Advanced Micro Devices saw its stock rise by about 7% on Tuesday after executives presented an assertive growth strategy for artificial-intelligence workloads at the Citi Global TMT Conference. Central to the companys outlook was a newly stated total addressable market of $2 trillion by 2030, a figure AMD tied to the companys expectation of expanding inference demand.
Company management emphasized that the balance of AI computing work is shifting. Quoting Citis summary of the presentation, AMD noted that "In the past twelve months, Inference has become the majority driver of AI computing," and that workloads are moving beyond basic conversational bots toward more independent, agentic AI systems. That change in workload profile was presented as a primary rationale for AMDs investment and product priorities.
Executives outlined an accelerated build-out of AMDs data-center business to capture the expanding inference market. The firm expects its data-center segment to more than double next year. As part of that push, AMD plans to introduce its next-generation AI accelerator, the MI450 GPU, this quarter, with production increases beginning in the fourth quarter and continuing into 2027.
Traditional server compute also factors prominently in AMDs roadmap. Management projected the server CPU business will grow more than 80% year-over-year in the second half of this year and exceed 70% year-over-year in the following year. Those gains in server processors and related compute capacity are intended to support the broader data-center and inference deployments the company detailed.
A concrete indicator of rising enterprise demand is AMDs Helios rack-scale system. The company counts Meta Platforms and two unnamed AI labs as anchor customers for Helios, and said those clients have provided demand forecasts that exceed their initial purchase agreements. AMD said Helios volume expectations for 2027 are already ahead of the companys earlier projections, and it is seeing orders come from new so-called "neo-cloud" providers as well.
On profitability, AMD acknowledged that data-center GPUs currently carry margins below the firms corporate average, but said its wider product mix is cushioning the impact. Management pointed to strong double-digit growth in higher-margin server and embedded segments during the second and third quarters as the mechanism preserving overall profitability.
Looking further down the product pipeline, AMD said it is in ongoing technical engagements with its three leading customers around next-generation MI500 and MI600 chips, underscoring continued collaboration on future hardware. The company also disclosed a partnership with Cerebras and indicated there are internal plans, not yet publicly detailed, aimed at low-latency inference markets.
Context note: The remarks, product timing and demand signals described here were presented by AMD management at the Citi Global TMT Conference and summarized in Citis event notes. The companys statements tied a long-term $2 trillion market estimate to the rising prominence of inference workloads through 2030.