NVDA Earnings Eve|NextEra Megadeal Redraws AI Power Map

The $67B Power Bet

The largest utility deal since Exxon bought Mobil is being driven not by demographic growth or rate-base expansion, but by the electricity demands of AI servers that do not yet exist. NextEra Energy's $67 billion all-stock acquisition of Dominion Energy, announced Monday, is a statement that the company believes hyperscaler capex commitments are real enough to justify a 23% premium — and a 5% hit to its own stock price on day one.

That immediate dilution is the tell. NextEra shareholders sold into the announcement while Dominion shareholders bought. Price action interpreted the deal as NextEra overpaying to acquire optionality, not current cash flow. What makes that institutional sell-off striking is the strategic arithmetic NextEra disclosed: the combined construction backlog stands at 130 gigawatts — more than the two companies' existing generation combined — enough to power 100 million homes. That backlog is almost entirely contracted to hyperscalers and industrial electrification, which means NEE is not acquiring Dominion's existing asset base so much as its permitting footprint and regulated utility territory in Virginia's data center corridor.

The key mechanism is Dominion's Virginia footprint. Loudoun County alone hosts the largest concentration of data centers in the world, and regulated utility zones compound in value when load growth is predictable and contracted. NextEra CEO John Ketchum said the deal is about being able to "grow while they grow" — matching hyperscaler expansion in real time rather than building ahead of demand. The combined entity would be the U.S. leader in total power generation and second nationally in nuclear output. But the market's immediate reaction — foreign and domestic institutional capital rotating out of NEE into Dominion on the spread trade — suggests investors are not yet pricing the 130-gigawatt backlog as fully executable.

What that rotation leaves open is whether the regulatory approval timeline compresses the deal's value before AI capex cycles forward.

The Lawsuit That Cleared a $134B Overhang

While the utility sector was absorbing the Dominion acquisition, a federal jury in Oakland delivered a verdict that removed a different kind of structural risk from the AI landscape — one that had been suppressing Microsoft's valuation for months. The jury dismissed Elon Musk's lawsuit against OpenAI and Sam Altman in under two hours, finding Musk had filed past the statute of limitations. The decision matters for MSFT because Musk's legal team had been seeking a $134 billion clawback from OpenAI and Microsoft jointly.

That number is not an abstraction. Microsoft holds roughly a 27% stake in OpenAI and had been named as a co-defendant on aiding-and-abetting grounds. A clawback of that scale, or even a structural unwinding of OpenAI's for-profit conversion, would have forced a revaluation of Microsoft's core AI partnership and potentially derailed OpenAI's planned IPO — in which Microsoft's stake would convert into public equity. The jury's unanimous decision closed that downside scenario. Microsoft shares moved up modestly on the verdict, with analysts citing relief buying rather than a new fundamental catalyst, as the litigation overhang had been discounting the stock's AI-related upside.

The mechanism deepening that the NEE chapter left unresolved is now visible: the capital that matters for AI infrastructure is not only flowing into power assets. It is also flowing into the software layer — specifically into Microsoft's OpenAI position, which the verdict has now de-risked for a potential IPO at OpenAI's current $852 billion private valuation. Retail and institutional holders who had been underweighting MSFT due to litigation risk absorbed today's verdict as a positioning signal. The question that follows is whether OpenAI's IPO timeline accelerates now that the largest legal obstacle has been cleared, and what that means for MSFT's stake valuation ahead of any lockup period.

NVDA's $725B Question

The two prior developments — the power infrastructure consolidation and the legal clearing of OpenAI's path — converge on a single earnings report after the close Wednesday. Nvidia reports fiscal Q1 results on May 20, and the stakes are unusually defined this cycle. Consensus calls for $79.17 billion in revenue, 80% year-over-year growth, and EPS of $1.78. But the actual question is not whether Nvidia clears that bar — it has beaten revenue estimates for 28 consecutive quarters — the question is whether management can confirm that the $725 billion in hyperscaler capex guidance upgrades disclosed after Q1 earnings season translates into durable Blackwell and Rubin demand.

The condition that changes the interpretation: Morgan Stanley raised its price target to $285 and KeyBanc to $300 ahead of the report, both expecting Blackwell shipments to accelerate from 150,000 to 200,000 units in Q1, adding between $5 billion and $7 billion in incremental revenue. CEO Jensen Huang has stated Nvidia has "high confidence visibility of $1 trillion-plus of Blackwell and Rubin" by end of 2027 — a forward claim that, if confirmed Wednesday, would make Wall Street's current $372 billion 2026 revenue estimate appear structurally low. The counter-signal is China: Huang acknowledged Nvidia's China market share has dropped to zero following the H20 export ban, with $17.1 billion in annual China revenue now permanently excluded. The Commerce Department has approved H200 sales to 10 Chinese firms, but Beijing is actively blocking those purchases while steering buyers toward domestic alternatives.

That zero China share is the variable the beat thesis cannot model away. If Nvidia's guidance absorbs the China revenue gap entirely through U.S. and international hyperscaler demand, institutional capital holding NVDA at a 27-times forward multiple has confirmation to add. If guidance implies any volume shortfall tied to China substitution, the same institutions that lifted the stock 21% year-to-date would reweight toward AMD and custom silicon plays. The verification threshold is the data center revenue guidance for Q2 — if it comes in above $75 billion, the Rubin ramp thesis holds; if it falls short, the China gap is showing up in the numbers. The NextEra megadeal established that power infrastructure is being built for AI demand that is real and contracted. The Musk verdict cleared the legal path for OpenAI's IPO. What Wednesday night determines is whether the chip layer at the center of both of those stories is still on the trajectory those deals were priced to assume.