Accelerating Climate Action Across AMD

Throughout 2024 and into 2025, we worked with external advisors to develop our Climate Transition Plan. The process included interviews with key AMD executive leaders and department managers, updated approaches to governance and incentives, robust assessment of risks and opportunities and detailed implementation plans and goal setting. The Plan, which brings a forward-looking lens to our environmental sustainability pillars, aligns with industry standards outlining our ambition, actions and accountability.^[i]

Ambition

AMD continues to pursue a science-based emissions reduction goal for our operations, to work closely with Manufacturing Suppliers and industry groups to support a 1.5°C pathway, and accelerate product energy efficiency to help others to achieve their climate goals.^[ii] Collaboration is central to our approach, largely because AMD has a fabless business model, with external suppliers manufacturing our products and customers integrating them into their devices. The upstream and downstream engagements with these partners are key to addressing Scope 3 emissions that occur beyond our operations.

CDP estimates that global supply chain emissions are on average 11.4 times higher than operational emissions.^[iii] The Semi Climate Consortium report estimates 63% of GHG emissions across electronics devices occur during the use phase.^[iv] Building on our 2020–2025 supplier and product goals, AMD set new 2030 targets focused on supply chain and product use. These include a 25% reduction in carbon intensity for Manufacturing Suppliers^[v] and a 20x increase in rack-level server AI energy efficiency from 2024 to 2030.^[vi]

Action

Our Climate Transition Plan directly supports our decarbonization efforts in a manner that encourages collaboration and further aligns our corporate strategy with our climate strategy. It includes a robust risks and opportunities assessment that incorporates climate scenario analysis to identify key areas that could impact the company’s path to decarbonization and long-term resilience.

These risks and opportunities include:

• Transition risks, such as the potential implementation of carbon taxes, which could increase operational and supply chain costs. Additionally, as demand for renewable energy grows, suppliers may face rising electricity costs, which could indirectly affect AMD.
• Physical risks, such as extreme weather events, which could cause disruptions in our supplier operations.
• Opportunities for continued innovation in energy efficiency across AMD products and in enabling climate solutions that may lead to reputational benefits and increased product demand.

The Plan’s implementation strategy shows how we are addressing these risks and opportunities. Examples include:

• In our operations, we are managing energy use and increasing our investment in renewable energy sourcing.
• In the supply chain, we are prioritizing and engaging Manufacturing Suppliers on activities spanning direct energy, renewable energy sourcing and materials sourcing.
• In our products, we are incorporating energy efficiency into every aspect of product design with a focus on data center sustainability.

 Accountability

The Plan describes our governance structure underlying these efforts, as well as financial incentives to accelerate progress and promote accountability. For example, we demonstrate how incentives are linked to annual performance metrics, promoting accountability and progress toward our energy and climate objectives. In addition to our new 2030 climate-related goals, our Plan also shows how we are actively supporting industry collaborations that advance low-carbon economy aligned with a 1.5°C scenario.

Read our Climate Transition Plan

Originally published in AMD 2024-25 Corporate Responsibility Report.

Footnotes

^{[i] Industry standard reference is based on Transition Plan Taskforce disclosure framework published in October 2023: https://www.ifrs.org/content/dam/ifrs/knowledge-hub/resources/tpt/disclosure-framework-oct-2023.pdf (accessed May 23 ,2025).}

^{[ii] The AMD GHG goal is aligned with the Science-based Target initiative’s (SBTi’s) 1.5-degree minimum target ambition of 4.2% linear annual reduction. The SBTi criteria considers multiple climate scenario models from the IAMC and IEA.}

^{[iii] CDP, “Environmental supply chain risks to cost companies $120 billion by 2026,” 2021, https://www.cdp.net/en/press-releases/environmental-supply-chain-risks-to-cost-companies-120-billion-by-2026 (accessed May 23, 2025).}

^{[iv] Transparency, Ambition, and Collaboration: Advancing the Climate Agenda of the Semiconductor Value Chain, https://discover.semi.org/rs/320-QBB-055/images/Transparency-Ambition-and-Collaboration-BCG-SEMI-SCC-20230919.pdf (accessed May 23, 2025).}

^{[v] AMD calculates the carbon intensity of supply chain emissions using the numerator of total manufacturing supplier emissions (metric tCO2e including suppliers’ scope 1 and 2 emissions from operations, as well as their Scope 3 Category 1 emissions from purchased goods) and the denominator of AMD reported net revenue. For suppliers, AMD surveys our top ~95% of spend to gather directly reported data, and where needed utilizes spend-based estimates using CEDA emission factors.}

^{[vi] AMD based advanced racks for AI training/inference in each year (2024 to 2030) based on AMD roadmaps, also examining historical trends to inform rack design choices and technology improvements to align projected goals and historical trends. The 2024 rack is based on the MI300X node, which is comparable to the Nvidia H100 and reflects current common practice in AI deployments in 2024/2025 timeframe. The 2030 rack is based on an AMD system and silicon design expectations for that time frame. In each case, AMD specified components like GPUs, CPUs, DRAM, storage, cooling, and communications, tracking component and defined rack characteristics for power and performance. Calculations do not include power used for cooling air or water supply outside the racks but do include power for fans and pumps internal to the racks. Performance improvements are estimated based on progress in compute output (delivered, sustained, not peak FLOPS), memory (HBM) bandwidth, and network (scale-up) bandwidth, expressed as indices and weighted by the following factors for training and inference.}