Lisa Su 3.1 4 ideas

AMD is expanding its partnership with Meta. Su states, "Every gigawatt of compute is, you know, double digit billions of revenue to us." She further notes, "We are sizing this AI accelerator market at $1 trillion over the next 5 years." The partnership validates AMD's hardware (likely MI300 series) as a viable alternative to Nvidia for top-tier hyperscalers. If Meta is deploying at a multi-gigawatt scale, and each gigawatt equals $10B+ in revenue, this signals a massive, concrete revenue pipeline for AMD that the market may have previously underestimated. For Meta, this diversification reduces vendor lock-in risk and secures necessary compute for their "superintelligence" goals. LONG AMD for the revenue unlock; LONG META for the infrastructure security. AMD fails to deliver on yield/performance; Nvidia counters with aggressive pricing; Meta reduces capex spend.

"I'm a big believer in there's no one size fits all. Frankly, you need a lot of different types of compute... We want to place bets on who we think are going to be the winners in AI innovation." The "no one size fits all" comment suggests the AI market is shifting from a monopoly (Nvidia-only) to an oligopoly where specialized chips (CPUs, GPUs, NPUs) from different vendors are required. This benefits the broader hardware ecosystem, not just the market leader. LONG the broad AI hardware basket. AI bubble bursts; oversupply of chips in 2025/2026.