Advanced Micro Devices, Inc. (AMD) Q3 2025 Earnings Call Transcript

Access earnings results, analyst expectations, report, slides, earnings call, and transcript.

Overview

The earnings call summary indicates strong financial performance, with a 28% YoY revenue growth and optimistic guidance, particularly in the data center and AI segments. The Q&A section reveals positive customer response and strategic partnerships, like with OpenAI, enhancing AMD's market position. However, management's avoidance of specific details on certain metrics slightly tempers the sentiment. Overall, the positive growth outlook and strategic advancements suggest a likely positive stock price movement.

Key Financial Performance

Revenue Revenue grew 36% year-over-year to $9.2 billion. This growth was driven by broad-based demand across data center AI, server, and PC businesses.

Net Income Net income rose 31% year-over-year. This increase was led by record EPYC, Ryzen, and Instinct processor sales.

Free Cash Flow Free cash flow more than tripled year-over-year. This was driven by record sales of EPYC, Ryzen, and Instinct processors.

Data Center Segment Revenue Revenue increased 22% year-over-year to a record $4.3 billion. This was led by the ramp of Instinct MI350 Series GPUs and server share gains.

Client and Gaming Segment Revenue Revenue increased 73% year-over-year to $4 billion. This was driven by strong demand for client and graphic processors and stronger sales of console gaming products.

Gaming Revenue Gaming revenue rose to $1.3 billion, up 181% year-over-year. This was due to higher semi-custom revenue and strong demand for Radeon GPUs.

Embedded Segment Revenue Revenue decreased 8% year-over-year to $857 million. The decline was attributed to end market mix and lower revenue.

Gross Margin Gross margin was 54%, up 40 basis points year-over-year. This increase was primarily driven by product mix.

Operating Income Operating income was $2.2 billion, representing a 24% operating margin. This was an increase from the prior year, driven by higher revenue.

Diluted Earnings Per Share Diluted earnings per share were $1.20, up 30% year-over-year. This was driven by higher revenue and profitability.

You have reached the limit Upgrade to access full content

Get started

Operating Highlights

Instinct MI350 Series GPUs: Revenue grew year-over-year driven by the sharp ramp of MI350 Series GPU sales and broader MI MI300 Series deployments. Multiple MI350 Series deployments are underway with large cloud and AI providers.

5th Gen EPYC Turin processors: Adoption accelerated rapidly, accounting for nearly half of overall EPYC revenue in the quarter. Sales of prior generation EPYC processors were also robust.

Ryzen 9000 processors: Desktop CPU sales reached an all-time high with record channel sell-in and sellout, driven by robust demand for gaming, productivity, and content creation applications.

Versal Prime Series Gen 2 adaptive SoCs: Began shipping to lead customers, extending leadership across adaptive and x86 computing.

Ryzen Embedded 9000 series: Introduced with industry-leading performance per watt and latency for robotics, edge computing, and smart factory applications.

Cloud adoption of EPYC CPUs: Hyperscalers launched more than 160 EPYC-powered instances in the quarter, including new offerings from Google, Microsoft Azure, and Alibaba. Adoption of EPYC in the cloud by large businesses more than tripled year-over-year.

AI partnerships and deployments: Announced a multiyear agreement with OpenAI to deploy 6 gigawatts of Instinct GPUs, with the first gigawatt of MI450 Series accelerators coming online in 2026. Oracle will also deploy tens of thousands of MI450 GPUs starting in 2026.

Gaming market growth: Gaming revenue increased 181% year-over-year to $1.3 billion, driven by semi-custom revenue and strong demand for Radeon GPUs.

Record revenue and profitability: Revenue grew 36% year-over-year to $9.2 billion, with net income rising 31% and free cash flow more than tripling.

R&D investments: Operating expenses increased 42% year-over-year, driven by aggressive investments in R&D to capitalize on AI opportunities.

ZT Systems manufacturing business sale: Completed the sale to Sanmina and entered a strategic partnership to accelerate large customer deployments of rack-scale AI solutions.

Next-generation Venice processors: On track to launch 2-nanometer Venice processors in 2026, with substantial gains in performance, efficiency, and compute density.

Helios rack-scale AI platform: Development progressing rapidly, integrating MI400 Series GPUs, Venice EPYC CPUs, and Pensando NICs to redefine performance and efficiency at data center scale.

OpenAI partnership: Establishes AMD as a core compute provider for OpenAI, with potential to generate over $100 billion in revenue over the next few years.

You have reached the limit Upgrade to access full content

Get started

Risk or Challenges

Data Center AI Business: The company is heavily investing in AI opportunities, but there are risks associated with the rapid scaling of data center AI business, including execution risks in launching next-gen MI400 Series accelerators and Helios rack scale solutions in 2026. Additionally, the reliance on partnerships with hyperscalers and AI companies could pose challenges if these relationships falter.

Client and Gaming Segment: While the segment showed strong growth, there is a risk of over-reliance on the gaming and PC processor markets, which are subject to cyclical demand and competitive pressures. A decline in gaming revenue is also expected in the next quarter.

