DeepSeek: disruption through open-source AI

Open-source AI models are becoming crucial drivers of accessibility and innovation in the evolving artificial intelligence landscape. DeepSeek stands out for its remarkable accomplishments, challenging the dominance of tech giants’ through disruption in the generative AI space.

This Chinese AI startup has made headlines with its advanced AI model, which performs comparably to OpenAI’s ChatGPT. As an open-source model, DeepSeek can be downloaded and installed on local infrastructure without a charge, reducing expenses by minimising reliance on third-party cloud services, and making AI adoption more cost-effective.

DeepSeek's capacity to operate on local hardware will significantly affect the semiconductor industry. Unlike cloud-based AI models requiring powerful GPUs, DeepSeek enables businesses to implement AI on-premises. The emergence of effective open-source AI models could reduce reliance on cloud-based technology and shift demand towards more affordable options. However, it might also drive advancements in edge AI computing, benefiting low-power AI chip manufacturers.

Foundries producing mature process nodes (such as 12nm, 28nm, and 40nm) may benefit, as these chips can effectively serve AI workloads. Currently, China remains the largest consumer of semiconductors among other countries. The Integrated Circuits market accounts for more than 80 % of the semiconductor market revenue and logic integrated circuits account for the major share of the revenue in this market.

Currently, the Asia-Pacific region accounted for more than 90% of global semiconductor assembly, testing, and packaging (ATP) capacity, led by China and Taiwan[1]. Countries such as Malaysia and Vietnam are forecasted to see a big increase in their shares of global chip production capacity. The semiconductor market will continue to be cyclical for the next few years due to major swings in the demand influenced by technological changes as well as changes in supply capabilities.

DeepSeek’s open-source model could potentially reduce cloud service costs for businesses and streamline AI adoption. This could potentially decrease the need for cloud-computing real-estate and reduce demand for hyperscale data centres. Conversely, it might increase the need for AI hardware creation, stimulating investment in the semiconductor industry. If foundries expand to meet the demand for less-than-leading-edge AI processors, the need for industrial real estate for chip fabrication and testing facilities may increase.

While the open-source concept offers many advantages, it also presents challenges. Maintaining the model's effectiveness and quality control is essential, and data privacy risks remains a concern. Open-source AI models could be vulnerable to security threats, unauthorised modifications, and compliance issues, requiring constant attention.

DeepSeek’s open-source model has lowered the entry barriers for AI development, allowing researchers and small businesses to take advantage of cutting-edge AI technologies without exorbitant costs. This democratisation promotes a more inclusive AI ecosystem, where innovation is available to a wider audience rather than just well-funded organisations. It has the potential to become an attractive choice for those desiring customisation and flexibility as it can be fine-tuned, retrained, or adapted for specific use cases. making it highly flexible for those who want tailored AI solutions. As we look to the future, this collaborative approach embodied by DeepSeek is set to shape the AI landscape, ensuring that cutting-edge developments are accessible across the board.

_{[1] Statista, 2025}