Nvidia CEO Jensen Huang Backs Both Open and Closed AI Models
Huang emphasizes that Nvidia hardware powers all AI infrastructure regardless of model openness.
1Hardware as the Foundation
Nvidia CEO Jensen Huang maintains a neutral stance on the competition between open and closed AI models. During the Forge event in San Francisco, he clarified that Nvidia prioritizes powering the underlying infrastructure. Whether companies utilize proprietary frontier models or open-weight alternatives, Nvidia remains the primary hardware provider for these diverse computational needs.
2Strategic Value of Specialization
Huang argues that the most significant economic value lies in specialized intelligence rather than general models. He suggests that businesses can create superintelligence by adapting open frontier models with proprietary domain knowledge and specific tools. This approach allows organizations to leverage their unique data assets while maintaining control over their AI-driven operational capabilities.
3Supporting the Open Ecosystem
The company's commitment to open models is evidenced by its partnership with Fireworks.ai, which recently received 25,000 Nvidia GPUs to expand operations in Europe. This collaboration supports the growing demand for platforms that enable companies to train and run models tuned to their specific needs, bypassing the limitations of locked-in proprietary systems.
4Addressing Escalating AI Costs
Fireworks.ai founder Lin Qiao highlighted that unchecked token spending is making AI adoption unsustainable for many businesses. Without cost-effective alternatives, scaling becomes financially prohibitive. The industry is responding with solutions like Genspark’s Gen-1 Slides, which delivers performance comparable to Anthropic’s Opus 5 at a fraction of the cost, specifically 1/17th the price.
5The Rise of AI Agents
Huang identified agent harnesses as the primary breakthrough of the current year, transforming static models into active agents. He anticipates that next year, Open Shell will become a critical development, enabling companies to deploy these agents safely and confidently. This shift marks a transition from simple text generation to complex, autonomous task execution.
6Future of Hardware Integration
Looking toward future industry milestones, Huang predicted that hardware-in-the-loop systems will be the next major breakthrough. This concept focuses on integrating AI directly into robotic systems to bridge the gap between digital intelligence and physical action. This prediction aligns with Nvidia’s ongoing investments and strategic interest in the evolving field of robotics.
7Balancing Industry Demands
While Huang acknowledges the appeal of affordable open models, he avoids dismissing the role of closed frontier models from companies like OpenAI and Anthropic. He recognizes that closed models have a solid place in the industry. Nvidia’s strategy effectively captures value from both segments by providing the essential compute for all AI architectures.
8Implications for Enterprise Scaling
The current AI landscape is shifting toward specialized, domain-specific intelligence to ensure scalability and cost efficiency. By providing the hardware backbone for both open and closed ecosystems, Nvidia positions itself to benefit from any market outcome. Companies must now focus on integrating domain knowledge and hardware-in-the-loop systems to achieve true operational superintelligence.