If Amazon doesn’t know how to eliminate carbon emissions, then who does?

Amazon’s chief sustainability officer said the company does not yet know how it will achieve its net-zero carbon target by 2040, despite maintaining the commitment. Amazon has 42GW of clean-energy capacity in its portfolio, but its carbon intensity increased in 2025 versus 2024 and the company is reportedly planning a 7.65GW natural-gas plant for an AI data-center campus. The project’s 35 turbines could emit up to 33 million tons of CO2 annually, highlighting a conflict between Amazon’s AI-driven growth ambitions and its climate commitments.

Analysis

The investable issue is not reputational damage alone; it is AWS’s marginal power cost and capacity-to-revenue conversion. If Amazon relies on dedicated gas generation to secure AI capacity, it may avoid the near-term revenue opportunity cost of constrained grid interconnection, but it also locks in fuel-price, carbon-policy and stranded-asset exposure that cleaner-powered hyperscale campuses can increasingly avoid. The market is likely to reward delivered AI capacity over the next 1-3 quarters, so this does not yet impair AMZN’s earnings multiple; the risk emerges over 6-18 months if power costs rise faster than AWS pricing or large customers demand auditable low-carbon compute.

The less obvious beneficiaries are gas-turbine and grid-equipment suppliers, plus merchant generators in constrained data-center regions: AI load growth raises the value of dispatchable capacity even where renewable procurement is expanding. That supports a selective long bias toward GE Vernova (GEV), Vertiv (VRT) and Eaton (ETN), although each already embeds substantial AI-infrastructure optimism. Renewable developers and storage providers may gain later, not immediately: hyperscalers’ inability to match hourly clean generation creates demand for firmed power, transmission and long-duration storage rather than simply more renewable-energy certificates.

Consensus may overread this as an ESG multiple risk for AMZN. The nearer-term interpretation is constructive for AWS: management is signaling that it will prioritize capacity availability, which protects cloud share during an AI supply bottleneck. The bearish version requires evidence that the power solution creates incremental capex, operating-cost pressure, permitting delays, or customer churn—not merely a gap between a long-dated climate target and a fully specified implementation plan.

Key falsifiers are AWS segment operating-margin guidance, disclosed power/capex commitments, regional gas basis and electricity-price moves, and customer procurement language around carbon-free cloud capacity. A material AWS margin decline or a capex step-up without corresponding backlog/revenue acceleration would turn the issue from narrative risk into an earnings risk.

Key Decisions for Investors

• Maintain AMZN as a hold/accumulate-on-weakness rather than initiate an ESG-driven short; reassess over the next 1-3 earnings reports if AWS operating margin falls materially while AI-related capex rises. The thesis fails if AWS backlog and revenue growth continue to absorb incremental power costs.
• Express the power-bottleneck theme through a 3-6 month basket long of GEV and ETN, sized modestly given elevated AI-infrastructure valuations; both benefit from incremental generation, grid and electrification spend regardless of the ultimate fuel mix. Take risk down if data-center capex guidance weakens or order/backlog growth decelerates materially.
• Prefer a relative-value pair of long GEV / short a broad clean-energy ETF such as ICLN for the next 3-6 months if seeking to isolate dispatchable-power and grid-capex demand from a lower-for-longer-rate-sensitive renewable-equity basket. Exit if policy support, falling storage costs, or hyperscaler contracted clean-power announcements materially narrow the timing advantage for thermal/grid equipment.
• Set an event watch for AWS disclosures on power procurement, dedicated-generation capex and carbon-free-energy matching. A quantified increase in self-generation or power costs without a parallel increase in AWS revenue guidance is the trigger to reduce AMZN exposure and consider downside hedges.