A Chinese startup called DeepSeek has just released its biggest AI model so far, and it’s causing quite a stir. The model, named DeepSeek V3.1, has 685 billion parameters, making it one of the largest open-source AI models out there. It was quietly launched on Hugging Face, a popular platform for sharing AI tools, with little fanfare. Even though it wasn’t heavily promoted, early tests suggest the model performs similarly to some of the top proprietary models from companies like OpenAI and Anthropic.

This move could make advanced AI more accessible worldwide. It also raises questions about how power in the tech world is shifting. Traditionally, US companies have led in AI innovation, but China’s recent advances challenge that dominance. For businesses, this could mean reevaluating where they get their AI tools from, especially if open-source Chinese models prove reliable and cost-effective.

DeepSeek’s Bold Step and Its Impact

DeepSeek’s V3.1 isn’t just big; it’s also more accessible than many other models. Large models have historically been the domain of US firms, who keep their most advanced versions private to protect their market edge. But DeepSeek’s open-source approach means developers and companies can experiment freely with this huge model. Experts say this could speed up the “commoditization” of AI — making powerful AI capabilities available to more people, not just big corporations.

Oishi Mazumder, a senior analyst, points out that US companies like OpenAI and Google will need to compete differently now. Instead of just focusing on size and raw power, they’ll have to emphasize trust, governance, and enterprise-ready ecosystems. The open access to frontier-scale AI challenges the old model of proprietary, subscription-based systems, pushing the industry to rethink its strategies.

What This Means for Business and Global Power

While DeepSeek’s model is impressive, experts believe it might not immediately shake up the US market. Many US companies still prefer domestic vendors who offer strong support and security, especially for sensitive enterprise use. For DeepSeek’s models to gain real traction in the US, geopolitical tensions would need to ease, and the company would have to demonstrate clear superiority over other open-source options from Western firms like Meta and Nvidia.

However, outside the US, the situation could be different. The open licensing and customization options of DeepSeek’s models might appeal to CIOs wanting more control over their AI systems. Still, the hefty computing resources needed to run such large models could be a barrier. Companies must consider infrastructure costs, compliance rules, and export restrictions before adopting these models.

Neil Shah from Counterpoint Research notes that DeepSeek might struggle to match the support and security standards of established enterprise AI models like Anthropic’s Claude Sonnet 4. But the bigger picture isn’t just about individual companies. It’s about the global fight for technological leadership. The US has long been ahead in semiconductors, operating systems, and cloud services, but China is pushing hard to lead in AI despite facing sanctions and trade restrictions.

Shah emphasizes that DeepSeek’s progress underscores the intensifying rivalry. Both the US and China are racing to innovate, regardless of geopolitical hurdles. This ongoing competition signals a future where AI development is a key battleground for global influence, with each side striving to outdo the other in technological prowess.

Inspired by

• https://www.computerworld.com/article/4042964/chinas-deepseek-launches-v3-1-raising-stakes-for-enterprise-ai-adoption.html

Sources

• gmpg.org
• cnbc.com
• ft.com
• podigee.com
• podigee.com