GE Vernova (GEV) is a top pick to play the AI power crunch, with orders booked through 2031 and a recent $7 billion deal with Microsoft that could drive exponential growth in the next 6–12 months.
Bloom Energy (BE) offers a medium-term fix with its portable fuel cells used by Meta, but the stock faces near-term risk after a key gas pipeline permit was rejected.
Micron (MU) presents a contrarian opportunity, as memory prices are still rising and demand is sold out through 2027, yet shares have pulled back ~20% on sentiment rather than weak fundamentals.
The electricity bottleneck is a durable, multi-year theme, so watch for new power purchase agreements or turbine orders as catalysts for the sector.

Energy Infrastructure & Power Generation (AI Bottleneck)

• Elon Musk privately acquired APR Energy for ~$1 billion, a company that owns ~1 gigawatt of mobile gas turbines (equivalent to one nuclear reactor, or enough to power ~600,000 H100 GPUs).
• This purchase highlights the massive bottleneck between available electricity and actually getting it connected to data centers. The US has abundant energy, but grid infrastructure permitting, transformers, and regulation create 5–7 year delays.
• Data center power demand in the US is exploding: ~23 GW in 2023 → projected 46.5 GW by 2026, yet only 5–12 GW of new supply is expected to come online this year.
• Three main modular solutions to bypass the grid: solar (large footprint, slow permitting), nuclear (commercially not ready until 2030s), and on-site gas turbines (quick fix, 6–12 months). Musk’s APR purchase allows him to bring turbines directly to data centers “behind the meter,” accelerating GPU deployment.
• The race for power is now a critical layer of the AI infrastructure stack, and some capital is rotating out of memory stocks into this theme.

Takeaways

• The electricity bottleneck is durable and agnostic to which AI model or chipmaker wins – all data centers need power. This makes it a broad, long‑term investment theme.
• Public companies involved in fast‑tracking energy for data centers could benefit from the scramble for electrons.
• Watch for announcements of large‑scale power purchase agreements, new turbine orders, or modular deployments as catalysts for the sector.

GE Vernova (GEV)

• Described as an “old staple” of the power industry that also builds modern gas turbines, high‑voltage transformers, and grid equipment.
• Orders are booked through 2031, providing highly predictable revenue. In 2025, orders doubled year‑over‑year to $7.1 billion.
• Signed a $7 billion deal with Microsoft and counts OpenAI as a primary customer, both racing to power massive GPU clusters.
• The stock has risen ~300% over three years, with steady growth rather than exponential spikes. The hosts expect growth could “go exponential” in the next 6–12 months as AI power demand becomes more apparent.
• Positioned as a “TSMC of power” because it combines legacy grid expertise with the ability to supply modular, quick‑deploy solutions.

Takeaways

• Bullish sentiment: GE Vernova is seen as a reliable, long‑term beneficiary of the AI electricity crunch, with strong backlog and established customer relationships.
• Investors may watch for further large contracts, margin expansion, or an acceleration in order growth as proof that the power theme is gaining momentum.

Bloom Energy (BE)

• Produces solid oxide fuel cells that convert natural gas into electricity more efficiently than simple gas turbines. These units are portable and can last 4–7 years, providing a medium‑term fix for data centers waiting on grid connections.
• Used by Meta and several data center operators in Mexico as a way to bypass multi‑year grid delays.
• Risk factor: The stock recently dropped after New Mexico regulators rejected (for a second time) a gas pipeline permit. Even these quick‑deploy solutions face the same red tape that slows the broader grid, which may limit near‑term upside until permitting challenges are resolved.

Takeaways

• Bloom Energy offers a compelling medium‑term solution, but the permitting overhang is a real risk that could create volatility.
• Positive regulatory developments or new customer announcements could reignite enthusiasm, but the stock currently faces the same “red tape” bottleneck as the broader energy build‑out.

Memory Stocks (Micron MU / SK Hynix / Samsung)

• Memory stocks (particularly DRAM and NAND flash) have seen massive gains – some prices rose 300–500% in nine months – but shares have recently pulled back ~20% from highs.
• Despite the stock drop, memory prices are still rising: DRAM was up ~20% in July alone. Demand remains extreme, with SK Hynix already selling out its supply for the entirety of 2027 through long‑term agreements (LTAs) with 13–15 customers locking in 40% of next year’s profit.
• Fundamentals for companies like Micron remain exceptionally strong: forward earnings multiple of ~7x, revenue growth of 350%, and gross margins around 85%.
• The sell‑off appears driven by emotional rotation of capital into the “new, exciting” power trade after a euphoric run, rather than a deterioration in business conditions. The hosts consider the pullback potentially overdone and temporary.

Takeaways

• Contrarian opportunity: If memory demand remains structurally in shortage until ~2030 (as no new fabs relieve the bottleneck soon), current price dips might be seen as buying opportunities – but only for investors comfortable with high volatility.
• Key risks: sentiment‑driven selling can continue, and a rotation into energy may keep memory stocks under pressure for a while even if fundamentals stay robust.
• Monitor future memory contract prices and any signs of oversupply. As long as prices rise and LTAs lock in profits, the bear case lacks concrete evidence.

Nuclear & Other Energy Plays

• Nuclear is a compelling long‑term solution (Microsoft reactivated the shuttered Three Mile Island reactor and locked in a 20‑year purchase agreement), but no public pure‑play reactor companies are expected to come online before the 2030s. A private company, Valor, is working on modular nuclear, but remains early‑stage.
• Independent Power Producers (IPPs) that own plants and sell electricity into the open market are another indirect play. They can sign decade‑long contracts with hyperscalers and bypass local permitting by generating power on their own sites.

Takeaways

• Nuclear and IPPs are part of the longer‑term energy thesis, but near‑term actionable public opportunities are more concentrated in turbine and fuel‑cell providers like GE Vernova and Bloom Energy.
• Watch for IPOs or new public vehicles in modular nuclear and advanced energy storage, as these may become the next wave of AI infrastructure investments.