Get access now
Sign up now to access private market investments on Citizen Mint’s platform.
POWER • INFRASTRUCTURE • ACCESS
Artificial intelligence has become one of the defining investment themes of the past several years. The public market gains have been real, but they have also been concentrated. J.P. Morgan Asset Management noted that the “AI-spawned Magnificent 7” drove 63% of S&P 500 returns in 2023, 55% in 2024, and 43% in 2025, numbers that underscore how much AI exposure many clients already carry through traditional market-cap-weighted portfolios.
For advisors, that raises a practical question: what does it mean to add more AI exposure on top of what clients likely already own?
The more useful framing may not be whether AI matters as a theme. Most thoughtful investors have settled that question. The harder conversation is about the broader investment ecosystem behind the AI buildout, one that extends well beyond chips, models, and the handful of technology companies dominating public indexes. Power infrastructure, data center real estate, energy storage, grid interconnection, private credit, and private companies that have not yet come to market all represent a different kind of participation in the same economy.
That is where private markets may offer something that public equities generally do not: differentiated exposure, with return profiles tied to income, collateral, real assets, or selective equity upside rather than continued multiple expansion in a concentrated group of mega-cap names.
The most important constraint limiting AI deployment may not be compute but electricity.
The International Energy Agency projects that global data center electricity consumption will roughly double from 485 terawatt-hours in 2025 to approximately 950 TWh by 2030, representing close to 3% of total global electricity demand by that date. Within that, AI-focused data centers are growing considerably faster. The IEA estimates that electricity demand from AI-specific facilities will triple over the same period, as energy-intensive inference and training workloads scale. Data center electricity demand rose 17% in 2025 alone, more than five times the 3% growth in overall global electricity demand that year.
In the United States, the scale of the shift is even more pronounced. According to the IEA’s analysis, U.S. data centers are on track to consume more electricity for processing data in 2030 than all energy-intensive manufacturing combined, including aluminum, steel, cement, and chemicals. Data centers are projected to drive nearly half of all U.S. electricity demand growth between now and 2030.
McKinsey estimates that more than $500 billion of data center infrastructure investment may be required through the end of the decade, excluding upstream transmission and distribution needs. Lead times for new power access in high-demand markets such as Northern Virginia can exceed three years, while some electrical equipment orders have stretched to two years or more.
This creates a tangible investment dynamic. When electricity access becomes a binding constraint on AI deployment, the infrastructure that delivers it (power generation, transmission, storage, and interconnection) shifts from commodity infrastructure to something closer to strategic input.
Even when a developer has the capital, land, and signed demand to build a data center or power project, they still need to connect to the grid. That process has become a significant obstacle.
Lawrence Berkeley National Laboratory’s most recent data show that more than 2,060 gigawatts of total generation and storage capacity were actively seeking grid connection as of the end of 2025, representing roughly twice the installed generating capacity of the current U.S. power plant fleet. The typical project reaching commercial operation in 2024 spent an average of 55 months in the queue, up from less than two years for projects that reached operation in the early 2000s. And historically, only about 13% of the capacity that entered interconnection queues between 2000 and 2019 ever reached commercial operation.
For investors, those numbers are worth sitting with. They do not mean that the buildout stalls. They mean that projects with established interconnection positions, executed agreements, and experienced development teams occupy a materially different risk position than projects still waiting for clarity.
Private capital has historically found meaningful roles in exactly these kinds of financing gaps: interconnection deposits, equipment procurement, pre-construction bridge financing, and capital for grid upgrades tied to confirmed commercial demand. The bottleneck is real, but it also creates opportunity for structured, collateral-backed lending that is not correlated to technology sector multiples.
Public AI exposure is typically expressed through the same cluster of mega-cap technology companies. Private market exposure can be structured quite differently, closer to the implementation layer, and with returns that may be driven by contractual cash flow, asset-backed income, or selective equity upside rather than market sentiment.
The key distinction is capital structure. Downside protection in private market investing does not come from the AI theme itself. It comes from where an investor sits in the capital stack, what collateral exists, how contracts are structured, and whether underwriting is grounded in asset-backed income, project economics, or venture-style equity appreciation. Each approach carries meaningfully different risk and return characteristics.
Power infrastructure lending: Energy projects tied to data center demand often require financing at various stages of development, including interconnection deposits, equipment procurement, construction costs, and refinancing of completed assets. Secured lending in this space may offer income with collateral or contractual protections, though key risks include project delays, permitting challenges, and counterparty quality.
