The AI revolution has been driven by the rapid evolution of Large Language Models (LLMs), but their high cost has remained a barrier to widespread adoption. DeepSeek has rewritten the rules, achieving a remarkable breakthrough in cost reduction while maintaining performance, unlocking new opportunities for businesses and investors alike.
Since the emergence of ChatGPT in late 2022, LLMs have transformed AI, enabling machines to generate and understand human-like text with astonishing accuracy. However, the substantial computational power and energy consumption required to build and operate these models have posed a major financial hurdle for companies deploying AI at scale. DeepSeek has upended this challenge with its R1 LLM, delivering performance comparable to OpenAI’s models at a fraction of the cost—more than 90% lower. This seismic shift makes AI more accessible, fosters innovation, and democratises cutting-edge technology.
Despite US export bans on advanced chips, DeepSeek’s team has optimised its hardware to push the boundaries of efficiency. While questions remain about the full costs behind model training, R&D, and reliance on prior open-source models, R1 represents a significant leap in AI affordability. The implications are profound, paving the way for broader AI adoption and commercial viability at a scale previously unattainable.
The financial markets have been quick to react. AI infrastructure stocks took a hit, with Nvidia losing $650 billion in market capitalisation in a single day amid fears that lower-cost AI models would reduce demand for high-performance chips and data centres. However, this initial reaction overlooks a key factor: cheaper AI models could actually fuel greater demand for computation and energy in the long term. More businesses and individuals gaining access to AI means increased usage, potentially driving an even larger market for infrastructure providers over time.
The market’s nuanced response was evident as Nvidia rebounded sharply the following day, indicating that investors are beginning to recognise the potential for sustained AI demand. Meanwhile, cloud computing giants like Microsoft (Azure) and Amazon (AWS) stand to gain from broader AI adoption, as lower costs enable more businesses to scale their cloud-based AI applications. Software development will likely accelerate, reinforcing the Jevons paradox—the principle that as efficiency increases, consumption rises rather than falls.
That said, the playing field in AI is shifting. DeepSeek’s success demonstrates that smaller firms can achieve groundbreaking advancements with fewer resources, challenging the dominance of established tech giants. This disruption raises the question of whether today’s AI leaders will maintain their competitive edge in the coming decade.
For investors, the key lies in navigating both risks and opportunities. The rapid evolution of AI presents new challenges, but it also opens doors to transformative growth. As AI adoption expands, the market will continue to adjust, and forward-thinking investors will be well-positioned to capitalise on the next wave of innovation.
DeepSeek’s breakthrough underscores a broader trend: the democratisation of AI. As costs decline and technology advances, the AI landscape is set to become more competitive, dynamic, and accessible than ever before.
Arbuthnot Banking Group PLC (LON:ARBB), trading as Arbuthnot Latham, provides private and commercial banking products and services in the United Kingdom. Founded in 1833, Arbuthnot Banking is based in London, United Kingdom.
