Why gas prices could surge

Noreva's forecast rests on a collision of three forces happening at once :

• Rapidly rising tech-sector demand: Data centers already account for a significant share of U.S. electricity consumption, and RBC Capital Markets projects that data center natural gas demand alone could reach 6.1 billion cubic feet per day by 2030—substantially higher than previous estimates . PwC's projections are even more aggressive, with some scenarios reaching 11.5 Bcf/d by 2035 .
• Slow-growing domestic supply: U.S. natural gas production growth is decelerating at the same moment demand is accelerating—a combination that historically drives sustained price increases .
• Rising LNG export volumes: The buildout of liquefied natural gas export terminals, particularly along the Gulf Coast, is creating a new source of demand that competes directly with domestic power consumers .

The price surge will not be uniform. Noreva expects regional differentials to widen, especially in locations where hyperscaler data center clusters overlap with constrained pipeline capacity and high LNG export demand . Think Louisiana, Texas, and the broader Gulf region.

The financial exposure

Natural gas prices have historically been one of the cheapest and most predictable inputs for large-scale computing. That is changing. Since fuel costs represent roughly half the operating expenses of a gas-fired power plant, a tripling of gas prices would directly translate into significantly higher electricity costs for data center operators . The Noreva report implies that hyperscalers have not adequately hedged this exposure .

This price risk comes at a moment of extraordinary capital intensity. The five largest U.S. hyperscalers have collectively committed between $660 and $690 billion in capex for 2026 alone, according to Futurum Group research . Already, investors are watching profit margins closely. If energy costs consume a much larger share of AI revenue, the math behind the entire AI buildout could shift .

The climate and regulatory backlash

Even if the price forecast proves too dire, the strategic risk is not just financial—it is political and reputational.

Emissions are surging. Google's total carbon emissions jumped 25% year-over-year in 2025, and Amazon's rose 16% . Microsoft's emissions were up 23% compared to a 2020 baseline . These numbers are increasingly hard to reconcile with the net-zero pledges these companies made before the AI era.

Meta has already retreated from its clean energy commitments. In July 2026, the company withdrew from the RE100 clean energy initiative, a pact it had honored for over a decade. Meta stated that renewable energy sources "simply cannot meet its escalating near-term power demands" .

Local opposition is mounting. More than 500 U.S. counties or municipalities now actively restrict or block new data centers from being built, according to a review by Heatmap . In 2026 alone, more than 50 planned data centers have already been canceled after facing pushback from locals—more than twice as many as were canceled in all of 2025 . Amazon, Microsoft, and Google have each canceled large-scale projects after facing local opposition .

The hidden risk to consumers

Noreva's report also carries an implication that extends beyond the tech sector: higher wholesale gas costs could spill over into residential electricity rates in affected regions . This creates a potential consumer backlash that could accelerate the regulatory restrictions that are already multiplying.

The bottom line

The hyperscalers' pivot to natural gas solved a short-term reliability problem. Wind and solar cannot yet run the 24/7 AI workloads, and grid interconnection timelines of three to seven years are fundamentally mismatched with data center construction timelines of 12 to 18 months . But this short-term fix opened a long-term price exposure that Noreva's report suggests is poorly hedged and potentially debilitating if the forecast plays out .

As energy research firm Noreva wrote, the era of cheap gas may be ending, and the AI industry has built its foundation on that assumption .

Why gas prices could surge

Noreva's forecast rests on a collision of three forces happening at once :

• Rapidly rising tech-sector demand: Data centers already account for a significant share of U.S. electricity consumption, and RBC Capital Markets projects that data center natural gas demand alone could reach 6.1 billion cubic feet per day by 2030—substantially higher than previous estimates . PwC's projections are even more aggressive, with some scenarios reaching 11.5 Bcf/d by 2035 .
• Slow-growing domestic supply: U.S. natural gas production growth is decelerating at the same moment demand is accelerating—a combination that historically drives sustained price increases .
• Rising LNG export volumes: The buildout of liquefied natural gas export terminals, particularly along the Gulf Coast, is creating a new source of demand that competes directly with domestic power consumers .

The price surge will not be uniform. Noreva expects regional differentials to widen, especially in locations where hyperscaler data center clusters overlap with constrained pipeline capacity and high LNG export demand . Think Louisiana, Texas, and the broader Gulf region.

The financial exposure

Natural gas prices have historically been one of the cheapest and most predictable inputs for large-scale computing. That is changing. Since fuel costs represent roughly half the operating expenses of a gas-fired power plant, a tripling of gas prices would directly translate into significantly higher electricity costs for data center operators . The Noreva report implies that hyperscalers have not adequately hedged this exposure .

This price risk comes at a moment of extraordinary capital intensity. The five largest U.S. hyperscalers have collectively committed between $660 and $690 billion in capex for 2026 alone, according to Futurum Group research . Already, investors are watching profit margins closely. If energy costs consume a much larger share of AI revenue, the math behind the entire AI buildout could shift .

The climate and regulatory backlash

Even if the price forecast proves too dire, the strategic risk is not just financial—it is political and reputational.

Emissions are surging. Google's total carbon emissions jumped 25% year-over-year in 2025, and Amazon's rose 16% . Microsoft's emissions were up 23% compared to a 2020 baseline . These numbers are increasingly hard to reconcile with the net-zero pledges these companies made before the AI era.

Meta has already retreated from its clean energy commitments. In July 2026, the company withdrew from the RE100 clean energy initiative, a pact it had honored for over a decade. Meta stated that renewable energy sources "simply cannot meet its escalating near-term power demands" .

Local opposition is mounting. More than 500 U.S. counties or municipalities now actively restrict or block new data centers from being built, according to a review by Heatmap . In 2026 alone, more than 50 planned data centers have already been canceled after facing pushback from locals—more than twice as many as were canceled in all of 2025 . Amazon, Microsoft, and Google have each canceled large-scale projects after facing local opposition .

The hidden risk to consumers

Noreva's report also carries an implication that extends beyond the tech sector: higher wholesale gas costs could spill over into residential electricity rates in affected regions . This creates a potential consumer backlash that could accelerate the regulatory restrictions that are already multiplying.

The bottom line

The hyperscalers' pivot to natural gas solved a short-term reliability problem. Wind and solar cannot yet run the 24/7 AI workloads, and grid interconnection timelines of three to seven years are fundamentally mismatched with data center construction timelines of 12 to 18 months . But this short-term fix opened a long-term price exposure that Noreva's report suggests is poorly hedged and potentially debilitating if the forecast plays out .

As energy research firm Noreva wrote, the era of cheap gas may be ending, and the AI industry has built its foundation on that assumption .