The artificial intelligence ecosystem is experiencing one of its most intense debates: should AI models be open or closed? Jensen Huang, CEO of Nvidia, has taken a strong stance by signing an open letter defending open-weights models. For Huang, the ability to download, modify, and run models on one's own infrastructure not only strengthens cybersecurity but also accelerates innovation and enables technological sovereignty. But behind this position lies a strategic calculation: Nvidia sells hardware to all market players, and a diverse ecosystem —with both open and closed models— maximizes its revenue.
The letter, backed by Meta, Microsoft, IBM, and Dell, emerges as a response to the growing anti-open-weights rhetoric from companies like Anthropic and OpenAI. These firms argue that open models, especially those from China, pose a security risk and enable malicious uses. However, Huang knows that if the United States restricts access to Chinese models such as DeepSeek, Kimi, or MiniMax, American companies will be left with limited options: increasingly expensive proprietary models from OpenAI or Anthropic, or uncompetitive local open models. His bet is clear: pressure US developers to produce high-level open models, while continuing to invest billions in AI startups, as he has done with OpenAI and Anthropic.
For companies looking to adopt artificial intelligence pragmatically, this debate has direct implications. At Q2BSTUDIO, we develop custom applications that integrate AI, intelligent agents, and automation capabilities, deployed on AWS or Azure cloud, with BI/Power BI dashboards that enable data-driven decisions. The availability of open models allows us to offer more flexible and cost-effective solutions to our clients, without relying on single vendors. For example, a customer service system based on AI agents can be trained on open models and hosted on private infrastructure, reducing recurring costs and ensuring data privacy.
Anthropic's stance, led by Dario Amodei, is more nuanced: it does not seek to ban open models, but to control those from authoritarian governments and mandate safety testing. Amodei also advocates restricting the export of high-performance accelerators to China and combating industrial distillation. Meanwhile, OpenAI, Meta, and xAI have shown support for Huang's letter, though concrete actions remain to be seen. OpenAI already launched GPT-OSS, a 120-billion-parameter open model, but it quickly became outdated compared to Chinese advances.
Cybersecurity is another key front. Nvidia has announced the Open Secure AI Alliance, aiming to equip defenders with open-source tools to protect against AI-powered cyber threats. This move directly connects with the need for companies to have robust cybersecurity. At Q2BSTUDIO we offer pentesting and security audit services, helping organizations identify vulnerabilities in both their custom applications and cloud infrastructures. Combining open models with a solid security strategy enables innovation without unnecessary risks.
For technology leaders, the choice between open and closed models is not binary. Each organization must assess its technical capacity, budget, and data sovereignty requirements. Large corporations can afford cutting-edge proprietary models, but SMEs and startups find in open models a gateway to high-performance AI without compromising financial viability. Moreover, AI agents, process automation, and business intelligence directly benefit from a competitive market where multiple options coexist.
Ultimately, Jensen Huang's push for open models is not a philanthropic gesture: it is a masterstroke to keep Nvidia as the leading hardware supplier in a diverse ecosystem. Companies that bet on flexible solutions, like those we develop at Q2BSTUDIO, must closely follow these developments. The ability to choose between open and closed models, deploy them on public or private cloud, and complement them with BI and cybersecurity tools defines the digital strategy of the future. And in that future, artificial intelligence will be both a battlefield and an opportunity for growth for those who know how to adapt.
