Beyond the GPU: The Resurgence of CPUs in the AI Era
For years, the narrative around artificial intelligence has been dominated by the GPU. While graphics processing units are essential for training massive models, a critical shift is occurring in how AI infrastructure is built. We are moving toward a hybrid model where the Central Processing Unit (CPU) regains its status as the “brain” of the operation.
The trend is clear: as AI workloads shift from foundational training to inference, agentic AI, and edge computing, the CPU’s role in coordinating systems and handling the control layer becomes indispensable. This isn’t just theory; it’s reflected in recent industry movements where server CPUs are being deployed alongside accelerators in increasing ratios.
The Shift Toward Agentic and Edge AI
The next wave of intelligence is moving closer to the conclude user. This transition from massive, centralized foundational models to “agentic AI”—AI that can take independent action to complete goals—requires a different hardware approach.
Edge computing and enterprise deployment demand chips that can move data efficiently and manage complex, distributed systems. This is where CPUs excel. By handling the orchestration of data and the control layer around accelerators, CPUs ensure that AI doesn’t just “feel” but actually “functions” within a business environment.
A prime example of this trend is the multiyear collaboration between Intel and Google, which focuses on combining Xeon processors with custom application-specific integrated circuits (ASICs) to optimize specific workloads.
Analyzing the Turnaround: Data and Market Sentiment
The financial indicators suggest a company regaining its footing. Recent reports show a significant beat in expectations, with first-quarter revenue reaching $13.6 billion—a 7% year-over-year increase that surged past management’s previous guidance of $11.7 billion to $12.7 billion.
Profitability metrics are also trending upward:
- Non-GAAP Earnings Per Share: Rose to $0.29 from $0.13 in the previous year’s first quarter.
- Adjusted Gross Margin: Expanded to 41%, up from 39.2%.
- Adjusted Operating Margin: Jumped to 12.3% from a previous 5.4%.
- Adjusted Net Profit: Reported at $1.5 billion, representing a 156% year-over-year increase.
This execution has caught the eye of Wall Street. For instance, HSBC recently upgraded Intel to “Buy” and significantly raised its price target from $50 to $95, signaling a renewed confidence in the company’s trajectory.
The Foundry Gamble: Risk vs. Reward
Despite the momentum in CPU and AI-adjacent business, a significant hurdle remains: the foundry business. Manufacturing chips is a capital-intensive endeavor that continues to weigh on the bottom line.
While foundry revenue grew 16% year-over-year to $5.4 billion, the segment still posted an operating loss of $2.4 billion. The future of the company depends on whether it can scale this manufacturing capability to a point of profitability without exhausting its capital reserves.
With a market capitalization of approximately $335 billion, much of the “comeback” may already be priced in. The risk now lies in the execution of the foundry strategy and the ability to maintain growth in a highly competitive AI landscape.
Frequently Asked Questions
Why are CPUs important for AI if GPUs do the heavy lifting?
GPUs are optimized for the parallel processing needed for training models, but CPUs are essential for coordinating systems, moving data, and managing the control layer, especially in inference and edge computing.
What is “Agentic AI” and how does it affect hardware?
Agentic AI refers to systems that can autonomously perform tasks. This shift increases the need for CPUs and advanced packaging to bring intelligence closer to the end user and manage more complex, distributed workloads.
Is Intel’s foundry business profitable?
No, the foundry business continues to be a source of loss. Despite revenue growth, it recently reported an operating loss of $2.4 billion.
Join the Conversation
Do you think the CPU will remain central to the AI boom, or will accelerators eventually take over entirely? Share your thoughts in the comments below or subscribe to our newsletter for more deep dives into semiconductor trends!
