 homes. Now imagine that you need to build dozens, or perhaps hundreds, of such facilities worldwide to meet the growing demand for training capacity for AI models.

The problem is not new — energy has always been a bottleneck for technological ambitions. But in the case of AI, it has taken on dramatic proportions. Traditional data centers can operate intermittently, can be flexible in power consumption. AI training centers cannot. Power outages mean losing weeks of work, corrupting models during training, enormous financial losses. That's why they require reliable, continuous, and predictable access to electrical power.

Practice illustrates this well. Meta and Microsoft announced plans to build gigantic data centers, but immediately faced resistance from local authorities and power grid operators. Ireland, which has attracted hundreds of billions in tech investments, is starting to deny new permits for data center construction — not due to lack of space or money, but because the local power grid cannot handle the load. This is not a temporary regulatory problem. It's a fundamental physical limitation.

The Sightline Climate report shows that this limitation will translate into real project delays. If half of announced data centers are not built on schedule, it means a change in the entire AI development timeline. Models will be trained more slowly, their availability will be delayed, competition between tech companies will shift from a technological race to a race for access to energy.

The renewable dream and real limitations

Of course, the answer seems obvious: switch to renewable energy. OpenAI, Google, Microsoft — everyone promises that their data centers will be powered by clean sources. This sounds beautiful in investor presentations and press releases. Reality is more complicated.

Solar and wind energy are variable. The sun doesn't shine at night, the wind doesn't blow constantly. To maintain continuous power, energy storage solutions are needed — batteries, heat storage systems, other technologies. These solutions exist, but are expensive and still developing. Building a data center powered entirely by renewable sources requires additional investments in energy storage infrastructure, which can account for 20-30% of the total project cost.

Moreover, even if energy is available, it must be nearby. Transmitting electricity over long distances involves losses and requires modernization of the transmission network. Most of the world's existing energy infrastructure is not prepared for a sudden spike in demand. Network modernization is a multi-year, expensive project requiring coordination between governments, grid operators, and private investors.

This is precisely what creates an opportunity for investors. Companies that can quickly and efficiently solve the problem of energy access for data centers will have access to billions of dollars in capital. These will no longer be startups fighting for market position — they will be key links in the infrastructure of the entire AI industry.

Where the energy business opportunity lies

If energy is the bottleneck, where are the most interesting opportunities for investors? The answer is not straightforward, but several directions can be identified:

• Energy storage — batteries, heat storage systems, water pumping systems. Companies like Tesla (with Powerwall) and other battery manufacturers will have growing demand. But also startups working on new storage technologies — from flow batteries to energy storage in the form of compressed air.
• Nuclear energy — small modular reactors (SMR) are gaining popularity. They can be installed near data centers, providing steady, low-emission power. Companies like NuScale Power or X-energy have the potential to become key players.
• Network infrastructure — modernization of transmission networks, smart grids, energy management systems. Both large corporations (Siemens, ABB) and newer companies specializing in network optimization software operate here.
• Geothermal energy — less popular, but promising. It allows for steady power supply regardless of weather conditions. Several startups are working on geothermal technologies that could be scalable.
• Data center energy efficiency — new cooling technologies, software optimization, reducing heat losses. Even if we don't increase available energy, we can reduce demand.

Each of these directions has its favorites among investors. But the common denominator is that none of them are traditional "AI startups" — these are infrastructure companies, often working in industries that have existed for decades. Their value will grow dramatically if they can deliver energy for the new generation of data centers.

Geopolitical games over energy resources

You cannot talk about energy for AI without mentioning geopolitics. Access to energy is access to power. Countries that have access to cheap power for data centers will have an advantage in the AI race. That's why governments are starting to intervene in this market.

The United States has invested billions in the Inflation Reduction Act and infrastructure to support the development of renewable and nuclear energy. Europe is working to change regulations to accelerate the construction of new power plants. China, which already has access to cheap coal, is building data centers at a pace that seems impossible for Western competitors to match.

For investors, this means that geopolitical decisions will have a direct impact on the value of their portfolio. A company that receives government support for nuclear energy development could suddenly become much more attractive. A country that grants permission to build a new data center will attract investment in energy infrastructure.

It also means that investing in energy for AI is not purely a technical challenge — it's a game of influence, subsidies, and access to natural resources. Venture capitalists who understand this dynamic will have an advantage.

Why venture capital missed this opportunity

It's worth wondering why for so long investors focused on AI software startups rather than energy. The answer is understandable, but also somewhat unsettling.

First, energy is boring. It doesn't have the same magic as AI. No one writes articles about how a battery will change the world — but articles about GPT-5 attract millions of readers. Venture capital lives on hype, and energy has no hype.

Second, the energy market is dominated by large corporations and governments. For a startup to enter this market means fighting established players, regulations, and bureaucracy. This is not a quick path to exit. It could be a path to enormous profits, but it requires patience and significantly larger capital than a typical venture round.

Third, energy is capital-intensive. Building power plants, transmission networks, or battery factories requires billions, not millions. This is not a venture capital model — it's a private equity or infrastructure investment model. Many venture capitalists simply don't have experience with this type of investment.

But now that the problem has become visible, now that the Sightline Climate report has shown that energy is a limitation, not just a curiosity — the perspective is changing. Large venture capital funds, infrastructure funds, and strategic investors are starting to pay attention to energy. This may mean that we are just entering a phase where investments in energy for AI will grow faster than investments in AI models themselves.

Future scenarios and real opportunities

What are realistic scenarios for the next 5-10 years? We will most likely see a combination of several trends.

First, acceleration of investments in nuclear energy. Small modular reactors are no longer just theory — they are in the commercial implementation phase. If a few projects succeed, it could trigger a wave of investment. For venture capital, this means opportunities in companies supplying components, software, and services to the nuclear sector.

Second, a boom in energy storage. Lithium-ion batteries will become cheaper, but new technologies will also emerge — sodium-ion batteries, heat storage, mechanical storage. Companies that become leaders in these technologies could achieve valuations comparable to today's tech giants.

Third, market consolidation. Large energy corporations will acquire startups working on new technologies. This means for investors the possibility of good returns on investment, but also the end of the era of independent energy startups.

Fourth, relocation of data centers. Centers will be built where energy is cheapest and most easily accessible — not necessarily near large cities or major markets. This could mean investment in infrastructure in less populated regions and countries.

Realistically speaking, an investor who today puts money into an energy company supporting AI infrastructure could achieve returns comparable to those venture capitalists achieved by investing in OpenAI or Anthropic five years ago. But with one difference — the risk is much lower because the market is more predictable and supported by governments.

A lesson for investors: look beyond the obvious

The entire story of energy for AI is a lesson for the venture capital industry. For years, investors chased the newest technology, the hottest trend, the most futuristic idea. But sometimes the biggest opportunities lie in the least sexy, most fundamental problems.

AI will not be a revolution without energy. Energy without infrastructure will not be available. Infrastructure without investment will not be built. It's a simple chain of cause and effect, but a chain where each link is equally important.

For venture capitalists who missed the first wave of AI investments, energy could be a second chance. For those who already have exposure to AI, investments in energy could be a natural extension of their portfolio, providing long-term value protection and additional profits.

The Sightline Climate report may turn out to be one of those documents that in retrospect will be seen as a turning point in the history of technology investments. A moment when the industry realized that before we can have AI — we must have energy. And it is precisely energy, not AI, that will be the next great battlefield for investors.