Mathpix Expands AI Training and Inference Infrastructure at DataVerge, Deploying Nvidia B300 GPUs to Power Real-Time AI Document Processing

DataVerge, owner and operator of Brooklyn’s largest carrier-neutral interconnection facility, announced that Mathpix, N AI-powered document automation and scientific communication company, is expanding its infrastructure at DataVerge’s Industry City facility to support AI training and real-time inference workloads.Building on their colocation relationship established in 2024, Mathpix is deploying Nvidia B300 GPU servers to meet growing demand for its document conversion and structured data extraction services.

Mathpix converts documents such as PDFs, handwritten notes, equations, and scanned files into structured, machine-readable text that feeds AI applications and automated workflows across enterprise, research, and financial markets. By colocating its AI training and inference servers at the Industry City facility rather than routing through distant cloud regions, Mathpix keeps processing close to where its New York-area users and enterprise customers are, cutting latency, improving reliability, and increasing cost savings.

Nvidia B300s are purpose-built for production-scale training and inference, delivering significantly higher throughput and memory capacity than prior generations. These GPUs give Mathpix the hardware foundation to handle larger models, more concurrent workloads, and enterprise API traffic at the speed and consistency its customers require.

“For Mathpix, AI training and inference performance are product features,” said Nico Jimenez, CEO, Mathpix. “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. For both fine-tuning models and real-time inference, milliseconds matter when processing user uploads, enterprise batch jobs, and API-driven workflows. DataVerge gives us the ability to deploy the B300s with the power density, connectivity, and hands-on support we need at a cost structure that makes sense.”

DataVerge supports GPU-intensive environments with high-density cold aisle containment pods, 24/7/365 remote hands, fast incident response, and deep operational experience managing production AI infrastructure. Its Meet-Me Room interconnects more than 35 top-tier carriers and network providers, giving Mathpix broad carrier choice, redundant routing paths, and optimized connectivity for latency-sensitive API traffic. The facility is designed to expand power, cooling, and network capacity together as demand increases, so Mathpix can scale in place rather than having to move to a larger facility. With 1.5 MW available today and an additional 3 MW coming online in Q2 2027, DataVerge offers Mathpix the headroom to grow on its own timeline.

“The B300 delivers a step change in AI inference capability, enabling larger models, more concurrent workloads, and faster response times at enterprise scale,” said Ray Sidler, CEO and co-founder, DataVerge. “But supporting that hardware in production requires high-density power and cooling, fast incident response, and the network architecture to keep latency low. AI inference companies don’t just need somewhere to put GPUs; they need production-grade infrastructure that’s close to their users and can scale without disruption.”