Survey: Financial Services Face AI Security and Performance Challenges
Artificial intelligence offers exciting new possibilities for financial services use cases from real-time credit decisions to trade execution, but only if the underlying infrastructure can keep up. According to the A10 Networks State of AI Infrastructure Report 2025, that constraint has become a major pain point. While financial firms lead all sectors in predictive analytics adoption, they also top our survey in citing network latency as a major problem. Unless they can modernize their infrastructure in tandem with AI innovation, these new capabilities simply won’t be viable in production.
In this blog, we’ll examine what the survey data reveals about how financial services organizations are deploying AI, where their infrastructure is falling short, and what the modernization agenda looks like for teams navigating these challenges.
Balanced Hosting Brings Operational Complexity
Seventy-one percent of financial services firms surveyed are using AI for predictive analytics, the highest rate of any industry. Use cases like risk assessment, fraud detection, and market analysis all depend on AI models that process large data volumes quickly and return reliable outputs under time pressure. To support these workloads, forty-six percent of financial services respondents use a balanced hybrid cloud approach—for example, keeping latency-sensitive and highly regulated workloads on-premises while development environments and scalable inference move to the cloud. This is a sound strategy, but it introduces operational complexity that compounds quickly when the underlying infrastructure isn’t purpose-built to handle AI traffic across both environments.
Performance Issues Come at a Cost
Across all organizations surveyed, 49 percent rate AI application performance as “important,” and another 23 percent rate it “extremely important.” In the financial services sector, the bar is set even higher. Lower latency enables more successful trades, while infrastructure that introduces delays can result in failed transactions and lost revenue. However, one-third of financial firms cite compute limitations as a current performance bottleneck, while 20 percent flag network latency—a rate higher than most other industries in the study.
The application delivery picture reinforces the concern. Half of all respondents say that their current infrastructure—including load balancers, application delivery controllers (ADCs), and related systems—can only “mostly” maintain the required performance and uptime for AI workloads. Only 16 percent say their infrastructure meets AI demands with capacity to spare.
Security: Significant Concern, Insufficient Response
Security is the single most widely cited infrastructure pain point across the survey, with 49 percent of respondents identifying it as their biggest limitation. For financial services firms, the stakes are especially high. Beyond the operational risks that affect every industry, they operate under regulatory frameworks—including DORA, GDPR, and sector-specific data sovereignty requirements—that make AI-related security failures a compliance event as well as a technical one.
Still, despite security ranking as the top concern, only 40 percent of organizations across industries have deployed AI-specific security solutions. Instead, a majority of companies are relying on existing infrastructure to protect AI workloads and traffic they weren’t designed to understand. Many respondents were aware of the risks they face, naming fears such as data leakage into AI training models, unauthorized AI access to sensitive systems, and the inability of legacy security tools to detect anomalous behavior at the model or prompt level.
Deploying Secure, High-performance AI Infrastructure
While compute is a perennial constraint for current AI initiatives, security poses an even more urgent challenge. Among the 79 percent of all organizations planning to modernize their infrastructure within 18 months, security infrastructure tops the priority list at 60 percent—including AI firewalls, WAF, API security, and DDoS protection. As they proceed, 62 percent of respondents prefer vendors with a platform strategy over standalone point products. This approach can be especially valuable in financial services, where adding multiple specialized vendors creates integration burden and visibility gaps that can exacerbate compliance challenges.
To learn more about how financial services organizations are approaching AI infrastructure modernization, download the A10 Networks State of AI Infrastructure Report 2025.
