Nvidia has built a formidable $2.2 trillion market cap by manufacturing artificial intelligence (AI) chips that power a new era of generative AI applications, utilized by startups and tech giants like Microsoft and Alphabet's Google.
Equally crucial to Nvidia’s dominance is its nearly two-decade-old CUDA software platform, which more than 4 million global developers rely on to build AI and other applications. This extensive software ecosystem has made it challenging for competitors to gain a foothold.
Now, a coalition of tech companies including Qualcomm, Google, and Intel is making strategic moves to loosen Nvidia’s grip. Their strategy focuses on targeting Nvidia's secret weapon: the CUDA software. These companies are part of a broader group of financiers and corporations aiming to reduce Nvidia's dominance in the AI landscape.
\"We're actually showing developers how you migrate out from an Nvidia platform,\" said Vinesh Sukumar, Qualcomm's head of AI and machine learning.
The initiative, spearheaded by the UXL Foundation—a consortium focused on open-source technology—plans to develop a suite of software and tools capable of powering various AI accelerator chips. Beginning with Intel's OneAPI, UXL aims to ensure that computer code can run seamlessly on any machine, irrespective of the underlying hardware.
\"It's about specifically – in the context of machine learning frameworks – how do we create an open ecosystem, and promote productivity and choice in hardware,\" explained Bill Magro, Google's director and chief technologist of high-performance computing.
The technical steering committee of UXL is set to finalize technical specifications early this year, with plans to refine these details to a mature state by year-end. The focus is on building a robust foundation that welcomes contributions from multiple companies, enabling deployment on any chip or hardware.
Beyond the initial members, UXL is aiming to attract major cloud-computing players such as Amazon and Microsoft's Azure, alongside additional chipmakers, to broaden its support base.
Since its inception in September, UXL has welcomed technical contributions from both founding members and external parties eager to leverage open-source technology. Intel's OneAPI is already operational, and the next phase involves establishing a standard programming model for AI computing.
With over $4 billion invested in 93 separate efforts aiming to dismantle Nvidia's software stronghold, the challenge is significant. Startups have raised more than $2 billion in 2023 alone to compete, up from $580 million the previous year, highlighting the intensifying effort to unseat Nvidia's leadership in AI.
Despite the substantial investments, success remains uncertain. Nvidia's CUDA platform remains robust, backed by continuous enhancements from both Nvidia and its developer community. \"But that's not what really matters,\" said Jay Goldberg, CEO of D2D Advisory. \"What matters is the fact that people have been using CUDA for 15 years, they built code around it.\"
Nevertheless, the push to create an open, competitive ecosystem in AI infrastructure signals a significant shift in the industry's dynamics, promising more choices and fostering innovation across the global tech landscape.
Reference(s):
Behind the plot to break Nvidia's grip on AI by targeting software
cgtn.com
