AMD is on a strong trajectory in the AI datacenter market, with its performance in the third quarter of 2025 signaling a potential hockey stick growth pattern. The company reported $9.25 billion in revenue, a 35.6% year-on-year increase and a 20.3% sequential rise. Operating income jumped 61.6% to $1.17 billion, while net income grew 61.2% to $1.24 billion, aided by accounting benefits. AMD ended the quarter with $7.24 billion in cash and equivalents, reflecting strong financial health. The datacenter segment remains AMD’s largest business, generating $4.34 billion in sales—up 22.3% year on year and 34% from the previous quarter. Operating income for the group reached $1.07 billion, representing a robust 24.7% margin. While AMD doesn’t break down Epyc CPU and Instinct GPU revenues separately, analysis suggests Epyc CPUs drove approximately $2.17 billion in sales, a 23% increase year on year. Of that, $1.61 billion came from hyperscalers and cloud providers, who have consistently adopted AMD’s CPUs for their infrastructure. Enterprise, telecom, academic, government, and other service providers contributed $561 million in Epyc sales, up 18%. Instinct GPU sales totaled just under $2 billion in the quarter, up 20.7% year on year. This growth is fueled by strong demand for AI inference workloads, particularly from cloud builders and hyperscalers. The MI300X continues to perform well, while the newer MI325X and MI355X are ramping steadily, with the MI355X already in shipment and expected to gain momentum through the first half of 2026. A major catalyst for AMD’s expansion is its acquisition of ZT Systems, finalized in August 2024. AMD has now completed a deal with contract manufacturer Sanmina to sell the server manufacturing arm of ZT for $3 billion. The transaction includes $2.55 billion in cash and $450 million in Sanmina equity tied to future business volume. This will boost AMD’s bottom line in Q4 and beyond, and the acquired 1,100 system engineers and architects have been integrated into AMD’s existing 500-person team, strengthening its in-house system design capabilities. The new “Helios” double-wide server racks, now an Open Compute Project standard, are being adopted by key customers and are set to be a key enabler of AMD’s next-gen AI systems. The upcoming MI430X and MI450X GPUs, featuring a new compute architecture, are expected to challenge Nvidia’s dominance in the datacenter AI accelerator market. CEO Lisa Su confirmed that AMD is revising its total addressable market for AI upward, and the company is preparing to provide updated guidance for Q4 2025 and beyond at its upcoming Financial Analyst Day. She also stated that the company is building a foundation for “well over $100 billion in revenues over the next few years.” While AMD’s datacenter group is already the largest segment, the company’s other divisions still collectively outpace it. However, with strong momentum in AI, a growing ecosystem, and key partnerships, AMD could see datacenter GPU sales surpass CPU sales within the next 18 to 24 months—marking a pivotal shift in its business model.