How an Innovative AI Model Impacted the Entire AI Ecosystem
The wild market fluctuations caused by the release of China-developed DeepSeek R1 underscored how breakthroughs in one part of the AI supply chain can impact the entire chain—from hardware and cloud services to software and applications. This report analyzes the impacts and offers a framework for financial institutions (FIs) to navigate future disruptions.
The combination of reinforcement learning and a mixture of experts model architecture allows DeepSeek to match the performance of other AI models while dramatically reducing the computational capacity required. These software-led innovations lower barriers to entry and alter competitive dynamics across the AI supply chain.
Despite efficiency gains, DeepSeek R1 raises significant security and privacy concerns, having failed some safety tests. Privacy regulators in Australia have raised flags, U.S. Navy has banned it outright, and institutions like Goldman Sachs are pausing adoption pending a security review.
From a geopolitical perspective, DeepSeek’s success underscores China’s progress in narrowing the AI gap with the U.S, after having been hamstrung in the AI arms race by U.S.-imposed restrixctions on premium AI chips. Improvements to software in this case have reduced reliance on top-tier hardware, and eased some of the pressure to scale renewable energy production. The report includes a breakdown of potential impacts for each element of the AI supply chain.
For FIs, this evolving landscape means adapting risk frameworks to include AI-specific third-party risks, diversifying providers, and conducting rigorous audits and security reviews. Understanding each layer of the AI supply chain is crucial for anticipating shocks and staying resilient in a fast-changing technological and geopolitical environment.
Additional content: