BloombergNEF: U.S. Data Centers' Natural Gas Consumption Over the Next Decade Will Surpass Most Countries
nashnova research
BloombergNEF now projects U.S. data centers will add roughly 15 Bcf/d of natural-gas demand over the next decade — by 2035 exceeding every country's current consumption except China, Russia, Iran, and the U.S. itself. One industry is on track to out-consume most nations.
How big is 15 billion cubic feet per day?
BloombergNEF forecasts U.S. data-center power generation will require an additional ~15 Bcf/d of natural gas by 2035.
This means → a single industry's gas appetite will surpass the total national consumption of every country on Earth bar China, Russia, Iran, and the U.S. itself.
In plain terms = the world's fifth-largest gas "consumer" may not be a country — it may be a collection of server farms.
The figure has more than doubled from BloombergNEF's December estimate of 6.9 Bcf/d, signaling the sector is expanding far faster than analysts expected just six months ago.
Why do data centers default to natural gas?
Gas wins on three counts: abundant reserves, low extraction costs, and dispatchable power — gas turbines can start and stop on demand.
This means → data centers need 24/7 uninterrupted power, and gas-fired generation delivers exactly that flexibility; wind and solar cannot match it alone.
BloombergNEF estimates 69% of electricity for newly grid-connected data centers will come from gas — making it the de facto first fuel of the AI-compute buildout.
Where does the power sector rank in U.S. gas-demand growth?
By 2035, U.S. power-sector gas consumption is projected to reach 54 Bcf/d, up 18 Bcf/d from 2025.
In the demand-growth ranking, the power sector sits second — behind only the 21 Bcf/d added by new LNG export terminals along the Gulf Coast.
This reflects two forces pulling on U.S. gas simultaneously: LNG exports outward, data-center power inward — stacked together, the supply pressure is unprecedented.
Can supply keep up?
BloombergNEF projects U.S. producers will add 35 Bcf/d of output between 2025 and 2035 — but that still leaves a 11 Bcf/d gap versus projected demand.
In plain terms = output will grow sharply, yet not sharply enough — the shortfall amounts to roughly a quarter of total incremental demand.
Henry Eaton, lead author of the report, acknowledged the forecast carries "a pretty wide error bar — upside and downside," adding that BloombergNEF's power-demand estimate is "definitely not low, but it's not the highest on the market either."
Is the era of cheap U.S. gas really ending?
Wood Mackenzie declared in July that "the decade of cheap Henry Hub gas is coming to an end," projecting power-sector gas demand will rise by 17 Bcf/d through the mid-2030s — closely aligned with BloombergNEF.
Chronometer Holdings founder Matthew Smith, in a video viewed 1.6 million times, predicted "fierce competition for natural gas" by the end of this decade, warning that "the biggest losers will be U.S. consumers."
This means → if both firms' supply-demand forecasts materialize together, the structural floor under U.S. domestic gas prices will rise — affecting everything from household heating to industrial feedstock costs.
Gas supply — can it keep pace or not?
BULL
Reserves are deep
Kimmeridge co-founder Ben Dell argues U.S. fields still hold vast untapped reserves.
History favors supply
Dell notes the gas industry has consistently met rising demand while lowering inflation-adjusted costs.
BEAR
The gap is quantified
BloombergNEF calculates a 11 Bcf/d shortfall — production growth may not be fast enough.
Two demands pulling at once
LNG exports and data centers are competing for the same molecule, creating a complex challenge for domestic fields.
In plain terms = there is plenty of gas underground, but the question is not 'is it there?' — it is 'can it be drilled fast enough?' Whether supply cadence can outrun demand expansion is the single most important test for the U.S. gas market in the years ahead.
市场有风险，内容仅供研究参考，不构成投资建议。