Qualcomm posts strong outlook despite tax hit

Qualcomm expects Q1 revenue of around US$12.2 billion and adjusted earnings of US$3.40 per share, both above analyst estimates.

The San Diego-based chipmaker, known for its smartphone processors, forecast strong demand for high-end Android phones.

However, Qualcomm reported a US$3.1 billion net loss for the last quarter after a US$5.7 billion writedown related to a US tax change.

Excluding certain items, profit was US$3 per share in Q4, beating the US$2.9 analyst estimate, while revenue rose 10% year-on-year to US$11.3 billion.

Phone-related sales reached US$7 billion, ahead of forecasts, while connected devices and automotive chips brought in US$1.8 billion and US$1.1 billion, respectively.

Qualcomm faces rising competition as Apple moves to use its own modem chips, and as regulatory pressures ease in China.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

Qualcomm data center AI chips face integration hurdles

AI200 and AI250 use up to 35% less power than comparable GPU systems 1. Many enterprises still struggle to leave Nvidia’s CUDA ecosystem (Compute Unified Device Architecture), Nvidia’s proprietary GPU programming platform 2.
Qualcomm scaled its mobile Hexagon Neural Processing Unit (NPU), its on device AI accelerator, to rack systems 2. Teams must retrain developers and plug into data center orchestration plus management tools to deploy or schedule AI jobs 2. Lower per chip throughput can force more racks to match GPU performance 3. Citi (Citigroup’s investment bank) pegs the 200 MW (megawatt) Humain deal at $1 billion 4. Christopher Danely at Citi doubts Qualcomm can close the multi year gap behind AMD and Nvidia in AI 4.

Integrators can build distinct inference options for cost sensitive enterprise AI

Cloud resellers with system integrators (firms that assemble plus maintain complete hardware/software solutions) can package AI200 and AI250 into turnkey (pre integrated, ready to deploy) rack setups 1 for workloads like chatbots or financial forecasting 1. Early movers that support Qualcomm alongside Nvidia give buyers flexibility while cutting vendor lock in risks (the cost and complexity of switching suppliers) 2.
AI250 memory delivers over 10x effective bandwidth versus current Nvidia GPUs 3. Integrators who quantify total cost of ownership (TCO) gains, including energy savings that can reach millions each year for large data centers 1, can win buyers seeking predictable costs as AI infrastructure spending could exceed $2.8 trillion through 2029 1.