Embedded Segment: The segment experienced an 8% year-over-year revenue decline, indicating potential challenges in maintaining growth in this area. The decline in operating income due to lower revenue and end-market mix also highlights vulnerabilities.

Geopolitical and Regulatory Risks: The company’s financial outlook excludes revenue from shipments of MI308 GPU products to China, indicating potential regulatory or geopolitical challenges that could impact revenue streams.

R&D and Operating Expenses: Operating expenses increased by 42% year-over-year due to aggressive investments in R&D and go-to-market activities. While these investments aim to capitalize on AI opportunities, they also pose a financial risk if the expected returns are not realized.

Supply Chain and Manufacturing: The sale of the ZT manufacturing business to Sanmina and reliance on Sanmina as the lead manufacturing partner for Helios could introduce supply chain risks if the partnership faces disruptions or fails to meet demand.

You have reached the limit Upgrade to access full content

Get started

Guidance & Outlook

Cloud Demand: Cloud demand is expected to remain very strong as hyperscalers significantly increase their general-purpose compute capacity to scale AI workloads. Many customers are planning substantially larger CPU build-outs over the coming quarters to support increased AI demands.

Next-Generation Processors: AMD remains on track to launch its next-generation 2-nanometer Venice processors in 2026. Venice silicon is performing well in labs, delivering substantial gains in performance, efficiency, and compute density. Multiple cloud OEM partners have already brought their first Venice platforms online.

Data Center AI Business: The data center AI business is entering its next phase of growth with customer momentum building ahead of the launch of next-gen MI400 Series accelerators and Helios rack scale solutions in 2026. These solutions are expected to redefine performance and efficiency at data center scale.

OpenAI Partnership: AMD announced a multiyear agreement with OpenAI to deploy 6 gigawatts of Instinct GPUs, with the first gigawatt of MI450 Series accelerators scheduled to come online in the second half of 2026. This partnership is expected to generate well over $100 billion in revenue over the next few years.

Oracle Partnership: Oracle will be a lead launch partner for the MI450 Series, deploying tens of thousands of MI450 GPUs across Oracle Cloud Infrastructure beginning in 2026 and expanding through 2027 and beyond.

Embedded Segment Growth: The Embedded segment is expected to return to growth in Q4 2025, driven by strengthening demand across multiple markets, including test and emulation, aerospace and defense, and industrial applications.

Revenue Outlook for Q4 2025: Revenue is expected to be approximately $9.6 billion, plus or minus $300 million, representing 25% year-over-year growth. This growth will be driven by strong double-digit growth in the Data Center, Client, and Gaming segments, and a return to growth in the Embedded segment.

Gross Margin Outlook: Non-GAAP gross margin for Q4 2025 is expected to be approximately 54.5%.

You have reached the limit Upgrade to access full content

Get started

Shareholder Return Plan

Share Repurchase Program: We returned $89 million to shareholders through share repurchases resulting in $1.3 billion in share repurchases for the first 3 quarters of 2025. Exiting the quarter, we have $9.4 billion authorization remaining under our share repurchase program.

You have reached the limit Upgrade to access full content

Get started

Key Q&A

Q:What is the CPU-GPU mix in Q3 and Q4, and how is the transition from MI355 to MI400 being managed?

A:The Data Center business had a strong Q3 with outperformance in server and data center AI, without MI308 sales. MI355 ramped sharply into Q3 and is expected to continue ramping in the first half of 2026. MI450 Series will come online in the second half of 2026, with a sharper ramp expected for the data center AI business.

Q:How is AMD planning its engagement with OpenAI and the visibility into 2027?

A:AMD is excited about its significant relationship with OpenAI, planning multiple quarters ahead to ensure power and supply chain availability. The first gigawatt deployment will start in the second half of 2026, with good visibility into the MI450 ramp. AMD is working closely with OpenAI and CSP partners to prepare for deployment.

Q:What is the view on discrete sales versus system sales for Helios, and what has been the customer response?

A:There is significant excitement around MI450 and Helios, with strong customer interest in rack scale solutions. Early customers for MI450 are expected to focus on rack scale solutions, though other form factors will also be available. The OCP reception was very positive, with customers bringing engineering teams to understand the system.

Q:What are the constraints for data center deployments in terms of components, power, and infrastructure?

A:The entire ecosystem must plan together. AMD is working with customers on power plans and supply chain partners to ensure capacity. AMD feels confident in its supply chain to support significant growth rates and compute demands, though the ecosystem will need to address tightness in power and supply.

Q:What is the sustainability of CPU demand for AI workloads, and are there supply constraints?

A:CPU demand has been broadening, with hyperscale clients forecasting significant builds into 2026. AI workloads are driving durable, multi-quarter demand. AMD is prepared for the ramp in 2026 and sees a positive demand environment for CPUs.

Q:What progress has been made with ROCm 7, and how does it compare competitively?

A:ROCm 7 has made significant progress in performance and framework support, with day 0 support for new models and frameworks. AMD continues to augment libraries and environments, focusing on newer workloads like reinforcement learning. The customer development experience has improved, and AMD will continue investing in this area.

Q:What is the framework for gross margins during the transition from MI355 to MI400?

