As we head into the final stretch of 2025, the explosive growth of artificial intelligence is reshaping not just technology, but the very infrastructure that powers our world. At Dark Stone Capital, our portfolio emphasizes high-growth semiconductors like NVIDIA (NVDA), which have thrived on AI demand. However, a looming challenge threatens to redefine the landscape: the AI energy crunch. With data centers consuming unprecedented amounts of electricity, power shortages could become the defining constraint for AI expansion starting in 2026. This post explores the intersection of AI growth and energy limitations, drawing on recent earnings insights, demand forecasts, and strategic investment shifts to help navigate what's ahead.

The Surging Energy Demands of AI Data Centers

AI's computational intensity is driving a massive uptick in electricity usage. Data centers optimized for AI workloads are projected to see electricity demand more than quadruple by 2030, with AI as the primary driver. Globally, these facilities could account for up to 8% of total power demand by the end of the decade, up from about 3% today. In the U.S. alone, electricity consumption from data centers is expected to surge 20-40% in 2025, with double-digit growth persisting through 2030.

This isn't abstract, it's already manifesting in recent earnings from major cloud providers, which serve as bellwethers for AI infrastructure needs. Amazon Web Services (AWS) reported Q2 2025 sales of $30.9 billion, up 17.5% year-over-year, with capital expenditures (capex) set to exceed $118 billion for the fiscal year—much of it directed toward data center expansions to support AI. Microsoft, meanwhile, highlighted Azure's 39% revenue jump in its FY25 Q4 (April-June 2025), with capex reaching $88.7 billion and plans for even higher spending in FY26 to bolster AI data centers. Google Cloud echoed this, posting $13.6 billion in Q2 2025 revenue and boosting its annual capex forecast to $85 billion, primarily for servers and technical infrastructure amid "strong and growing" AI demand.

These figures underscore how AI is fueling record investments in data centers, directly benefiting semiconductor giants like NVDA and TSMC. TSMC, for instance, has raised its 2025 sales growth guidance to 30% on robust AI chip demand, with forecasts pointing to 33% revenue growth through 2026. Yet, as these buildouts accelerate, energy constraints are emerging as a critical risk factor.

Forecasting the Bottlenecks: Power Shortages on the Horizon

Experts warn that electricity supply could become the most binding constraint on U.S. AI dominance. Elon Musk has highlighted this, predicting a major AI power crunch by 2026 as data centers scale to gigawatt levels. The International Energy Agency (IEA) projects data center electricity use could triple by 2030, driven by AI, while U.S. grids face potential shortfalls of up to 36 GW for compute by 2028.

Key challenges include:

• Grid Interconnection Delays: Queues for connecting new data centers to the grid can stretch 5–7 years, with transformer lead times at 36–60 months.
• Rising Demand vs. Supply: AI could double global electricity needs by 2030, straining infrastructure in critical sectors.
• Regional Vulnerabilities: In California, rolling blackouts have already exposed the fragility of a grid under pressure from aggressive EV mandates and AI expansion. In Texas, extreme weather and rapid AI cluster growth magnify risks for a notoriously unstable grid. And in Northern Virginia, “Data Center Alley” already consumes more electricity than some states, forcing utilities to delay new projects.

Recent discussions on X emphasize this urgency, with users noting that utilities are demanding curtailments during peaks and that fusion or advanced nuclear might be needed for long-term abundance.

Potential Regulatory Shifts and Their Impact

Governments are responding, but unevenly. In the U.S., policies like the "One Big Beautiful Bill Act" have reduced solar tax credits, potentially slowing renewable expansions. Meanwhile, China is surging ahead, adding over 277 GW of solar in 2024 alone. Regulatory changes could include:

• Mandates for data centers to fund grid upgrades or pay impact fees to avoid passing costs to consumers.
• Incentives for clean energy integration, such as Microsoft's deals for nuclear and geothermal.
• Stricter efficiency standards for AI models and hardware to curb demand.

These shifts could favor companies investing in sustainable power, while penalizing those reliant on fossil fuels amid carbon concerns.

Actionable Strategies: Hedging with Energy and Infrastructure Plays

For investors like us at Dark Stone Capital, the energy crunch presents both risks and opportunities. While NVDA and TSMC remain core holdings, diversifying into energy-resilient stocks can hedge volatility. Consider:

• Renewables Leaders: NextEra Energy (NEE) is committing $120 billion to solar and wind, positioning it as a key enabler for AI data centers. Brookfield Renewable (BEP) offers exposure to hydro and diversified clean energy.
• Grid and Infrastructure Tech: Companies like GE Vernova (GEV) and Constellation Energy (CEG) are surging on AI-driven demand, with GEV up 400% since IPO.
• Nuclear and Advanced Energy: Talen Energy (TLN) and Oklo (OKLO) are poised for growth in nuclear, with OKLO potentially seeing 10-25% short-term upside on contracts.

By allocating 10-20% of portfolios to these areas, we can mitigate risks from power volatility while capitalizing on AI's energy tailwinds.

Looking Ahead: Positioning for a Power-Constrained Future

The AI energy crunch isn't a distant threat it's already influencing Q3 and Q4 2025 earnings trajectories and could define 2026 market dynamics. At Dark Stone Capital, we're refining our strategies to balance AI upside with energy realities, viewing this as an opportunity to lead in thoughtful, forward-looking investments. As breaking news on grid expansions and regulatory moves unfolds, staying agile will be key.

Disclosure: Dark Stone Capital holds positions in NVDA and may initiate positions in mentioned energy stocks. This is not investment advice; consult a financial advisor.