Technology has always been at the heart of trading, but the pace of change today is faster than ever. Firms are constantly evolving—seeking ways to process more data, execute trades more efficiently, and uncover new opportunities in increasingly competitive markets.
As we move into 2025, several emerging technologies are reshaping the industry. GPUs are becoming more central to high-performance computing, AI-driven trading strategies are gaining traction, and firms are investing heavily in cloud infrastructure to handle massive workloads. At the same time, quantum computing is edging closer to real-world applications, and alternative datasets are playing a bigger role in decision-making.
This article, in two parts, takes a closer look at some of the most significant technology trends in trading right now. Drawing on insights from across the industry, it highlights where firms are focusing their efforts, how these advancements are changing the landscape, and the technical skills that will be most in demand. In part one, we focus in on Hardware Acceleration with GPUs, Generative AI and the use of Big Data and Alternative Datasets. Part two explores what’s happening in Quantum Computing, Cloud-Enhanced HPC, and Real-Time Risk Assessment. So plenty to cover!
Hardware Acceleration with GPUs
For years, the trading industry has been locked in a latency “arms race,” often involving FPGAs or even ASICs to shave off microseconds. While ultra-low latency still matters, many major players have reached a point of diminishing returns—everyone’s latencies are converging. Because of that, more trading firms are now turning to GPUs (Graphics Processing Units) for a competitive edge in areas beyond just speed, particularly machine learning, data analytics, and massively parallel computations.
A prime example of this trend is XTX Markets’ recent announcement that it plans to build a $1bn data centre hub in Finland. As one of the biggest players in algorithmic trading, XTX’s investment reflects the growing need for extreme computational power to support AI-driven trading and large-scale data analysis.
All the major players have ongoing investments in this area, but here are some examples:
- XTX Markets is making a $1bn investment in a high-performance computing data centre in Finland to power its algo-trading operations.
- Jump Trading is reputed to invest heavily in GPU-powered HPC (High-Performance Computing) clusters for complex data analysis and advanced backtesting.
- Citadel Securities also makes use of GPU solutions for certain research workloads, complementing its ultra-low-latency infrastructure.
Key Technical Skills
- GPU Programming & Parallel Computing – Deep expertise in CUDA, OpenCL, and SYCL for optimising computations on NVIDIA and AMD GPUs. Experience with SIMD (Single Instruction, Multiple Data) architectures is also valuable for high-speed calculations.
- Low-Latency Trading Infrastructure – Understanding of FPGA-to-GPU communication, direct market access (DMA), and kernel bypass networking (DPDK, RDMA, or Solarflare NICs) to minimise latency in trading applications.
- High-Performance Computing (HPC) for Trading – Knowledge of multi-GPU scaling, distributed computing frameworks (MPI, NCCL), and GPUDirect RDMA to enable ultra-fast processing of market data.
- Machine Learning & AI for Quantitative Research – Experience with deep learning frameworks (TensorFlow, PyTorch), reinforcement learning for trading strategies, and AI model optimisation (TensorRT, cuDNN, and ONNX Runtime).
- Numerical Computing & Risk Modelling – Proficiency in GPU-accelerated libraries (cuBLAS, cuSPARSE, RAPIDS AI, XLA, and H2O.ai) for Monte Carlo simulations, derivatives pricing, and real-time risk calculations.
- Trading Systems & Market Data Processing – Familiarity with tick-by-tick data processing, order book reconstruction, and GPU-powered statistical arbitrage models.
- Cloud-Based GPU Computing – Experience deploying GPU-accelerated workloads in AWS (EC2 P4d, G5 instances), Google Cloud (A100, H100 TPUs), or Azure (ND A100 v4 series).
- Performance Optimisation & Profiling – Expertise in NVProf, Nsight Compute, NVIDIA Apex, and PyTorch Lightning to fine-tune GPU performance and minimise bottlenecks in trading algorithms.
Further Reading
- XTX to Build $1bn Data Centre in Finland
- NVIDIA – Creating a New Era of Intelligent Trading
- GPU Optimization For Trading Algorithms
- As the Latest FPGA Technology from AMD Sets the Gold Standard, where Next for Ultra-Low Latency Trading?
