As tech giants compete for gigawatts, energy access is becoming the defining constraint—and investment opportunity—of the AI era.
Artificial intelligence is reshaping the global economy, but a fundamental constraint is emerging: access to electricity. The bottleneck limiting AI deployment isn’t computational capacity or algorithmic sophistication—it’s the availability of reliable, grid-scale power to run the data centers that make AI possible.
“Data centers are very power hungry.“
— Brian Kessens, CFA
Senior Portfolio Manager, Tortoise Capital
Across the U.S., proposed data centers are being denied permits or delayed indefinitely due to insufficient grid capacity. Power purchase agreements are becoming strategic negotiations involving multi-gigawatt commitments and billion-dollar capital expenditures. Companies that have secured access to electricity for the long-term have gained competitive advantages in the AI race. Those without face genuine constraints on their ability to scale operations.
Energy infrastructure, long viewed as a background utility, has become a strategic asset determining who can build and maintain AI capabilities at scale. For investors, this shift presents a fundamental reframing of where value accrues in the AI economy.
This article examines four critical dimensions:
- The Supply Constraint: Why Electricity Is the New Bottleneck
- Tech as Energy Buyer: The Strategic Shift to Captive Power
- Natural Gas and Nuclear: Powering the Buildout
- Infrastructure as Investment: Where AI Capital May Flow Next
#1 The Supply Constraint: Why Electricity Is the New Bottleneck
The challenge is straightforward: electricity demand from data centers is surging now, but building new generation capacity takes years to build. Natural gas plants require three to five years from planning to operation. Nuclear facilities take seven to ten years or longer. Transmission infrastructure upgrades introduce additional multi-year timelines.
The interconnection queue has become a significant bottleneck, with projects awaiting years for approvals or grid upgrades. Even fast-tracked projects face extended delays.
The result is a widening gap between projected electricity demand and planned supply additions. Utilities in several states have begun implementing moratoriums on new large-load customer connections, including data centers, until grid capacity can be expanded. Others require multi-year wait periods or demand that customers fund transmission upgrades themselves before interconnection approval.
“We’re at all-time highs for demand—and the outlook has never been stronger.”
— Matthew Sallee, CFA, Senior Portfolio Manager, EVP, Head of Investments, Tortoise Capital
For the first time in decades, electricity supply has become the binding constraint on technology deployment.
#2 Tech Companies Become Energy Buyers—and Builders
The strategic response from major technology companies has been direct: if the grid cannot provide the power they need, they will secure it themselves. This manifests in two approaches, both representing fundamental changes in how tech companies think about infrastructure:
A. Power Purchase Agreements (PPAs)
B. Behind the Meter (BTM)
A. Power Purchase Agreements
First, tech firms are negotiating Power Purchase Agreements (PPAs) directly with utilities and energy producers. Recent agreements involve multi-gigawatt commitments extending as long as 20 years, effectively making tech companies among the largest energy buyers in the economy. Microsoft, Google, Amazon, and others are committing billions to long-term power contracts, treating electricity supply as strategic infrastructure requiring the same capital commitment as data center construction itself.
These PPAs secure guaranteed power delivery, provide price certainty over long timeframes, and signal to utilities that demand is real and durable—helping accelerate approval and construction of new generation capacity.
B. Behind-the-Meter
The second approach is more dramatic: behind-the-meter generation. Rather than relying on the grid, some companies are exploring dedicated power plants co-located with data centers. This model eliminates interconnection queue delays, reduces transmission losses, and provides complete operational control over power reliability. Some companies are integrating multiple power generation technologies to elevate the reliability of their BTM configurations: natural gas, diesel, renewables, and even nuclear. BTM also insulates communities and consumers from higher electricity rates by easing reliance on the central power grid, lowering peak demand, and sparing the public from the cost of expensive infrastructure buildouts.
“Why BTM? For the right situation: reliability, speed to power, and cost control.”
— Matthew Sallee, CFA, Senior Portfolio Manager, EVP, Head of Investments, Tortoise Capital
This represents a remarkable evolution. Energy access is no longer a facilities department question—it has become a boardroom-level strategic priority requiring capital allocation, long-term planning, and integration with core business objectives. Some tech companies are becoming energy operators.
Competitive Implications
The competitive implications are significant. Companies that successfully secure long-term power supplies gain positioning advantages in AI development. Those unable to secure sufficient electricity face real constraints on their ability to expand, train larger models, or deploy new services.
But securing power is only part of the equation. The question of supply sources remains central.
#3 Natural Gas and Nuclear: Powering the Buildout
Where will new electricity supply come from? This has become a central question in AI infrastructure discussions. Renewable energy sources have expanded substantially, but they face fundamental challenges meeting the specific requirements of large-scale data center operations.
AI workloads operate continuously, creating base-load demand requiring 24/7 power availability. Solar and wind are intermittent by nature, generating power only when environmental conditions permit. Battery storage technology is advancing, but cannot yet power multi-gigawatt data center operations through extended periods without sun or wind.
This operational reality is driving renewed focus on dispatchable generation sources capable of providing continuous, reliable power regardless of weather or time of day. Two energy sources meet these requirements: natural gas and nuclear.
| Natural Gas | Natural gas offers the fastest path to new baseload capacity. Modern combined-cycle gas plants, which efficiently generate power while re-using waste heat, can be permitted, constructed, and brought online in three to five years under favorable conditions. They provide flexible output that can ramp up or down to meet variable demand. Natural gas infrastructure, particularly pipeline networks, is already extensively developed across the United States. |
| Nuclear | Nuclear power provides different but complementary characteristics. Existing nuclear plants deliver consistent baseload generation with minimal operational variability and near-zero emissions. For data center operators seeking long-term contracts with stable pricing and reliability, nuclear represents an attractive option. The nuclear sector is experiencing renewed investment interest driven partly by AI infrastructure requirements. Small Modular Reactors (SMRs) represent an emerging approach that could reduce construction timelines to five to seven years while offering more flexible deployment than traditional large-scale nuclear plants. Several tech companies have made direct investments in SMR development, viewing them as potential medium-term solutions. |
“We think natural gas and nuclear will be those future supply sources.”
— Rob Thummel, Senior Portfolio Manager, Tortoise Capital
Solving Multiple Constraints
The constraint extends beyond generation. Transmission infrastructure, cooling systems, and grid interconnection all require substantial capital investment and multi-year buildout timelines. Energy infrastructure companies capable of delivering integrated solutions—generation, transmission, pipelines, and grid access—occupy strategically valuable positions in the AI economy.
This infrastructure requirement is reshaping how investors think about AI expansion.
#4 The Infrastructure Investment Opportunity
“It’s a new arms race… and energy is central to who wins.”
— Mark Marifian, Head of Product, Tortoise Capital
AI investment narratives typically emphasize chips, software, and the cloud. The focus is on companies building and deploying AI capabilities. The energy story represents a different access point to the same secular trend: investing in companies making AI deployment possible.
Energy infrastructure investments typically involve long-duration physical assets with established regulatory frameworks. Natural gas pipelines, power generation facilities, and transmission networks generate relatively predictable cash flows over multi-decade operating lives. What has changed is the strategic necessity of these assets.
Energy infrastructure is no longer simply utility-like infrastructure serving steady demand. It has become essential enabling infrastructure for one of the most significant technology transitions in decades. Companies operating natural gas production, pipeline networks, utility-scale generation, and grid infrastructure are positioned at the intersection of AI growth and physical constraint.
This infrastructure bottleneck is structural rather than temporary. AI adoption remains in early stages across most industries. As these technologies scale, electricity demand will compound while infrastructure buildout requires time. This creates a sustained opportunity window beyond quarterly earnings cycles or product launches.
The investment thesis is straightforward:
- AI growth requires physical infrastructure at unprecedented scale.
- Current supply cannot meet projected demand.
- Building new infrastructure takes years.
- Companies capable of delivering this infrastructure may capture durable value streams.
This is engineering necessity driving capital allocation.
Rethinking Where AI Value Accrues
The AI revolution extends beyond semiconductors and data centers to the essential infrastructure of energy supply. As AI scales from experimental to core, companies with access to abundant, reliable electricity will have structural and competitive advantages.
This creates unexpected investment opportunities deeply integrated with technology and innovation. Energy companies are becoming enablers of digital transformation, occupying strategic positions in value chains previously dominated by software and hardware.
From the perspective of risk, energy infrastructure investments involve long development timelines, regulatory considerations, and exposure to policy shifts around energy sources. Demand projections for AI infrastructure, while compelling, remain forecasts. These capital-intensive assets require patience and appropriate risk tolerance, characteristics typical of infrastructure investing generally.
The race to power AI may determine where value accrues in the digital economy. Energy sits at the center of that race. For advisors building portfolios positioned for the AI era, energy warrants evaluation not as a defensive allocation, but as strategic exposure.
Watch the Full Discussion: “Why Energy Now?”
Hear directly from Tortoise Capital’s senior portfolio management team as they explore the transformation of the energy sector, the drivers of surging electricity demand, and why energy has become essential to the AI revolution.
Important Information
Nothing contained in this communication constitutes tax, legal, or investment advice. Investors must consult their tax advisor or legal counsel for advice and information concerning their particular situation. This communication contains certain statements that may include “forward-looking statements.” All statements, other than statements of historical fact, included herein are “forward-looking statements.” Although Tortoise Capital believes that the expectations reflected in these forward-looking statements are reasonable, they do involve assumptions, risks and uncertainties, and these expectations may prove to be incorrect. Actual events could differ materially from those anticipated in these forward-looking statements as a result of a variety of factors. You should not place undue reliance on these forward-looking statements. This communication reflects our views and opinions as of the date herein, which are subject to change at any time based on market and other conditions. We disclaim any responsibility to update these views. These views should not be relied on as investment advice or an indication of trading intention. Discussion or analysis of any specific company-related news or investment sectors are meant primarily as a result of recent newsworthy events surrounding those companies or by way of providing updates on certain sectors of the market. Tortoise Capital, through its family of registered investment advisers, does provide investment advice to Tortoise-related funds and others that include investment into those sectors or companies discussed in this communication. As a result, Tortoise Capital does stand to beneficially profit from any rise in value of the sectors broadly discussed, including individual companies contained within.
Past performance is no guarantee of future results.