BBVA Accelerates Derivative Pricing with Riskfuel AI Integration
By Lauren Towner · 20 November 2026
In brief: BBVA has deployed Riskfuel's AI neural surrogates within its ePricer platform, enabling the bank to price complex derivatives and autocallables in seconds rather than hours.
| Performance Metric | 2.27 million valuations per second |
|---|---|
| Processing Time | Under 1 minute for thousands of trades |
| Input Features | Over 4,000 |
| Hardware Platform | Intel Xeon 6 processors |

BBVA has integrated Riskfuel’s neural surrogate technology into its ePricer platform, moving AI-driven derivatives pricing from proof-of-concept into live production. For fintech professionals, this represents a significant shift in capital markets workflows, enabling the near-instantaneous evaluation of complex autocallable structures that previously required hours of computationally intensive stochastic simulation.
What was announced
BBVA is deploying a deep neural network surrogate developed by Riskfuel to accelerate the pre-trade structuring of autocallable products. The tool is now part of the bank's ePricer platform, used by structuring teams to design and compare trades for clients. Unlike generic models, this surrogate is trained on data from BBVA’s own proprietary pricing model, ensuring consistency with the bank's internal risk frameworks. It processes over 4,000 input features, including interest rate curves, full volatility surfaces, and discrete dividend schedules, alongside structural elements like knock-in barrier levels and autocall triggers.
The deployment follows rigorous testing where the surrogate was measured against BBVA’s reference model across 100,000 test trades with maturities reaching five years. Performance benchmarks conducted with Intel on Xeon 6 processors showed the system capable of approximately 2.27 million valuations per second. This speed allows structuring desks to evaluate thousands of candidate trades in less than a minute—a process that formerly took hours. Crucially, the technology runs on general-purpose Intel hardware using built-in matrix acceleration, meaning BBVA does not need to invest in specialized AI accelerators or separate operating models to maintain the system.
"What is important here is the combination of speed and consistency with our own pricing models. By integrating Riskfuel's technology into ePricer, we can dramatically accelerate the evaluation of different autocallable structures while keeping BBVA's pricing framework at the core of the process. That gives our structuring teams a much more efficient way to explore and compare alternatives."
Francisco Gómez Casanova, Executive Director, XVA Quantitative Analysis, BBVA Corporate & Investment Banking.
The companies involved
BBVA is a global financial services group and one of the largest financial institutions in the world, with a particularly strong presence in Spain, Mexico, and South America. The bank has increasingly positioned itself as a digital leader in the Eurozone, frequently integrating advanced technology into its corporate and investment banking (CIB) operations. Its CIB division provides a range of services to corporate and institutional clients, including the structured products handled by the ePricer platform.
Riskfuel is a financial technology firm specializing in the use of artificial intelligence to accelerate the pricing and risk management of complex financial derivatives. The company focuses on "neural surrogates"—AI models that learn to replicate the behavior of traditional, slower quantitative models. By replacing computationally heavy simulations with high-speed neural networks, Riskfuel aims to remove the latency bottlenecks that have traditionally limited the ability of trading desks to explore complex product structures in real-time. This partnership marks a transition for the firm from experimental testing to a practical, day-to-day tool within a major global bank’s workflow.
What FF News has reported before
FF News has followed BBVA’s aggressive adoption of artificial intelligence across its various business units. Recently, the bank announced it was leveraging Transformer AI models to enhance personalization within its mobile banking app. The institution has also focused on the security implications of these technologies, partnering with NVIDIA to secure autonomous AI agents. Beyond AI, BBVA has seen significant growth in its digital-first initiatives, including hitting the one million customer milestone in Italy through its 100% digital model and launching integrated mobile eSIM services in collaboration with Gigs and ING.
What this means
This deployment signals a shift in how investment banks handle the "compute crunch" of modern derivatives trading. For years, the industry has struggled with the trade-off between model accuracy and computational speed. By successfully moving a neural surrogate into a live production workflow, BBVA is proving that AI can do more than just summarize text or chat with retail customers; it can handle the heavy lifting of quantitative finance. This puts pressure on other Tier-1 banks to move beyond legacy Monte Carlo simulations. The reliance on standard Intel hardware rather than specialized GPUs also suggests a more sustainable, cost-effective path for AI integration in capital markets.
Frequently asked questions
How does BBVA use Riskfuel for derivative pricing?
BBVA uses Riskfuel's neural surrogate technology within its ePricer platform to handle pre-trade structuring for autocallable products. The AI replicates BBVA's internal pricing models to provide high-speed valuations that remain consistent with the bank's existing risk framework.
What are the performance benefits of Riskfuel's AI?
Riskfuel's technology allows for approximately 2.27 million valuations per second when running on Intel Xeon 6 processors. This enables structuring teams to evaluate thousands of candidate trades in less than a minute, a task that previously took several hours.
Does Riskfuel require specialized AI hardware?
No, Riskfuel's neural surrogates run on general-purpose Intel Xeon processors. By utilizing built-in matrix acceleration like Intel AMX, the system avoids the need for specialized accelerator hardware or proprietary AI operating models.
About the companies
Riskfuel
1 story on FF News since 2026
Companies in this story: BBVA, Riskfuel
People in this story: Francisco Gómez Casanova, Ryan Ferguson