Chip firm AMD posted record revenue for its data center segment, recording 22 percent Year-Over-Year (YoY) and 34 percent Quarter-over-Quarter (QoQ) growth in Q3 2025.

Totaling $4.3 billion for the quarter and beating analyst expectations, the company said the growth was driven by strong demand for its 5th-generation Epyc Turin CPUs and Instinct MI350 Series GPUs.

The segment’s results reflected growth largely across the board for the chip designer, which posted $9.3 billion in revenue for the three-month period ending September 27, a 36 percent YoY increase, and up 20 percent QoQ.

Operating income for the quarter was $1.3bn, while net income was $1.2bn. Operating expenses totaled $3.5bn, an increase of 30 percent YoY and primarily driven by AMD’s continued investment in R&D to “capitalize on significant AI opportunities and go-to-market activities for revenue growth,” the company said in a statement.

"We delivered an outstanding quarter, with record revenue and profitability reflecting broad-based demand for our high-performance Epyc and Ryzen processors and Instinct AI accelerators," said Dr. Lisa Su, AMD chair and CEO. "Our record third quarter performance and strong fourth quarter guidance marks a clear step up in our growth trajectory as our expanding compute franchise and rapidly scaling data center AI business drive significant revenue and earnings growth."

For Q4 2025, AMD has forecast revenue of approximately $9.6bn, plus or minus $300 million. The midpoint of the guidance represents YoY revenue growth of around 25 percent, the company said, noting that it expects this figure to reflect strong double-digit growth in its data center segment, in addition to its client, gaming and embedded units.

Partnerships and compute demand continue to fuel growth

Speaking on the company’s earnings call after the results were published, Su said sales of its Turin CPUs accounted for almost half of all Epyc sales during the quarter, with AMD also seeing launches of more than 160-powered instances from hyperscalers during Q3.

Regarding its GPUs, Su name checked Oracle for its deployment of the first publicly available MI355X instances, before going on to note MI350 series offerings from neoclouds, including Crusoe, DigitalOcean, TensorWave, Vultr, during the quarter.

However, despite agreeing a deal with the Trump administration in August for the shipment of its Instinct MI308 to China, for the second quarter in a row, AMD said its quarterly results did not include any revenue generated by their sale.

Su described the situation as “pretty dynamic” on the earnings call, and said the company’s Q4 outlook also doesn’t include any revenue from Instinct MI308 shipments. Despite this, she went on to state that AMD had received “some licenses” for the chip and is continuing to work with customers, “so we'll be able to update that more in the next couple of months.”

Looking ahead, Su said that AMD remained on track to launch its 2nm Venice processors in 2026, adding that its AI data center business was also “entering its next phase of growth,” with customer momentum growing ahead of the MI400 accelerator series and Helios rack-scale solution launches in 2026.

“In summary, our AI business is entering a new phase of growth and is on a clear trajectory towards tens of billions in annual revenue in 2027,” Su said, after rattling through the recent partnerships announced by the company, most notably with OpenAI for the deployment of 6GW of Instinct GPUs, and with Cisco and G42 in the United Arab Emirates for a forthcoming Instinct MI350X GPU cluster.

“The demand for compute has never been greater as every major breakthrough in business, science, and society now relies on access to more powerful, efficient, and intelligent computing. These trends are driving unprecedented growth opportunities for AMD.”

Su said the company would provide more details on its AI data center growth plans at AMD’s Financial Analyst Day on November 11.