Summary of Key Points
The focus of AI investment is shifting from GPUs (Graphics Processing Units) to the entire computing infrastructure, with CPUs (Central Processing Units) becoming the new beneficiaries. Intel’s second-quarter performance far exceeded expectations due to the surge in demand for data center CPUs. The company not only raised its capital expenditure forecast for 2026 but also revealed that customer demand has surpassed supply capacity. Its rival, AMD, has also increased its CPU market size forecast, believing that agent-based AI will continue to drive demand growth. As AI moves from model training to inference deployment and the implementation of intelligent agents, CPUs are gaining renewed importance in key areas such as task scheduling and data management, expanding the benefits across the entire computing ecosystem beyond GPUs.
Detailed Analysis
1. Intel’s Outstanding Performance: AI Boosts CPU Demand
Intel’s second-quarter results were impressive: revenue reached $16.1 billion, a year-on-year increase of 25% (the fastest in fifteen years), exceeding market expectations by $1.7 billion; adjusted net profit was $2.2 billion, with earnings per share at $0.42—twice what analysts anticipated. The gross margin also increased from 29.7% last year to 41.8%. Most notably, the revenue from its Data Center and AI Group (DCAI) grew by 59%, more than twice the company’s overall growth rate.
Why such strong performance? CEO Pat Gelsinger attributed it to “unprecedented” demand driven by AI, with data center CPU demand already exceeding Intel’s expanded production capacity. As a result, Intel has raised its capital expenditure for 2026 from $18 billion to $20 billion and plans to increase spending in 2027 as well—essentially building more factories and purchasing additional equipment to meet customer orders (it has already signed ten long-term supply agreements, but these are not enough). After the announcement, Intel’s stock price rose by 11% post-market. What matters most to the market is not the numbers themselves, but the confirmation that AI is driving CPU demand.
2. AI Transition from Training to Execution: CPUs Finally Gain Momentum
In the past two years, people have primarily purchased AI equipment based on GPUs (such as NVIDIA’s A100) because training large models requires extensive parallel computing, which GPUs excel at. However, the situation has changed:
- Inference Phase: After model training, these models are used for tasks like answering questions (e.g., ChatGPT) or generating images. At this stage, CPUs are needed to schedule tasks, manage data transfer, and handle storage—CPU act as the “project managers” that coordinate the work of GPUs.
- Agent-Based AI: AI assistants capable of performing complex tasks require coordination between multiple steps (researching, analyzing data, executing actions), all of which rely on CPUs.
Therefore, AI servers now need both GPUs and more powerful CPUs, leading to increased demand for these components.
3. AMD’s Optimistic Outlook: Could CPU Market Double?
Intel’s rival, AMD, has also revised its forecasts, raising the global CPU market size by 2030 from $120 billion to $220 billion—almost doubling. AMD’s reasoning is similar to Intel’s: agent-based AI workloads will continue to drive CPU demand growth. This indicates that the entire industry recognizes new opportunities for CPUs, with the AI ecosystem expanding beyond GPUs to a broader range of computing infrastructure.
4. Intel’s Other Business Areas Are Performing Well
Intel’s foundry business (manufacturing chips for other companies) is also making progress: second-quarter revenue from this segment grew by 31%. Its advanced 18A process has entered the risk production phase, bringing it one step closer to mass production. The company has also secured a manufacturing contract with cybersecurity firm Fortinet, its first public external customer. Management notes that its advanced packaging technology (which combines multiple chips) has accumulated significant orders, potentially supporting future growth in this area. This represents another growth driver for Intel as it transitions its business model.
5. Future Challenges: Can Capacity, Manufacturing, and Foundry Services Meet Demand?
Despite the current positive trends, Intel still faces several challenges:
- CPU Production Capacity: With demand exceeding supply, can it quickly expand production to meet orders on time?
- Advanced Manufacturing Processes: Will the 18A process be successfully mass-produced? This is crucial for Intel to remain competitive by producing more advanced CPUs.
- Foundry Business: Can it attract more external customers? Given that foundry services were not Intel’s strong point in the past, competing with companies like TSMC and Samsung presents significant challenges.
The ability to overcome these issues will determine Intel’s long-term growth prospects, and the market will closely monitor whether these goals are achieved.
Conclusion
AI no longer relies solely on GPUs; CPUs, as the core of computing infrastructure, are experiencing a resurgence driven by AI. Both Intel and AMD have recognized this opportunity, as evidenced by their performance and revised forecasts. However, for Intel to turn this short-term success into long-term advantages, it will depend on its ability to expand production capacity, develop advanced technologies, and grow its foundry business. For individuals, this means that investment opportunities in the AI ecosystem are no longer limited to GPUs; companies related to CPUs also deserve attention.