Advanced Reasoning Models Like DeepSeek-R1 Are Driving Growth in GPU Demand, Says Together AI

Prevailing perceptions in the tech realm have been radically upended by Together AI. The conventional fear that advanced reasoning models would necessitate less infrastructure has been blown out of the water. As it turns out, these sophisticated models, such as the trailblazing DeepSeek-R1, are demanding an increase rather than a decrease in infrastructure. This unanticipated consequence has been a boon for Together AI, which has seen substantial business expansion fueled by the escalating need for more robust infrastructure. Consistently staying ahead of the curve, Together AI has recently added a sizeable $305 million to its war chest in a Series B round led by General Catalyst and Prosperity7. Ever since its conception in 2023, Together AI’s crusade has been to simplify enterprise usage of open-source large language models (LLMs). The company’s trajectory took an upward swing in 2024 with its feature-rich enterprise platform that enables the efficient deployment of AI in various environments, including virtual private cloud (VPC) and on-premises environments. