AMD said at Citi’s 2026 Global TMT Conference on Sept. 8, 2026 that the artificial intelligence opportunity could reach $2 trillion by 2030.
The company said its data center business is expected to double by 2027, with revenue reaching about $70 billion. Around $40 billion of that figure is expected to come from AI GPUs, while the rest would come from server CPUs.
AMD also raised its view of the server CPU market, saying the segment could reach $220 billion by 2030, up from its earlier estimate of $60 billion.
On supply, AMD said it has secured $29 billion to $30 billion in purchase commitments and received the largest increase in capacity allocation among all TSMC customers. Even so, the company said supply remains constrained in wafers, HBM memory, advanced packaging and substrates.
For product timing, AMD said MI450 will begin volume production shipments in the third quarter of 2026, with a larger ramp in the fourth quarter. The company guided for a third-quarter gross margin of 56%. Because MI450 is still ramping, AMD said gross margin is expected to edge lower in the fourth quarter of 2026 and in 2027.
Chief Financial Officer Jean Hu said, “The speed, scale and rise of AI are unprecedented.”
AMD said it has already named Meta, OpenAI and Anthropic as three major anchor customers, all involving multi-gigawatt deployments and multi-generation partnerships. Helios is also scaling at the rack level.
The company added that it has completed its acquisition of Taalas and is working with Cerebras to integrate the Helios system with Cerebras’ wafer-scale engine technology.
In server CPUs, AMD said the business is expected to grow by more than 80% year over year in the second half of 2026, and by more than 70% as it moves into 2027. Enterprise server revenue grew by more than 70% in the second quarter of 2026.