Intel is formulating a new strategy to enhance its competitive stance against industry leaders Nvidia and AMD in the rapidly expanding artificial intelligence chip market. The company aims to commence shipments of a novel AI graphics processing unit (GPU), named Crescent Island, before the conclusion of 2025. According to Intel's data center chief, Kevork Kechichian, this chip is specifically engineered for AI inference workloads and will incorporate more affordable memory and cooling technologies compared to rival offerings from Nvidia and AMD.
Measured Approach After Previous Setbacks
In an interview with the Financial Times, Kechichian revealed that Intel is adopting a more cautious strategy following earlier attempts to penetrate the AI accelerator market that did not yield the anticipated outcomes. The executive emphasized that the company is "starting with the basics" as it endeavors to rebuild its foothold in AI-focused semiconductors. "We decided to start rebuilding our muscles in AI ... [but] we are not particularly aiming for [the training market] based on past experience," Kechichian stated.
Focus on Inference Over Training
The Crescent Island GPU is dedicated to inference tasks, which occur when users interact with AI models and receive responses. This contrasts with AI training, where Nvidia currently holds a commanding market share. Intel's prior AI training chip, Gaudi, encountered difficulties in gaining market traction, and its planned successor was scrapped last year. By targeting inference, Intel hopes to carve out a niche in the AI chip landscape.
Cost-Effective Infrastructure as a Differentiator
A key aspect of Intel's strategy is reducing deployment costs for customers. The Crescent Island chip will utilize LPDDR5 memory instead of the high-bandwidth memory (HBM) commonly found in Nvidia and AMD AI accelerators. Additionally, the chip will be air-cooled, eliminating the need for liquid-cooling infrastructure often required for high-performance AI systems. This approach could make Intel's solution more accessible to a broader range of enterprises.
Kechichian noted that the chip has been under development for the past 18 months and will initially ship in limited quantities to customers before a wider rollout. This measured release allows Intel to gather feedback and fine-tune the product.
Leadership and Strategic Shifts
The Crescent Island project represents Intel's first major AI infrastructure initiative under CEO Lip-Bu Tan, who assumed leadership last year following organizational changes. Tan has prioritized cost reduction, product strategy refinement, and strengthening Intel's position in key semiconductor markets. His leadership is steering Intel toward a more focused and efficient approach.
Potential Sales in China Amid Export Restrictions
Intel is also exploring whether versions of Crescent Island could be sold in China while adhering to US export regulations. Current trade policies have limited Nvidia and AMD's ability to sell certain advanced AI chips in the Chinese market. "There are tiers of [the chip] that might be OK there ... and we'll confirm that over time: clearly there is demand for that particular price point in that particular market," Kechichian remarked. This could open up a significant revenue stream for Intel.
In-House Manufacturing Push
The company is also striving to manufacture more of its future data center products internally rather than relying on external foundries. Intel recently launched advanced PC and server chips produced in-house after years of substantial dependence on external manufacturing partners. "For all data center products we are moving aggressively into our own foundry. That's the intent in general," Kechichian added. Producing chips internally could further reduce costs and provide a competitive edge over rivals that continue to rely on external manufacturers for advanced AI processors.
Intel believes that this integrated approach will not only lower expenses but also enhance control over production and innovation. As the AI chip market intensifies, Intel's renewed focus on cost-effective solutions and in-house manufacturing may help it regain lost ground against Nvidia and AMD.
