Chinese LLM developer DeepSeek has the potential to forge more efficient, secure and personalised financial services at a fraction of
their current costs. Reuters reported this week that while
Europe has struggled to keep pace with the US in regards to AI, a low-cost LLM alternative could help democratise this technology.
Bernstein analysts have estimated that DeepSeek’s pricing is 20 to 40 times cheaper than equivalent models from
OpenAI. “OpenAI charges $2.5 for 1 million input tokens, or units of data processed by the AI model, while DeepSeek is currently charging $0.014 for the
same number of tokens,” the article read.
While these cost savings are substantial, and innovative engineering techniques are leveraged rather than relying on large computational resources, fintech startups now have a new avenue to compete with larger firms. Here’s everything you need to know about
DeepSeek and its impact on fintech.
Why is DeepSeek in the news?
DeepSeek is currently being investigated for replicating OpenAI data and censoring negativity against China by a number of European countries, while at the same time, US President Donald Trump unveiled
Stargate, a $500 billion AI project in collaboration with OpenAI, Softbank and Oracle.
The Chinese organisation, although having launched in 2023, has recently dominated headlines after it revealed that DeepSeek’s V3 required under $6 million worth of computing power from
Nvidia H800 chips
– not their most advanced chips. Since then, DeepSeek has overtaken ChatGPT as the most popular productivity application.
Benefits of DeepSeek on the fintech industry
David Krause, emeritus professor, finance department, Marquette University in Wisconsin, published a
paper
last week on the democratisation of AI and its global implications. He highlights that these new developments have welcomed a “new paradigm in fintech by making high-performing AI models accessible at significantly lower costs.”
Krause lists the key aspects of DeepSeek’s potential:
- Cost-Effectiveness: “DeepSeek-R1 delivers performance comparable to GPT-4 but at a fraction of the cost, developed for just $6 million compared to GPT-4’s $100 million.”
- Open-Source Collaboration: “DeepSeek fosters wider access to advanced AI tools, encouraging collaboration and innovation within the global AI community.”
- Efficient Engineering: “DeepSeek’s innovative design, including multi-head latent attention (MLA) and mixture of experts (MoE) architectures, minimises computational requirements while maintaining high performance.”
- Strategic Research Focus: “DeepSeek's ‘reasoning’ model, designed to compete with a state-of-the-art offering by OpenAI, engages in self-dialogue before answering a query, a process that enhances the quality of its responses but also increases
electricity usage and raises costs as output quality improves.”
- Challenge to Silicon Valley: “The company's success may compel larger tech firms to reassess their strategies and adopt more cost-effective approaches to AI development.”
DeepSeek fintech use cases
The technological innovation that DeepSeek provides can improve the level at which AI is currently being leveraged. Using open source AI models can automate tasks like credit scoring, compliance checks, or customer service. Further to this, with a democratic
use of AI, real-time analysis can empower financial institutions to make quicker and more informed decisions.
In addition to these operational improvements to customer experience, enabling personalised financial services or tailored investment strategies allows organisations to enhance products and services through customisation. While this is all possible with
AI, it can be done at a cheaper rate with DeepSeek.
What this means is that financial services can be made more accessible, especially for underserved populations or in emerging marketing. Financial products can also be simplified, perfect for those communities with low financial literacy, or with high rates
of fraud risk. By challenging the traditional financial system and introducing alternative models for payments, lending or investment, DeepSeek can ensure scalability and growth of the sector.
In Tomasz Godziek, portfolio manager of the Tech Disruptors fund at J. Safra Sarasin Sustainable Asset Management’s view, DeepSeek’s model that “performs competitively against top-tier models such as OpenAI o1 and Anthropic's Claude 3.5 Sonnet on various
benchmarks” has sent shockwaves through equity markets.
Godziek says: “DeepSeek did not use Nvidia’s most advanced chips, as US export restrictions prevent Chinese firms from accessing them. While some skepticism remains regarding the actual training costs, the expenditure was just a fraction of what US firms
have been investing.”
He adds that techniques such as MLA and MoE has meant that this marks a “pivotal moment for AI investment. While it raises concerns for AI infrastructure providers, it could drive significant new opportunities in software, consumer internet, and inference
technologies such as autonomous driving.”
Challenges of rapid AI implementation in fintech
Returning to Krause’s paper, he also explains the challenges that come with rapid implementation of technologies like AI. “These include data security and privacy risks, regulatory concerns, ethical dilemmas, and the potential for market overcrowding. As
AI tools become more accessible, addressing these challenges will be critical to ensuring sustainable and equitable industry growth. Here’s a summary of what needs to be considered to mitigate these challenges:
- Implement robust cybersecurity measures,
- Ensure strong data governance frameworks,
- Maintain confidentiality of data,
- Comply with privacy laws and local regulations,
- Create a consistent regulatory approach across borders,
- Invest in transparent and accountable AI systems that mitigate bias and promote fairness,
- Emphasis the need for reskilling initiatives that help workers transition into new roles, and
- Focus on creating unique value propositions.