Eight Major Tech Giants Form Alliance to Challenge NVIDIA’s Dominance

In the battle for AI dominance, Intel, AMD, Google, Microsoft, Meta, Broadcom, HPE, and Cisco form UALink to rival NVLink.

AI chip giant NVIDIA is facing a coordinated counterattack from competitors in the artificial intelligence field. Arculus System CEO James Yang points out that NVIDIA’s success is not just from selling chips, but from leveraging its software ecosystem platform and powerful network technology (NVLink) to serve its customers. He highlights that competitors are continuing to integrate, with eight major tech giants, led by Intel and AMD, forming the UALink alliance to challenge NVIDIA’s dominance.

The eight tech giants—Intel, AMD, Google, Microsoft, Meta, Broadcom, HPE, and Cisco—have a combined market value of $7.6 trillion. They have established the UALink (Ultra Accelerator Link) alliance to set new interconnect technology standards for AI data centers, rivaling NVIDIA’s NVLink. This scenario, akin to “eight major factions besieging Bright Summit,” reflects the industry’s anxiety over NVIDIA’s leading position.

NVIDIA dominates the data center and AI GPU market with a market share exceeding 90%. On the 25th, NVIDIA’s stock price ended a three-day losing streak, rising over 2% in early trading, with a market value nearing $3 trillion. James Yang analyzes that the key to success for IC design companies at this stage is no longer just the chips themselves but increasingly relies on system and ecosystem support. NVIDIA’s moat is built on three core advantages: leading GPU chip technology, a comprehensive software ecosystem such as the CUDA platform, and robust network technology, making it difficult for competitors to overtake in the short term.

While the UALink alliance is gaining momentum, it will take considerable time to launch actual products, with the first product expected by 2026. By then, NVIDIA may have already introduced the new Rubin GPU architecture. James Yang believes that time to market is crucial for capturing the AI blue ocean. With a background in chip design, Yang is well aware of the pain points encountered in the IC design process, from IP to custom chips, and views EDA tools as the trump card for building ecosystems. He notes that Arculus System’s GreenEDA can quickly assist customers in providing comprehensive performance analysis during the architectural design phase, significantly saving engineers’ time.

Currently, major EDA companies were established during the IDM era. Arculus System, with its inherent advantages, begins from a fabless perspective, closer to engineers’ needs. As the only global optimization for architectural design processes, Yang reveals that facing today’s 3D stacking, Arculus can provide performance and thermal power analysis in the early stages, helping to solve AI chip energy consumption issues.

The AI chip market competition continues, creating enormous peripheral business opportunities. The market landscape remains uncertain. As AMD CEO Lisa Su previously stated, “In such a rapidly developing market, I don’t believe in the existence of competitive moats.”

New source: https://www.ctee.com.tw/news/20240626700061-439901