Is Fintech's AI Bubble About to Burst?

May 2025

Fintech & Payments

Daniel Bedford

Senior Research Analyst

AI entered the fintech sector with sky-high expectations, being touted as a solution to everything from fraud detection and customer service to automated lending and hyper-personalised financial tools; leading many fintechs to embrace ‘AI-first’ strategies to stay ahead of their rivals.

However, cracks are clearly beginning to show, with a recent survey finding that 42% of companies had scrapped the majority of their AI projects by early 2025, up from just 17% a year earlier.

Generative AI, in particular, has proven difficult to translate into meaningful returns. Between 70% and 85% of deployments fall short of ROI expectations. As reported by The CFO in January 2025, only 47% of AI initiatives are turning a profit - while 14% are actually generating negative returns. With profitability slipping out of reach, many firms are now shifting focus to productivity gains instead, signalling a major recalibration in how success is measured.

Klarna’s Cautionary Case Study

Klarna’s widely publicised shift from human to AI-driven customer service quickly ran into trouble. While the AI assistant was able to handle two-thirds of chats and deliver rapid responses, users encountered familiar frustrations: inaccurate answers, robotic tone, circular interactions, and confusion around escalation when the AI failed.

CEO Sebastian Siemiatkowski later acknowledged that an overemphasis on cost-cutting had compromised service quality; a mistake mirrored across the industry, as more consumers report poor experiences with chatbots lacking empathy or clarity. In financial services, where issues are often complex and stakes high, customers continue to favour human interaction.

Klarna’s reversal underscores a broader truth: AI can streamline basic queries, but it still struggles with nuance, emotional intelligence, and context. Human agents remain essential to delivering the kind of trust, understanding, and resolution that customers expect.

Key Pressure Points 

Alongside challenges around ROI and declining customer satisfaction, several other obstacles continue to stand in the way of successful AI adoption in fintech:

• Data privacy issues - Handling sensitive financial data with AI raises major privacy and compliance risks, especially as regulations tighten and data volumes grow.
• Security risks - AI systems can introduce new cybersecurity threats; making robust fraud detection and real-time risk monitoring essential.
• Poor data quality - Poor or inconsistent data undermines AI accuracy; leading to unreliable outputs and compliance risks.
• Scalability challenges - Integrating AI into legacy systems and scaling across operations is complex and costly, especially for established financial institutions.
• Bias risks - AI models can amplify biases in training data, as its lack of explainability and transparency can risk unfair decisions and regulatory scrutiny, particularly in areas like lending and underwriting. 

An AI Bubble Burst...Or a Case for Evolution?

As previously mentioned, generative AI projects have had mixed prospects. Many AI projects stall after prototypes, due to poor data, rising costs, and unclear value fuelling executive scepticism about AI investments. In January 2024, we found that GenAI spend by banks reached $6 billion globally, with spend forecast to reach $85 billion in 2030; however, disappointing returns could trigger cutbacks. Reduced AI investment may slow productivity gains and innovation, potentially worsening broader economic slowdowns or recession risks.

Despite this, AI has delivered notable successes and improvements to many processes, notably back-office ones. Successful firms within fintech have prioritised AI use cases with clear ROI, such as fraud detection and compliance; avoiding the trap of chasing every AI trend. This targeted approach improves success rates and sustains investment.

Klarna’s desire to return to human agents after AI-driven quality drops highlights the need for AI to augment, not replace, humans. Hybrid models combine AI efficiency with human empathy and judgment, boosting customer satisfaction and long-term value. While fintech’s AI bubble faces risks from hype and economic pressures, a shift toward strategic, hybrid approaches offers a sustainable path forward - balancing innovation with practical outcomes.

The Path Forward 

As AI adoption outpaces regulation, fintechs must take the lead on proactive compliance and bias mitigation. This means implementing transparent algorithms, regular audits for fairness, and clear documentation to ensure decisions can be explained to both customers and regulators.

Staying ahead of evolving rules helps build trust, and reduces the risk of costly legal or reputational setbacks. Rather than investing in AI for its novelty, fintechs should focus on projects with clear ROI and scalable impact. This involves rigorous evaluation of each initiative’s business case, ongoing performance measurement, and the flexibility to scale up successful pilots or quickly sunset those which are underperforming. Such discipline ensures resources are directed to solutions that deliver real value.

Fintechs should prioritise user experience by designing AI tools that are transparent, intuitive, and easy to escalate to human support when needed. This not only addresses scepticism but also fosters loyalty and long-term engagement. The future of AI in fintech depends on responsible governance, disciplined investment, and a relentless focus on customer needs. Those who balance these priorities will lead the next wave of sustainable, impactful innovation.

While the AI bubble in fintech is not collapsing outright, it is undergoing a painful correction. The era of unchecked experimentation is giving way to strategic, ethical, and ROI-driven adoption. Fintechs who navigate this shift, by balancing automation with human oversight, will survive the bubble’s deflation and emerge stronger for it.

As a member of Juniper Research’s Fintech & Payments team, Dan analyses developments in financial and payments markets, and how these interplay with emerging technologies, such as blockchain and AI. His recent reports include Trade Surveillance Systems, QR Code Payments, and Future Leaders 100: Fintech.