A:Gross margins typically improve with new product generations. During the transition, there is a normalization period, but AMD prioritizes top-line revenue growth and gross margin dollar expansion while driving gross margin percentage up.

Q:How is AMD planning customer penetration and growth expectations for 2026 and beyond?

A:AMD is excited about its road map and has seen great traction with large customers like OpenAI. AMD expects multiple customers to achieve significant scale with the MI450 generation and is dimensioning the supply chain to support this growth.

Q:What grew more in the data center business in Q3: servers or GPUs?

A:Both servers and data center AI grew nicely year-over-year, with servers performing slightly better.

Q:What is AMD's guidance for data center growth in Q4?

A:Data center is expected to grow double digits sequentially, with both server and data center AI contributing to the growth. MI350 is also expected to ramp.

Q:How has the OpenAI deal influenced AMD's market position and customer engagement?

A:The OpenAI deal has accelerated customer interest and engagement, with customers engaging at higher scales. AMD is also dimensioning its supply chain to support multiple customers at similar scales.

Q:What is AMD's readiness to ship MI308 if the market opens up?

A:AMD has some work in process for MI308 and is working with customers on the demand environment. The situation is dynamic, and AMD will update as it evolves.

Q:How does AMD differentiate in the AI compute market with competitors also signing deals?

A:AMD focuses on time to market, total cost of ownership, deep partnerships, and a strong product road map. MI450 Series is well-positioned for both inference and training, and AMD is planning for future generations like MI500.

Q:What is AMD's view on the AI silicon TAM opportunity?

A:AMD sees the AI compute TAM increasing beyond the previously stated $500 billion, with updated numbers to be provided at the Analyst Day.

Q:How does the OpenAI relationship contribute to the development of AMD's software stack?

A:The OpenAI relationship strengthens AMD's software stack, particularly with Triton. AMD is also working with other large customers and AI-native companies to enhance the ROCm stack, using AI to accelerate development.

Q:What is the useful life of GPUs, and are CSPs planning to extend it?

A:There are indications that CSPs may extend the useful life of GPUs, especially for older generations like MI300X, which are still being deployed for inference.

Q:What is AMD's approach to unique agreements like the OpenAI deal?

A:The OpenAI deal was unique, focusing on deep partnership and aligned incentives. AMD is open to exploring other creative opportunities to address demand, leveraging its technical and operational capabilities.

Q:Review of Unclear Management Responses

A:Management avoided providing specific details on the CPU-GPU mix for Q3 and Q4, the exact growth percentages for servers versus GPUs, and the precise guidance for MI308 revenue. Additionally, they used vague language like 'strong double digits' without clarifying the exact figures, and did not provide a clear framework for gross margins during the MI355 to MI400 transition.

You have reached the limit Upgrade to access full content

Get started

Earnings Word Cloud

The most frequently occurring keywords in this quarter's earning call

AMD Transcript

Advanced Micro Devices, Inc. (AMD) Presents at Barclays 23rd Annual Global Technology Conference Transcript

Neutral12-10

Advanced Micro Devices, Inc. (AMD) Presents at UBS Global Technology and AI Conference 2025 Transcript

Neutral12-3

Advanced Micro Devices, Inc. (AMD) Q3 2025 Earnings Call Transcript

Positive11-4

Advanced Micro Devices, Inc. (AMD) Announcement of Strategic Partnership between AMD and OpenAI Conference Transcript

Neutral10-7

AMD Slides

AMD Q2 2025 slides: Revenue surges 32% despite export restriction charges

2025-08-05

AMD Report

ADVANCED MICRO DEVICES INC 10-K

10-K

2025-02-05

ADVANCED MICRO DEVICES INC 10-Q

10-Q

2024-07-31

ADVANCED MICRO DEVICES INC 10-Q

10-Q

2024-05-01

ADVANCED MICRO DEVICES INC 10-K

10-K

2024-01-31

Frequently Asked Questions

Where does this earnings call transcript come from?

All transcripts are sourced directly from the official live webcast or the company’s official investor relations website. We use the exact words spoken during the call with no paraphrasing of the core discussion.

How soon is the transcript available after the earnings call ends?

Full verbatim transcripts are typically published within 4–12 hours after the call ends. Same-day availability is guaranteed for all S&P 500 and most mid-cap companies.

Is the transcript edited or altered in any way?

No material content is ever changed or summarized in the “Full Transcript” section. We only correct obvious spoken typos (e.g., “um”, “ah”, repeated 10 times”, or clear misspoken ticker symbols) and add speaker names/titles for readability. Every substantive sentence remains 100% as spoken.

Why do some answers appear as “Unclear” or “Inaudible”?

When audio quality is poor or multiple speakers talk over each other, we mark the section instead of guessing. This ensures complete accuracy rather than introducing potential errors.

Who creates the AI Summary and Key Q&A highlights shown above the transcript?

They are generated by a specialized financial-language model trained exclusively on 15+ years of earnings transcripts. The model extracts financial figures, guidance, and tone with 97%+ accuracy and is regularly validated against human analysts. The full raw transcript always remains available for verification.