Google and Amazon both released sustainability reports revealing a troubling trend: their carbon emissions are climbing sharply, and artificial intelligence is largely to blame. Google's total carbon emissions increased 25% compared to the previous year, while Amazon's rose 16%. Neither company explicitly names AI as the culprit, but a close examination of their reports reveals that the explosive growth in AI infrastructure is making it significantly harder for both tech giants to meet their net-zero climate commitments.

Why Are Tech Giants' Emissions Suddenly Spiking?

For years, Google and Amazon managed their carbon footprints by purchasing renewable energy to power their data centers and offices. That strategy worked reasonably well when data center growth was modest and predictable. But the AI boom has fundamentally changed the equation. Both companies acknowledge that their energy consumption has surged as AI adoption has accelerated, and they're now struggling to keep pace with the power demands.

The problem isn't just the electricity that AI systems consume. The real climate challenge comes from what's called "Scope 3 emissions," a category that includes pollution a company doesn't directly control but is responsible for through its supply chain and products. For Google and Amazon, this includes the manufacturing of graphics processing units (GPUs), semiconductors, and the construction of new data centers themselves.

Google's Scope 3 emissions doubled since 2019, increasing by 2.1 million metric tons in the most recent year alone. Amazon's Scope 3 emissions spiked even higher, driven largely by capital goods like data centers and warehouses. Amazon noted in its sustainability report that it "added more data center capacity globally than any other company" in 2025, including more than 1.2 gigawatts of capacity in the final quarter alone.

Amazon

What Makes AI Infrastructure So Carbon-Intensive?

The carbon footprint of AI infrastructure stems from multiple sources, each presenting its own environmental challenge:

• Semiconductor Manufacturing: Producing the GPUs and memory chips that power AI systems requires enormous amounts of energy. Many of the world's leading chip factories are located in Asia, where electrical grids still rely heavily on fossil fuels, making the manufacturing process particularly carbon-intensive.
• Chemical Emissions from Chip Production: The chemicals used in semiconductor manufacturing are potent greenhouse gases, capable of warming the atmosphere thousands of times more than an equivalent amount of carbon dioxide.
• Data Center Construction: Building new data centers requires steel and cement, both heavy polluters. While startups are developing low-carbon alternatives, these technologies aren't yet ready to scale to the levels that tech companies need.
• Fossil Fuel Fallback: As renewable energy sources struggle to keep pace with AI's power demands, tech companies are increasingly turning to natural gas power plants to fill the gap, reversing years of progress toward cleaner energy.

How Are Tech Companies Responding to the Challenge?

Both Google and Amazon have pledged to achieve net-zero carbon emissions in the coming years, but their recent reports suggest those goals are now in serious jeopardy. To meet their commitments, these companies will need to make substantial changes across multiple fronts. The path forward requires simultaneous action on several interconnected problems, each with significant financial and operational implications.

The companies face a complex set of challenges that can't be solved with a single approach. Renewable energy purchases alone won't be enough, especially as data center power demands continue to climb. Instead, tech giants will need to invest heavily in emerging technologies and practices that can reduce emissions across their entire supply chain.

According to the TechCrunch analysis of their sustainability reports, both companies will need to ramp up renewable energy purchases, invest in advanced steel and cement manufacturing techniques, and purchase millions of tons of carbon removal credits to offset unavoidable emissions. These aren't minor adjustments; they represent fundamental shifts in how these companies source energy, build infrastructure, and account for their environmental impact.

The challenge is particularly acute because many of these solutions are still emerging. Low-carbon steel and cement production exist but aren't available at the scale needed. Carbon removal technologies are improving but remain expensive and unproven at the massive scale required. Meanwhile, the demand for AI computing power continues to accelerate, making the window for action increasingly narrow.

Steps Tech Companies Can Take to Reduce AI's Carbon Footprint

• Accelerate Renewable Energy Procurement: Expand long-term contracts for wind and solar power, and invest in battery storage to smooth out intermittency issues that currently force companies to rely on fossil fuel backup power.
• Support Low-Carbon Manufacturing: Fund research and early deployment of green steel and cement production methods, and commit to purchasing these materials even at a premium to create market demand.
• Optimize AI Model Efficiency: Invest in research to make AI models more computationally efficient, reducing the raw processing power required to train and run them.
• Invest in Carbon Removal: Purchase carbon removal credits and fund direct air capture technology development to offset emissions that can't be eliminated through other means.
• Diversify Energy Sources Responsibly: If natural gas is necessary as a bridge fuel, prioritize facilities with carbon capture technology and set clear timelines for transitioning away from fossil fuels.

The situation isn't hopeless. Both companies acknowledge the problem and have the resources to address it. But the scale of the challenge is becoming impossible to ignore. As AI continues to reshape the technology industry, the environmental cost of that transformation is becoming increasingly visible, and increasingly difficult to manage within existing climate commitments.