Renewable power and battery storage: The U.S. Energy Information Administration expects a record 86 gigawatts of utility-scale generating capacity to be added to the grid in 2026, with solar accounting for 51% of planned additions and battery storage for 28%. Developers plan to add 24 gigawatts of utility-scale battery storage in 2026, more than 60% above the 15 gigawatts added in 2025. Real asset exposure to this buildout may offer cash flow durability, though merchant power risk, equipment costs, and interconnection delays are important underwriting considerations.
Data center infrastructure: CBRE’s North America Data Center Trends H2 2025 report found that primary market vacancy fell to a record low 1.4% at year-end 2025, even as primary market supply increased 36% year over year to meet accelerated hyperscale demand. Primary markets posted record net absorption of approximately 2,498 megawatts in 2025. Exposure here can include real estate, power distribution, cooling infrastructure, and site development, areas that connect to AI demand through physical capacity rather than software economics. Concentration risk among hyperscale tenants and ongoing power access challenges are among the factors to underwrite carefully.
Interconnection and grid access: For projects that have secured or are pursuing their place in the grid queue, shorter-duration infrastructure-linked financing may offer an alternative profile tied to specific project milestones, with refundability provisions and documentation quality as key variables.
Venture and growth equity: Many of the companies building the AI stack are still private, including firms focused on enterprise workflow automation, cybersecurity, vertical AI applications, data infrastructure, developer tools, energy technology, and compute optimization. Access to these companies may offer upside that is not available in public markets. Liquidity constraints, valuation risk, and the competitive dynamics of a well-funded sector are important considerations in any evaluation.
Venture debt and specialty lending: Financing AI and technology companies through structured debt rather than pure equity may offer income combined with warrants or other upside participation. Revenue quality, cash burn trajectory, and refinancing risk warrant careful attention in this segment.
There is a compelling case for including some private venture or growth exposure in the broader AI theme. But the sector’s size and momentum do not automatically make individual investments attractive.
According to the NVCA 2026 Yearbook, using PitchBook data, U.S. venture firms closed 15,352 deals worth $320 billion in 2025, a 51% increase in deal value from 2024 and the second-highest annual total on record. AI accounted for 65.4% of all deal value, up from roughly 50.9% in 2024. The top five AI companies collectively raised nearly $60 billion, and nontraditional investors (hedge funds, sovereign wealth funds, corporates, and endowments) participated in about 30% of deals while accounting for 83% of total investment value.
That concentration raises reasonable questions. Capital has flowed heavily toward a small number of large platforms, while the broader ecosystem of earlier-stage companies competes for a smaller share of the attention. Advisors evaluating private AI exposure should separate genuine innovation from momentum-driven capital formation. The underwriting question is whether a given company has durable customer relationships, a credible path to unit economics, defensible data or distribution advantages, and a valuation that leaves room for future return independent of sector sentiment.
The NVCA data also highlight a structural gap worth noting: 859 unicorn companies are currently valued at $4.34 trillion in aggregate, but only 30 to 40 actually achieved exits in 2025. Liquidity remains constrained, which matters for advisors managing clients against any timeline.
For many clients, AI exposure already exists, carried silently through market-cap-weighted equity allocations that have tilted heavily toward the same group of large technology companies. The planning question is whether that exposure is appropriately sized, whether it is too dependent on continued valuation expansion in public markets, and whether it leaves meaningful opportunity unaddressed.
Private markets offer a different way to participate in the same economic shift. The layers include secured lending tied to energy infrastructure, real assets serving data center demand, battery storage that supports grid reliability, interconnection-related financing, and selective exposure to private companies building the next wave of AI applications. Each layer carries a different combination of return potential, income, collateral, duration, liquidity, and risk.
None of this replaces careful due diligence or advisor judgment about suitability. Private market investments are complex, illiquid by nature, and appropriate only for investors who meet relevant eligibility requirements and can tolerate the associated risks. But for advisors thinking about the AI economy as a multi-layered investment theme rather than a single trade, the opportunity set is considerably broader than most public market portfolios reflect.
At Citizen Mint, this is the conversation we are built to support, helping advisors identify the infrastructure, financing, and private company opportunities that may represent the next phase of the AI buildout, with the rigor that institutional allocations deserve.
Sign up now to learn the benefits of private market investing for financial advisors
Sign up now to access private market investments on Citizen Mint’s platform.