Qualcomm (QCOM.US) held its Q4 FY2025 earnings call, expressing excitement about its entry into the data center sector, which it views as a "new chapter" in diversification. The company highlighted its core competitive advantage in energy efficiency—maximizing token generation with minimal power consumption. This strategy is supported by two key initiatives: a highly competitive, energy-efficient CPU for AI cluster head nodes and general computing, and a new architecture specifically designed for AI inference.
The development of its AI 200 and AI 250 chips and accelerator card solutions is progressing well, with further details and performance metrics set for release early next year. Qualcomm also confirmed discussions with a "hyperscale" customer, expressing satisfaction with the outcome. Additional details will be shared as part of the company’s future data center roadmap updates.
**Key Takeaways from Q&A:**
1. **Data Center Competitive Edge**: Qualcomm emphasized energy efficiency as its primary advantage, leveraging a high-performance CPU and a new AI inference architecture. The AI 200 and AI 250 solutions align with this strategy, with specifics to be disclosed in early 2026.
2. **Samsung Partnership**: Despite market speculation about Samsung’s potential shift to in-house modem chips, Qualcomm reiterated its baseline assumption of 75% share in future models, with occasional outperformance (e.g., 100% in Galaxy S25).
3. **AI Deployment Costs**: While avoiding premature disclosures, Qualcomm noted accelerated revenue growth expectations for its data center business, now projected for FY2027 (vs. FY2028 earlier), driven by its partnership with Humain and AI accelerator progress.
4. **Smartphone Growth Drivers**: Revenue growth in non-Apple QCT (18% YoY) stemmed from premiumization—both higher-tier device adoption and increased content per device. Qualcomm expects this trend to persist.
5. **China Market Momentum**: Strong demand for new Snapdragon-powered devices in China reflects organic adoption, not inventory buildup, with global launches extending into early 2026.
6. **Differentiated AI Architecture**: Qualcomm’s approach prioritizes cost and energy efficiency over traditional GPU-HBM combinations, targeting superior compute density for token generation.
7. **Margins and Investments**: The 100+ bps YoY margin decline reflects strategic investments in data centers, shifting OpEx from mature businesses to high-growth areas.
8. **XR and Data Center Potential**: The XR market, particularly AI glasses, is outperforming initial expectations, while data center opportunities could scale to multi-billion-dollar revenue streams in coming years.
Qualcomm reaffirmed its confidence in sustained growth across automotive, IoT, and smartphone segments, with QCT revenue projected to rise 11–13% QoQ in December quarter, led by Android premium devices.