Oracle Sets 2035 Carbon-Free Power Goal for AI Data Centers

Oracle has set a goal to match 100% of the electricity used by its Custom AI Data Centers with carbon-free electricity by 2035, as growing AI workloads increase pressure on power grids and data-center emissions. The company has also committed to reducing carbon emissions per megawatt of IT capacity by 20% by 2030 from a 2020 baseline. Oracle says its offices and public cloud data centers already match 100% of their annual electricity use with carbon-free power, while its newer AI infrastructure will be tracked separately because of its higher energy demands.

 

Oracle Is Separating AI Infrastructure From Its Existing Cloud Targets

 

Oracle’s climate strategy now draws a clear line between its existing public cloud and office operations and its Custom AI Data Centers, which include large, dedicated AI clusters and high-performance computing environments. The distinction matters because Oracle has already reached annual carbon-free electricity matching across its offices and public cloud data centers, while the newer AI facilities present a different scale of power demand.

The company’s 2035 target applies specifically to electricity used by those Custom AI Data Centers. At the same time, Oracle’s 20% reduction target for carbon emissions per megawatt of IT capacity gives it a second measure focused on how efficiently emissions are managed as computing capacity expands, rather than relying only on electricity procurement.

 

Read more: Schneider Electric, Greenly Expand SME Decarbonization Program

 

Large Power Procurements Are Following AI Growth

 

Oracle says it is already backing new grid and electricity projects in regions where its cloud and AI infrastructure is expanding. In Texas, the company is supporting more than 1.7 GW of wind energy, while in New Mexico it has issued a request for proposals to source 2 GW of carbon-free electricity. It is also working with utilities and developers on battery storage and new clean-power supply in Michigan and Wisconsin.

These projects show that Oracle’s AI strategy is increasingly tied to regional energy planning. Rather than treating data-center electricity as a procurement issue alone, the company is looking at generation, storage and grid capacity together, which will become more important as large AI facilities add significant new loads to local power systems.

 

Carbon Capture Is Being Explored for Fuel Cell Emissions

 

Oracle has also committed $1 million to research carbon capture and sequestration in New Mexico. The work will examine whether carbon dioxide produced by Bloom Energy fuel cells can be captured and stored permanently, while also looking at other decarbonisation options for those systems.

The investment is small compared with Oracle’s broader energy procurement plans, but it highlights a different part of the problem. Some infrastructure may continue relying on technologies that produce direct emissions while grid capacity and cleaner generation are being built, so Oracle is exploring ways to reduce those emissions during the transition rather than relying on electricity sourcing alone.

 

Explore OneStop ESG Marketplace: Building energy & decarbonization

 

Supplier and Travel Emissions Show Progress Outside Data Centers

 

Oracle says 100% of key suppliers representing more than 80% of its spend now have environmental programmes in place, while more than 80% have set emissions-reduction targets. The company also reports that employee air-travel emissions were 68% lower in 2025 than in 2019.

Those figures give the wider climate strategy more context, but the main challenge is shifting towards digital infrastructure. AI growth is increasing demand for power, cooling and grid connections, meaning future progress will depend increasingly on how Oracle builds and operates new capacity rather than only on reducing emissions from corporate operations.

 

The 2030 and 2035 Targets Will Show Whether AI Growth Can Be Decoupled From Emissions

 

Oracle’s next test is whether it can maintain rapid expansion in AI capacity while reducing emissions intensity by 20% per megawatt by 2030. After that, the 2035 carbon-free electricity goal for Custom AI Data Centers will show how successfully the company can secure enough cleaner power as its infrastructure footprint grows.

The company has already started building the energy supply around that expansion through wind procurement, storage projects and new power requests. The harder measure will be whether those investments keep pace with the rate at which AI infrastructure is added and whether lower emissions per megawatt translate into meaningful reductions across a much larger data-center portfolio.

 

Source: Oracle