Generative AI in Trading
Generative AI has rapidly advanced, offering sophisticated models that assist in creating novel trading strategies and summarising complex research data. A significant recent development is the emergence of DeepSeek, a Chinese AI company that has introduced a cost-effective model rivalling established platforms like OpenAI’s ChatGPT. DeepSeek’s model has demonstrated efficiency and effectiveness, prompting discussions about its potential impact on the trading industry.
In addition to proprietary developments, trading firms are increasingly exploring the integration of third-party generative AI tools. For instance, some proprietary trading firms have utilised ChatGPT to enhance daily operations, such as aiding traders in self-reflection to identify cognitive biases that may affect decision-making.
- DeepSeek, originally a quant hedge fund, DeepSeek has transitioned into AI development, creating models that could influence trading strategies.
- Point72, founded by Steve Cohen, Point72 has launched the Turion Fund, focusing on AI assets and raising nearly $1.5 billion. The fund achieved a 14% return within its first three months, reflecting the firm’s commitment to integrating AI into its investment strategies.
- Various Proprietary Trading Firms are experimenting with ChatGPT to improve efficiency and productivity, particularly in areas like research and communication.
Key Technical Skills
- Programming Proficiency: Expertise in languages such as Python, R, and Java, which are essential for developing and implementing AI models.
- Machine Learning and Deep Learning: A solid understanding of machine learning algorithms and deep learning frameworks like TensorFlow and PyTorch is crucial for developing predictive models.
- Natural Language Processing (NLP): Skills in NLP are vital for processing and analysing large volumes of textual data, enabling the extraction of meaningful insights from unstructured information.
- Data Engineering: The ability to design and manage data pipelines ensures that AI models have access to clean, relevant, and timely data.
- Model Fine-Tuning and Optimization: Experience in refining AI models to enhance performance, including techniques like hyperparameter tuning and model compression.
- Ethical and Regulatory Awareness: An understanding of the ethical considerations and regulatory frameworks governing AI use in finance is essential to ensure compliance and responsible deployment.
Further Reading
- DeepSeek Chief’s Journey From Math Geek to Global Disruptor
- Point72’s New AI Fund Near $1.5 Billion After Double-Digit Returns
- How Generative AI Can Be a Game Changer in Online Trading
Big Data and Alternative Datasets
Data has always been the backbone of trading, but today, firms are looking far beyond traditional financial reports and market feeds. Alternative data—ranging from satellite imagery and credit card transactions to social media sentiment—has become a core input for trading strategies. These non-traditional datasets provide insights that might not be reflected in price movements yet, helping traders get ahead of the market.
Advances in AI and machine learning have made these vast datasets more valuable than ever. Rather than relying solely on structured reports, firms now apply predictive analytics to everything from shipping logs to web traffic patterns. The ability to extract actionable signals from unstructured, high-volume data is becoming a key competitive advantage.
Two of the firms leveraging this approach are:
- Goldman Sachs has been mining alternative data sources like geolocation tracking and consumer sentiment to refine forecasts for retailer performance. By analysing foot traffic and transaction trends, the firm aims to predict earnings surprises before official reports. (Reuters)
- Balyasny Asset Management recently revamped its equities division, integrating AI-powered data analysis into its research process. The firm is shifting towards a more data-driven investment strategy, leveraging machine learning to identify patterns in alternative datasets. (Business Insider)
Key Technical Skills
- Data Science & Machine Learning – Extracting meaningful signals from noisy, unstructured data requires a strong grasp of predictive modelling, NLP, and anomaly detection techniques.
- Big Data Engineering – Working with alternative datasets often means handling high-volume, high-velocity data streams. Experience with tools like Apache Spark, Hadoop, and cloud-based storage solutions is key.
- Financial Domain Knowledge – Knowing what data matters is as important as knowing how to process it. A solid understanding of market dynamics helps traders and quants separate signal from noise.
- Data Ethics & Compliance – Many alternative datasets contain sensitive or private information. Understanding GDPR, SEC, and FCA regulations ensures firms remain compliant while leveraging data effectively.
Further Reading
- Investors Mining New Data to Predict Retailers’ Results
- Hedge Funds Planning Increased Budgets for Alternative Data in 2025
- The Rise of Alternative Data & Machine Learning in Finance
Coming Up Next: Part 2
The trading landscape is evolving rapidly, and we’ve only covered the first few key trends. Stay tuned for Part 2, where we’ll dive into Quantum Computing, Cloud-Enhanced HPC, and Real-Time Risk Assessment—technologies that are redefining what’s possible in trading.