Samsung backs Nvidia AI chip rival in $230 million funding round as GPU alternatives boom
Source Entity
US Top News and Analysis
Two startups, Cornelis and Euclyd, have raised significant capital to challenge Nvidia's dominance in the AI hardware market. These firms are focusing on specialized networking fabrics and alternative chip architectures to optimize AI performance.
The Shift Toward Hardware Diversification in AI
The artificial intelligence landscape is witnessing a strategic pivot as investors pour hundreds of millions into companies aiming to dismantle Nvidia’s current market monopoly. Recent funding rounds, including a $205 million injection for Cornelis and a $231 million series for Dutch startup Euclyd, signal that the industry is moving beyond generic GPU reliance. As AI models grow in complexity, the bottleneck for performance is shifting from pure processing power to the efficiency of data movement and specialized inference architectures.
Breaking the Bottleneck: Cornelis and Active Compute Fabric
Cornelis, a 2020 spin-off from Intel, is targeting a critical inefficiency in current AI infrastructure: the time GPUs spend idling while waiting for data. By introducing its 'Active Compute Fabric,' the company seeks to enable simultaneous processing and data transmission. This approach addresses the 'Nvidia tax,' where users are often locked into proprietary networking ecosystems. By offering an open architecture, Cornelis provides a modular alternative that permits the integration of diverse accelerators, potentially lowering the barrier to entry for enterprises seeking scalable AI performance.
Euclyd’s Architectural Challenge to GPUs
While Cornelis focuses on networking, Euclyd is targeting the fundamental architecture of the chip itself. Founded in 2024, the company is developing a system tailored specifically for 'inference'—the process of running AI models once they are trained. Unlike Nvidia’s GPUs, which were originally optimized for gaming and repurposed for AI, Euclyd is designing a processor and memory architecture from the ground up to handle the specific demands of inference. With backing from Samsung, the startup joins a growing cohort of firms attempting to prove that general-purpose GPUs may not be the most efficient long-term solution for AI workloads.
The Strategic Role of Samsung and Venture Capital
Samsung’s participation in the Euclyd funding round underscores a broader trend: incumbent hardware giants are eager to diversify the AI supply chain. By backing European startups, these investors are hedging against the risks associated with Nvidia’s near-monopoly. This capital infusion is vital for startups attempting to bridge the gap between theoretical chip design and mass-market deployment, a challenge that historically requires massive R&D spending.
Future Trends: Toward Open and Specialized Hardware
Looking forward, the success of these companies will depend on their ability to integrate seamlessly into existing data centers. The industry is currently locked in a cycle where Nvidia’s hardware is deeply optimized with its own networking protocols. If firms like Cornelis and Euclyd can prove that open-source networking and specialized inference chips offer superior cost-to-performance ratios, we may see a fragmentation of the market. This transition would likely lead to a more competitive landscape, reducing the reliance on single-vendor ecosystems and driving down the cost of AI development.