DeepInfra Opens Toronto AI Data Center with 1,000+ NVIDIA Blackwell B300 GPUs
DeepInfra opened its first international data center location in Toronto, deploying more than 1,000 NVIDIA Blackwell B300 GPUs to expand capacity for large-scale AI inference workloads. The 1.7 MW facility marks the company’s ninth data center location and extends its infrastructure footprint beyond the United States as enterprises increase production deployments of generative AI and agentic applications.
The Toronto cluster follows DeepInfra’s $107 million Series B funding round announced in May 2026. The company said the investment will support global expansion of its purpose-built inference cloud, developer tooling, and next-generation model deployments. DeepInfra owns and operates its GPU infrastructure and processes nearly five trillion tokens per week across a platform supporting more than 200 open-source models through OpenAI-compatible APIs.
DeepInfra positions the Toronto deployment as part of a distributed infrastructure strategy designed to place GPU capacity closer to customers, users, and data. The company operates eight U.S. data center locations and said additional international deployments remain under evaluation. DeepInfra reported in May that its revenue had tripled since the beginning of 2026 and that nearly 30% of its weekly token volume came from agent-based systems.
• Facility capacity: 1.7 MW
• GPU deployment: More than 1,000 NVIDIA Blackwell B300 GPUs
• Infrastructure footprint: Nine data center locations, including eight in the United States and the new Toronto cluster
• International expansion: Toronto marks DeepInfra’s first data center location outside the United States
• Series B funding: $107 million announced in May 2026
• Investors: Round co-led by 500 Global and Georges Harik, with participation from A.Capital Ventures, Crescent Cove, Felicis, NVIDIA, Peak6, Samsung Next, Supermicro, and Upper90
• Platform scale: Nearly five trillion tokens processed per week
• Model portfolio: More than 200 open-source models supported through OpenAI-compatible APIs
• Infrastructure strategy: DeepInfra owns and operates its GPU infrastructure and optimizes its computing stack for high-throughput inference workloads
“Enterprises are moving from experimentation to production at unprecedented speed, and that shift demands infrastructure that is both scalable and globally distributed,” said Nikola Borisov, CEO and co-founder of DeepInfra. “This Toronto cluster is a foundational step in expanding our capacity beyond the U.S. and ensuring customers can run AI workloads closer to where their users and data reside.”
🌐 Analysis: DeepInfra’s Toronto deployment illustrates the continued specialization of AI infrastructure around inference as production workloads generate sustained demand for GPU capacity, low latency, and predictable token economics. The company’s expansion also places it within a growing field of inference-focused infrastructure providers and GPU cloud operators investing in owned or dedicated compute capacity rather than relying exclusively on general-purpose public cloud infrastructure.
DeepInfra Purpose-built AI inference cloud operating GPU infrastructure for production-scale deployment of open-source, proprietary, and agent-driven AI models. |
| Overview | DeepInfra is a privately held AI infrastructure company focused on high-throughput inference. The company owns and operates its GPU infrastructure and provides managed access to open-source and proprietary AI models through APIs and dedicated deployments. Its platform is designed for organizations moving AI applications from development into production, where token throughput, latency, GPU utilization, security, and infrastructure availability become operational requirements. |
| Why It Matters | DeepInfra represents the emerging class of specialized inference cloud providers building infrastructure around the sustained compute requirements of production AI. The company operates the underlying GPU capacity rather than functioning only as a model API aggregator, placing it within the expanding market for AI inference infrastructure, GPU clouds, agentic AI deployment, and managed model serving. |
| Founded | 2022 |
| Headquarters | Palo Alto, California |
| CEO / Key Leadership | Nikola Borisov, Co-Founder and CEO; Yessenzhar Kanapin, Co-Founder; Georgios Papoutsis, Co-Founder. |
| Core Technologies | AI Inference Infrastructure GPU Infrastructure Distributed Systems Model Serving Agentic AI Workloads OpenAI-Compatible APIs Dedicated Model Deployments |
| Key Products / Platforms | Managed AI inference cloud; serverless model APIs; OpenAI-compatible APIs; dedicated model deployments; support for open-source and proprietary AI models; GPU infrastructure for production inference; text generation, embeddings, image generation, speech, and other model-serving workloads. |
| Funding / Major Milestone | In May 2026, DeepInfra announced a $107 million Series B to expand its inference cloud and global infrastructure capacity. The round was co-led by 500 Global and Georges Harik, with participation from A.Capital Ventures, Crescent Cove, Felicis, NVIDIA, Peak6, Samsung Next, Supermicro, and Upper90. In July 2026, DeepInfra announced its first international data center location in Toronto, a 1.7 MW facility planned to host more than 1,000 NVIDIA Blackwell GPUs, bringing the company’s infrastructure footprint to nine data center locations. |
| Target Markets | AI Infrastructure AI Inference Agentic AI AI-Native Startups Enterprise AI GPU Cloud Services |
| Editorial Coverage | Converge Digest coverage focuses on DeepInfra’s expansion of production-scale AI inference capacity, GPU infrastructure deployments, data center footprint, funding and infrastructure investment, model-serving architecture, support for agentic AI workloads, and competition among specialized inference clouds, neocloud providers, hyperscalers, and AI infrastructure platforms. |
| Industry Context | As AI deployment shifts from model development and training toward continuous production inference, infrastructure providers are building specialized platforms optimized for token generation, latency, throughput, GPU utilization, and geographically distributed capacity. DeepInfra competes within this developing infrastructure layer alongside specialized inference providers, GPU clouds, neocloud operators, hyperscale cloud platforms, and companies developing vertically integrated AI compute services. |
| Profile Updated | July 2026 |
| Related Knowledge Hubs | DeepInfra • AI Infrastructure • Data Centers |
