Stacy Rasgon on Qualcomm's AI Chip Strategy and Market Potential

Qualcomm's AI Accelerator Strategy

• 💡 Qualcomm has been selling AI accelerators for years, with the AI 100 being a notable example.
• 🚀 The company is now introducing next-generation AI accelerator chips targeting the AI inference market.
• 📈 While it's too early to determine their competitive success, these new chips represent significant option value for Qualcomm.

AI Inference Market Dynamics

• 🎯 The AI inference market is expected to be more fragmented than AI training, which is dominated by Nvidia.
• ⚡ Inference workloads may be more specialized, with power efficiency being a key consideration, creating room for more players.
• 💰 Even a small slice of the potentially massive inference market could be substantial for companies like Qualcomm.

Competition and Nvidia's Position

• ⚠️ While Nvidia currently dominates AI training, the inference market presents opportunities for other companies.
• 🧩 The primary question for the market is the overall size of the AI opportunity, not necessarily which specific technology (GPU vs. ASIC) will win.
• 📈 As long as Nvidia can continue to significantly improve performance with each generation, their margins are unlikely to be threatened, even with increasing competition.
• 📊 Nvidia's ability to double pricing while increasing performance tenfold (e.g., Ampere to Hopper) demonstrates their pricing power and customer value proposition.