Unique Research · Insight · September 10, 2024 · Source: Goldman Sachs Communacopia & Technology conference interview — Lisa Su (AMD Chair & CEO) with Brian Sozzi (Yahoo Finance)
At Goldman’s 2024 Communacopia & Technology conference, AMD Chair and CEO Lisa Su sketched a bullish AI-chip outlook and AMD’s roadmap. She expects the global AI market to reach $400 billion by 2027 and argued that an AI super-cycle is only just beginning.
“We are making strategic investments for a major leap over the next five years,” she said. “There is no one-size-fits-all in this technology race — the key is end-to-end AI.” AMD, she added, is accelerating its AI blueprint to stay central in that race.
MI300: AMD’s fastest-growing product line
Su highlighted the MI300X GPU accelerator — among AMD’s fastest-growing lines since launch last year. Specs called out in the talk: up to 192GB memory and roughly 153 billion transistors, enough to train frontier LLMs such as OpenAI’s ChatGPT. The product, she said, marks AMD’s rapid rise in AI silicon and a credible challenge to Nvidia’s long dominance.
She also framed MI300X as AMD’s historically fastest-growing line and guided that AI-related chips could deliver up to ~$4.5 billion of revenue in 2024 — versus roughly $100 million of AI chip revenue the year before — crediting foresight and product execution.
Three-year cadence: MI325 → MI350 → MI400
AMD plans an annual product cadence: MI325 later this year, MI350 in 2025, and MI400 in 2026 — a ladder aimed at Nvidia’s high-performance AI franchise. Su: “We are confident we can take a meaningful position in both large-language-model training and inference.”
Demand signals and top customers
Jefferies analyst Blayne Curtis noted AMD raised MI300 expectations in Q2 against a market that had priced in softness — overdone supply-chain worry, with no pullback from top accounts. Microsoft, Meta, and Oracle were cited as AMD’s top three AI customers. Street sketch in the piece: ~$5B sales this year and $8–9B next year as the expansion base.
Investing through the super-cycle
Su admitted the AI surge outran her five-year-ago expectations: “Five years ago I did not foresee AI at this scale. We are now making major investments for the next five.” Faster roadmaps plus continuous silicon refreshes are AMD’s bet on exponential AI growth — and on becoming a durable force in the high-performance AI-chip market.
Note (from CN close): Unique Research offered a small pool of free tickets to an AI-marketing event for practitioners who commented interest — limited, first-come.
— END —