Cold weather and data centres drive up US greenhouse gas emissions

US greenhouse gas emissions rose 2.4% in 2025 after two years of declines, driven by a nearly 7% increase in residential natural gas consumption during an unusually cold start to the year and a 13% surge in coal-fired power to meet higher electricity demand. Rapid growth in data centres and cryptocurrency mining boosted power needs, while higher natural gas prices (linked to large gas exports) caused fuel switching to coal even as solar generation jumped 34% and hybrid vehicle registrations rose 25%; analysts also noted a slowdown in coal plant retirements and said policy changes had limited impact on 2025 emissions.

Analysis

Market structure: Winners are data‑centre owners/lessors (Digital Realty DLR, Equinix EQIX), AI/GPU suppliers (NVDA) and US LNG exporters (Cheniere LNG) as persistent export demand supports a higher US natural gas floor; losers are gas‑dependent utilities facing higher input costs and marginal coal plants only temporarily profitable. Coal producers (Peabody BTU, Arch Resources ARCH) get an earnings uptick but limited pricing power — coal is now marginal fuel driven by spikes in Henry Hub rather than structural demand, so volume gains are volatile. Cross‑asset: higher fuel burn implies upside for gas/coal prices, positive commodity real yields and upside pressure on CPI — expect upward pressure on 10y yields (+10–30bps if winter remains cold) and a stronger USD on energy export strength.

Risk assessment: Tail risks include an abrupt regulatory clampdown on crypto/data‑centre power (state moratoria) or federal carbon pricing which would re‑price stranded coal/utility assets; probability medium but impact high (≥30% hit to coal equities). Time horizons: immediate (days) — weather volatility drives energy names; short (1–3 months) — data‑centre leasing and quarterly LNG shipments; long (12–36 months) — structural natural gas floor from sustained LNG exports vs. accelerated renewables. Hidden dependencies: pipeline constraints, regional RTO dispatch rules and global LNG cargo availability can flip economics quickly. Key catalysts: NOAA winter forecasts, weekly EIA gas storage, Cheniere shipping cadence over next 90 days.

Trade implications: Establish 1.5–3% long positions in DLR and EQIX (split) to capture leasing tightness, with 6–12 month horizon and 10% stop; add 2% long CHR (Cheniere LNG, ticker LNG or LNG equivalents) as a hedge if Henry Hub >$4.50/MMBtu persists for 3+ months. Short 1–2% positions in BTU or ARCH as a tactical pair vs DLR (pair: long DLR, short BTU) — close if coal EUR/USD‑adjusted premium compresses by 20% or Henry Hub drops below $3.50. Buy 3–6 month NVDA calls (one‑to‑two month IV watch) to play sustained AI/data‑centre capex; alternatively sell covered calls to harvest elevated premia.

Contrarian angles: Consensus may overstate a sustained coal revival — history (2015 gas rebound) shows coal reversals when Henry Hub falls; thus coal longs are likely overstretched. Conversely, the market underprices structural upside in US gas producers and LNG shippers from durable export growth — consider selective long exposure to midstream producers (EQT, Cheniere) on dips. Unintended consequence: data‑centre demand accelerates grid modernization — buy selective transmission/storage names (NextEra Grid / utilities with regulated ROE catalysts) on pullbacks. Monitor thresholds: Henry Hub <$3.50 or >$5.00 and weekly EIA storage +/-10% vs 5‑yr avg as trade triggers.

Key Decisions for Investors

• Establish a 2–3% long position split between Digital Realty (DLR) and Equinix (EQIX) over 6–12 months to capture data‑centre leasing tightness; set a 10% stop‑loss and review after quarterly earnings or material guidance changes.
• Initiate a 2% tactical long in Cheniere Energy (LNG) / US LNG exporters if Henry Hub remains >$4.50/MMBtu for four consecutive weeks; target 6–12 month hold, take profits if Henry Hub drops below $3.50.
• Open a 1.5% short position in Peabody Energy (BTU) or Arch Resources (ARCH) paired with the DLR long (long DLR, short BTU) to express view that coal gains are marginal/temporary; close if BTU outperforms coal margin by 20% or Henry Hub falls < $3.50.
• Buy 3–6 month NVDA call options (size 0.5–1% notional) to play continued AI/data‑centre GPU demand; if implied volatility rises >30% above 90‑day realized, consider selling covered calls instead.
• Rotate 3–5% of discretionary capital from long‑duration growth into solar manufacturers (FSLR) and regulated transmission/storage names if weekly EIA gas storage deviates ±10% from 5‑yr avg or NOAA issues a prolonged cold forecast for next 60 days.