TL;DR:
- Naver, a leading South Korean web portal giant, has replaced Nvidia with Intel as its chip supplier for AI-powered map services.
- This shift comes amid discontent over Nvidia’s GPU price increases and a global GPU shortage.
- Naver’s AI model, critical for distinguishing information on its map app, was previously powered by Nvidia GPUs.
- The shortage of Nvidia’s GPUs has led to soaring prices, making system upgrades more expensive and challenging.
- Intel has improved its AI inference capabilities, making it a viable alternative to Nvidia’s GPUs.
- Naver conducted extensive testing before deploying Intel CPU-powered AI servers.
- The company plans to continue using Intel’s next-generation CPUs for system upgrades.
- Analysts suggest this transition may reduce Nvidia’s influence in the AI processor market.
- Microsoft has partnered with AMD to compete with Nvidia’s dominance in the global AI processor market.
Main AI News:
In a notable strategic move, South Korea’s prominent web portal giant, Naver Corp., has opted to transition its main chip supplier for its AI-powered map service, Naver Place. The company has moved away from Nvidia Corp., a long-standing partner, in favor of Intel Corp., a decision that has sparked intrigue in the tech industry.
Naver’s reliance on Nvidia’s graphic processing unit (GPU)-based server for its AI-powered location information service has been a longstanding practice. However, in light of Nvidia’s GPU price hikes and a global shortage of these critical components, Naver made the pivotal switch to Intel’s central processing unit (CPU)-based server.
This transition is indicative of a broader trend among global technology firms expressing dissatisfaction with Nvidia’s pricing strategies and the scarcity of its GPUs, which has created a challenging environment for AI server users worldwide.
Naver’s AI model plays a pivotal role in discerning authentic information from misleading data when users search for “nearby restaurants” and similar queries on the Naver Map app. This process involves intricate data processing, inference, and verification, which have historically been powered by Nvidia’s GPU chips.
However, securing GPU accelerators, particularly Nvidia’s H100, which boasts multiple chips for data processing and computing, has grown increasingly arduous due to surging demand for generative AI devices. The skyrocketing prices, with AI accelerators using Nvidia’s GPUs now fetching as much as 80 million won ($59,193), have added to the predicament of AI server users looking to upgrade their systems. Furthermore, the lead time for obtaining Nvidia’s GPU-based accelerators after placing an order has stretched to a daunting 52 weeks.
Nvidia’s GPUs are instrumental in the operation of generative AI devices like ChatGPT, a chatbot developed by Microsoft Corp.-backed OpenAI. These semiconductors enable rapid processing of vast datasets, making them invaluable components.
Nvidia’s heavy reliance on Taiwan Semiconductor Manufacturing Co. (TSMC) for chip packaging has encountered challenges, as TSMC’s packaging process lines are nearly at full capacity. Consequently, Nvidia is exploring alternatives and seeking partnerships with other foundry players, including Samsung Electronics Co.
While GPU servers have historically outperformed CPU servers by processing AI inference tasks ten times faster, Intel has made significant strides in enhancing its AI inference capabilities, offering viable alternatives to Nvidia’s GPUs.
Naver conducted extensive testing of its Intel CPU-powered AI server before its implementation. The Korean company is poised to continue utilizing Intel’s next-generation CPUs, such as the Sapphire Rapids CPU, to elevate its location and information search services.
Industry analysts speculate that Naver’s collaboration with Intel could potentially erode Nvidia’s dominance in the AI processor market, paving the way for new contenders.
In parallel, Microsoft Corp. has forged a partnership with Advanced Micro Devices Inc. (AMD), a fabless chip designer, to fortify the US chipmaker’s presence in the AI processor domain. This strategic alliance aims to challenge Nvidia’s dominant position, which currently commands an estimated 80% market share in the global AI processor arena, according to media reports.
Intel’s fourth-generation Sapphire Rapids Xeon scalable processors. Source: KED Global News Network
Conclusion:
Naver’s strategic shift from Nvidia to Intel for its AI map app servers reflects growing discontent with Nvidia’s pricing strategies and GPU shortages in the technology industry. Intel’s improved AI inference capabilities provide a viable alternative, potentially diminishing Nvidia’s market influence. Microsoft’s collaboration with AMD further underscores the competitive landscape in the AI processor market, challenging Nvidia’s dominant position.
