Impact in Action | Q2 2026
Artificial intelligence is often described as a software revolution, but its growth is creating a very physical set of demands. Behind every AI model is a network of data centers, power lines, cooling systems, construction materials, and electricity contracts. As usage expands, the infrastructure required to support AI is becoming one of the most important environmental questions facing the technology sector.
Many of the same companies leading the AI buildout have also spent the past several years building reputations as climate leaders. Microsoft, Amazon, Google, and other industry leaders have made ambitious public commitments, invested heavily in renewable energy and carbon credits, and helped push corporate sustainability from a niche concern into a mainstream business priority. Now those commitments are being tested by a level of electricity demand that few expected when the original goals were set.
The question is not whether these companies still care about sustainability; it is whether their earlier targets can remain credible as AI changes the scale, timing, and complexity of their energy needs.
The Scale of the Challenge
Training and running large AI systems requires sustained, intensive computing power at a scale that traditional cloud services have never approached. The result is a paradigm shift in electricity demand. Global data center electricity use is currently on track to more than double by 2030, climbing from about 415 terawatt-hours in 2024 to roughly 945 terawatt-hours, which is an amount slightly larger than Japan currently uses to power its entire economy.[1] Data centers already account for an estimated 4%-5% of total U.S. electricity consumption, and some industry analysts project that by 2030, several states with heavy concentrations of data centers could see that figure climb above 20% of local electricity usage.[2]
New renewable energy generation cannot meet this surge alone, at least not on the current timeline. Utilities are responding in part by building new natural gas plants and delaying the retirement of older coal facilities, with the development of new U.S. gas-fired capacity nearly tripling in 2025. Large technology companies are attempting a difficult balancing act: increasing near-term fossil-fuel power consumption while trying to maintain their long-term carbon neutrality goals that were set before AI’s power needs were fully understood.
Prior Promises, New Challenges
The following examples show how rapidly evolving technology and power needs are complicating the path forward for companies that have been highly visible in their climate efforts.
- Alphabet, the parent company of Google, has reported emissions nearly 50% above its 2019 baseline, a shift it attributes primarily to data center electricity needs tied to AI. Google still describes its 2030 net-zero goal as its target, even as executives have acknowledged that AI’s growth makes reaching that goal on schedule considerably harder.[3]
- Amazon, one of the world’s largest corporate purchasers of renewable energy, reported a 16% increase in carbon dioxide emissions from 2024 to 2025, representing its largest single-year jump since it began tracking this data.[4] Electricity-related emissions alone rose 34%, which the company attributed largely to data center growth. Amazon is keeping its 2040 net-zero commitment in place even as it acknowledges that near-term growth will likely continue pushing emissions higher before any longer-term decline takes hold.
- Microsoft committed in 2020 to becoming carbon negative by 2030. Instead, its total emissions have risen 23.4% above that baseline, largely driven by a nearly 31% jump in supply-chain emissions tied to data center construction and AI hardware. The company’s reporting points to meaningful progress elsewhere, including land protection and recycling targets it says it has met or exceeded ahead of schedule.[5]
Taken together, these examples show a similar pattern: the companies are not abandoning their climate goals, but the path from commitment to execution has become more complex, more capital-intensive, and more dependent on downstream infrastructure decisions.
Carbon Credits and Net-Zero
For many large companies, achieving “net-zero” goals depends not only on reducing their own emissions, but also on using carbon credits to address emissions that are difficult or impractical to eliminate.
In simple terms, a carbon credit allows a company to pay for an emissions reduction that happens somewhere else and attribute it to itself. A company might purchase credits tied to projects such as forest restoration, methane capture, renewable energy development, or direct carbon removal technology, then describe its remaining footprint as neutralized.
The financialization of these carbon credits simultaneously provides a valuable avenue for financing carbon-reduction efforts as well as a way for large companies to appear responsible without fully eliminating emissions. The important caveat is that not all credits carry equal weight. A high-quality credit should represent a reduction that is real, independently verified, and would not have happened anyway. If a forest was never genuinely at risk of being cleared, or a renewable project would have been built regardless of credit revenue, the credit may add little real climate benefit.
Genuine Progress Towards Long-Term Goals
As AI adoption increases demand for electricity, land, materials, cooling systems, and new infrastructure, companies will need to explain not only what their long-term targets are, but also how they plan to make progress in a changing environment. Adaptation and clear communication will be central as AI usage and applications continue to evolve rapidly.
While the path from ambition to execution has become more complex, that does not mean that climate commitments should be disregarded. Rather, the question shifts to what tools companies can use along the way. Renewable energy purchases, efficiency improvements, and carbon credits are all likely to play a role. Achieving long-term climate goals will depend on whether those efforts represent credible progress or simply create the appearance of progress while underlying emissions continue to rise.
Resources
[1] Energy and AI. IEA
[2] Clean Energy Resources to Meet Data Center Electricity Demand. US Dept of Energy
[3] Alphabet Inc: Report or Address Data Center Climate Goals. As You Sow
[4] Amazon reveals it used more energy than New Zealand in 2025 as its carbon footprint rises 16%. TechRadar
[5] Corporate Sustainability Commitments. Microsoft
