Political limitations
There is no escaping the political background surrounding DeepSeek as a Chinese-developed open-source LLM. This means that DeepSeek is unlikely to ever see production-grade adoption in most Western institutions because of trust-related concerns; however, the mix of performance and efficiency claims does intrigue many data users and there are merits regarding its technical and commercial viability.
DataIQ members can click here to read the full findings on the benefits and concerns.
Strategic tensions
Although there are benefits touted by DeepSeek, data leaders must understand the limitations that exist and are likely to continue to exist. Smaller, efficient models like DeepSeek strongly hint towards lower costs, but the reality is that most organisations will still rely on expensive proprietary APIs.
Click here to discover all the strategic tensions of using DeepSeek.
Multimodal architectures become the norm
Data leaders across the industry are aggressively approaching generative AI as modular, multi-cloud, and multi-model opportunities. The last few years (particularly the 2020 pandemic) have shown that flexibility and resilience are core to success and evolution, so being able to embrace new tools that can be easily inserted, such as DeepSeek, is just another string to the bow: LLMs are now treated as interchangeable components in orchestration pipelines.
DataIQ members can learn about the multi-modal approaches senior data leaders are using to experiment with DeepSeek.
