Analysis
Google's total carbon emissions rose 25% and Amazon's rose 16% over the past year, according to the companies' own sustainability disclosures covered by TechCrunch on July 2, 2026, with both hyperscalers explicitly acknowledging that AI-driven energy use is now putting their publicly stated net-zero targets at real risk rather than remaining a manageable, offsettable side effect of growth.
The reporting identifies data center construction and equipment, particularly semiconductor manufacturing, as a primary driver of the increase. Chip fabrication is extremely energy-intensive, and many of the world's leading-edge chip factories are located in Asia, where electrical grids remain dominated by fossil fuels -- meaning a meaningful share of the emissions growth tied to AI buildout is happening upstream, in the supply chain, rather than in either company's own directly operated data centers. Compounding the problem, many of the specialized chemicals used in semiconductor manufacturing are themselves potent greenhouse gases, capable of warming the atmosphere thousands of times more per unit than an equivalent amount of CO2.
Both companies are responding, at least in part, by investing in natural gas power plants to keep pace with AI's power demands -- a notable shift for organizations that have spent years publicly emphasizing renewable energy purchasing as their primary decarbonization strategy. That pivot amounts to an implicit admission that a pure renewables-plus-battery-storage approach can no longer scale fast enough to match the pace at which AI infrastructure is being built out, at least not without accepting real near-term reliability or timeline risk.
“The reporting identifies data center construction and equipment, particularly semiconductor manufacturing, as a primary driver of the increase.”
To actually deliver on their existing net-zero pledges, the reporting notes, both companies will need to substantially ramp up renewable energy purchases, invest heavily in lower-carbon steel and cement manufacturing for data center construction, and purchase large volumes of carbon removal credits -- a materially more expensive and operationally complex path than the renewable-energy-matching strategies both companies have relied on to date.
The disclosure lands the same week Google separately reported its data center electricity use rose 37% in 2025, the largest single-year increase in company history, reinforcing that the emissions and raw energy-consumption stories are two closely related but distinct dimensions of the same underlying AI-power-demand problem -- one measuring carbon output, the other measuring raw electricity draw, both trending sharply upward at once.
For founders and investors in energy, grid infrastructure and climate tech, two of the best-resourced companies in the industry both posting double-digit emissions growth in the same year is concrete, first-party validation that AI's power and carbon costs are now a first-order strategic constraint, not a marginal externality -- exactly the dynamic driving continued investor interest in energy infrastructure deals elsewhere in this issue. For ESG-focused allocators, the shift toward natural gas investment despite public net-zero commitments is a clear signal that near-term operational pragmatism is winning out over long-stated climate targets at even the most prominent tech companies.
What to watch: whether Google and Amazon's natural-gas investments continue expanding as a stopgap for grid reliability, whether Scope 3 emissions from chip manufacturing keep rising as a share of both companies' total footprints, and whether other hyperscalers' 2026 sustainability disclosures show similarly accelerating emissions tied to AI buildout.
