CPU Returns to the AI Battlefield: Who Will Prevail? Intel, AMD, and NVIDIA Spark a Three-Way Battle for Dominance

As artificial intelligence (AI) applications evolve from simple model computations to autonomous 'AI agents' capable of planning and executing multi-step tasks, the central processing unit (CPU)—once considered a supporting player—is now standing shoulder-to-shoulder with the graphics processing unit (GPU) as a critical battleground. Intel (INTC-US), Advanced Micro Devices (AMD-US), and NVIDIA (NVDA-US) are each ramping up their strategies, and a three-way power struggle is beginning to take shape in the market.

Intel reported its second-quarter earnings on Thursday (23rd), posting quarterly revenue of $16.1 billion. The standout performer was the Data Center and AI business segment, which saw a 59% year-over-year increase in revenue, reaching $6.3 billion—surpassing market expectations.

Company executives attributed this growth to a resurgence in server CPU demand driven by AI agents, revealing that they have signed multiple long-term contracts with customers lasting three to five years, some of which include dual commitments on both purchase volume and pricing.

Intel CEO Pat Gelsinger stated bluntly: 'In the data center space, CPU demand is taking off, exceeding our growing supply capacity.'

Strong demand has also given Intel greater pricing power. The company's CFO confirmed that market acceptance of price adjustments has been better than expected.

In fact, Intel has already raised prices for certain consumer and server processors this month, with consumer products seeing increases of $30 to $50, and server products experiencing hikes ranging from hundreds to over a thousand dollars.

Intel responded that these price adjustments are part of a routine review based on supply chain conditions and costs.

Wall Street investment firm Wedbush believes that, given the ongoing tight supply of server CPUs, Intel still has room to raise prices further without negatively impacting end-market demand.

From GPU Dominance to CPU Power-Sharing: NVIDIA and AMD Clash

For years, the focus of AI computing has been almost entirely on GPUs. However, with the rise of AI agents, where every token's computational cost is scrutinized, CPUs—excelling in task scheduling and logical planning—are beginning to play a more pivotal role in this arms race.

Notably, Intel is not the only player in this wave.

NVIDIA recently released benchmark data for its next-generation Vera Rubin platform, with its flagship product being the Vera CPU, specifically designed for the era of AI agents.

NVIDIA emphasized that through its proprietary Olympus core architecture, the Vera CPU achieves twice the single-thread performance compared to competitors' chiplet-based designs, three times higher inter-core bandwidth, and 40% lower memory latency.

However, just two days later, AMD directly countered at its annual AI summit 'Advancing AI 2026,' unveiling its sixth-generation server CPU 'Venice' to compete with Vera, claiming that Venice delivers 2.2 times the throughput performance of Vera and leads by 20% in per-core performance even in out-of-the-box configurations before any performance tuning.

Moreover, AMD forecasts that by 2030, the traditional data center CPU market will reach $220 billion, and the total addressable market it competes in will grow to $2 trillion.

Bank of America analyst Vivek Arya highlighted the underlying technological divergence in this competition, noting that what the market truly cares about is whether AI requires higher single-core clock speeds or more core counts.

NVIDIA is betting on the former, advocating for reduced latency between tool calls and task switching. AMD, on the other hand, firmly believes in the advantage of 'quantity,' continuously increasing core and thread counts to pursue rack-level concurrent throughput capacity.

In contrast to the technical rivalry between NVIDIA and AMD, Intel is choosing to respond with capital investment.

The company announced it is raising its annual capital expenditure plan from the originally planned $18 billion to $20 billion, with approximately $5.7 billion dedicated specifically to expanding production capacity for its Xeon 6 series server processors using the Intel 3 process.

As the three major players intensify their investments, the market expects the CPU industry landscape to gradually shift from Intel's current 'one dominant, multiple strong' structure toward a new three-way balance of power.

One institution predicts that by 2030, ARM-architecture commercial CPUs introduced by NVIDIA and others could capture 36% market share; Intel's share may decline from its current 41% to 24%; while AMD is expected to maintain a relatively stable position between 25% and 27%.

Academic Research Confirms: CPU Importance Has Been Underestimated

The industry's optimism isn't solely based on corporate claims. Joint research by Georgia Tech and Intel found that in five AI agent task tests, CPU computation time accounted for over 50% of the total task duration in every case.

Market research firm TrendForce also forecasts that in the AI agent era, the CPU-to-GPU configuration ratio will narrow significantly from the previous 1:4 to 1:8 range to between 1:1 and 1:2.

China's CITIC Securities further suggests that even 1:1 to 1:2 might just be the starting point, with optimistic scenarios seeing ratios as high as 8:1. In such a case, the CPU-related market size could exceed $2 trillion, surpassing the current GPU market.