Rhodium sees US emissions dropping to 2030, then hitting an AI-and-oil wall
Clean energy buildout buys a decade of declines, but post-2030 trajectory faces conflicting pressures from data centers and new export capacity.
The US is heading for a decade of measurable emissions cuts, then the trajectory goes opaque. Rhodium Group’s annual "Taking Stock" report, released Wednesday (2026-07-29), projects domestic greenhouse gas emissions will keep falling through 2030, driven by a clean energy buildout that sees developers completing around 50 GW of solar, storage and wind annually.5
That is an improvement on Rhodium’s 2023 report, which had already baked in Biden-era Inflation Reduction Act subsidies and forecast a 29% to 42% drop by 2030. The new baseline lands inside that range. Lead author Hannah Kolus described the scenarios as "a reasonable bound around where emissions are headed" rather than a definitive forecast.5
The catch is Biden’s Paris Agreement pledge of 50% reductions by 2030. That target is now out of reach. The arithmetic is simple: at the high end of the Rhodium range — 42% — the remaining gap is too large to close in four years without either a policy revolution or a demand collapse that no one is forecasting.5
The real uncertainty starts after 2030. Two forces pull in opposite directions. The first is AI-driven electricity demand, which Rhodium flags as a major variable that could push emissions higher as data centre loads rise. The second is the war with Iran, which has scrambled energy markets in ways that cut both ways for the transition.5
That conflict has already reshaped production patterns. US oil and gas companies initially held back on new drilling, uncertain whether President Trump’s promises of a quick resolution were credible. By late April to early May, that hesitation ended. EIA data cited by analysts show US crude output rising from 13.6 million bbl/day in 2026 to a record 14.2 million bbl/day by mid-2027.1
Exports hit their highest levels ever. The IEA ordered its members to release 412 million barrels from strategic reserves. China, hedging against supply risk, slashed imports by 20%. The net effect: more American barrels into global markets, displacing some higher-carbon sources but also locking in long-cycle fossil infrastructure that will operate well past 2030.1
Yet the same war has accelerated parts of the clean energy push. China’s photovoltaic cell exports surged 346% year-on-year to reach $39.96 million, while lithium-ion battery exports rose 20.8% to $780 million.3
A separate report from June (2026-06-30) showed the US led global CO₂ emissions growth in 2025, accounting for roughly 30% of the increase. Coal consumption, widely feared to stage a comeback across major economies, rose less than 2% — nowhere near enough to offset the US-led surge in oil and gas output.4,1
ICE Brent crude front-month traded at $90.15/bbl on Wednesday (2026-07-29), up 0.72%. ICE TTF front-month moved to €60.52/MWh, a 4.72% gain. The VIX closed at 20.66, up 13.52%. [LIVE PRICES]
The collision matters for positioning. The clean energy buildout is real and measurable — 50 GW per year of solar, storage and wind is enough to shift the generation mix materially through 2030. But beyond that, the AI load curve and the inertia of new oil and gas export capacity create a risk of emissions plateauing — or even rising again — just when deeper cuts would be needed.5,1
Analysts say countries emerging from the Iran conflict may prioritise stable energy supplies above transition timelines, repeating the pattern seen after the 1970s oil shocks. "I don’t see this as being pro-transition or anti-transition," one analyst told E&E News. "I see it as states looking for what works and minimising their exposure."2
The report’s middle scenario — that the US stays within Rhodium’s bounds for the rest of the decade — is the base case. The question is not 2030. It is 2032, when the dual pressures of AI-driven load growth and post-war fossil investment will either have been absorbed by a scaled-up grid or will have blown through the baseline.5