In a strategic move to bolster its AI capabilities, Mathpix is expanding its operational infrastructure at DataVerge’s Industry City facility in Brooklyn. This expansion addresses the rising demand for its document automation and structured data extraction services, which are crucial for enterprises seeking efficiency in data processing.
Mathpix, known for transforming complex documents like PDFs and handwritten notes into structured, machine-readable formats, is deploying NVIDIA B300 GPU servers. This follows their initial colocation agreement established in 2024, positioning Mathpix to enhance its AI training and inference workloads while maintaining low latency and high performance.
Choosing to colocate at DataVerge instead of relying on distant cloud regions allows Mathpix to keep its processing infrastructure close to its primary user base in New York. This proximity reduces latency, enhances reliability, and ultimately results in cost savings. As Nico Jimenez, CEO of Mathpix, noted, "For Mathpix, AI training and inference performance are product features. Our customers expect near-instantaneous document conversion, which means our infrastructure needs to be close to them and built for the demands of modern AI workloads."
Expanding AI Capabilities with NVIDIA B300
The deployment of NVIDIA B300 GPUs represents a significant upgrade in Mathpix's hardware capabilities. These servers are designed for production-scale training and inference, offering superior throughput and memory capacity compared to previous generations. This enhancement enables Mathpix to manage larger AI models, handle more concurrent workloads, and efficiently support enterprise API traffic. Jimenez highlighted the importance of speed in processing user uploads and enterprise batch jobs, stating that "milliseconds matter" in their operations.
The partnership with DataVerge further strengthens Mathpix's infrastructure. DataVerge’s Brooklyn facility features high-density cold aisle containment pods and 24/7 operational support, ensuring optimal conditions for GPU-intensive environments. The facility’s extensive interconnectivity with over 35 top-tier carriers offers Mathpix multiple redundant routing paths, optimizing connectivity for latency-sensitive applications.
Future Growth and Scalability
Looking ahead, Mathpix is well-positioned for growth as DataVerge plans to expand its power and cooling capabilities. Currently providing 1.5 MW, an additional 3 MW will be available by Q2 2027, allowing Mathpix the flexibility to scale its operations without relocating. This expansion will give Mathpix the necessary headroom to grow in line with the increasing demand for its services.
Ray Sidler, CEO and Cofounder of DataVerge, commented on the significance of the B300's capabilities, stating, "The B300 delivers a step change in AI inference capability, enabling larger models, more concurrent workloads, and faster response times at enterprise scale. Supporting that hardware in production requires high-density power and cooling, fast incident response, and the network architecture to keep latency low."
As Mathpix continues to enhance its infrastructure, its commitment to advancing technological progress in document automation remains strong. By prioritizing performance and proximity to its users, the company aims to redefine efficiency standards in AI-driven applications, particularly within enterprise and research environments. Mathpix's expansion at DataVerge not only represents growth for the company but also marks a significant evolution in AI infrastructure capabilities to meet modern workflow demands.
The stories that move AI & crypto markets — before the market reacts.
Free. 7am ET. Five stories. 62,400 readers.
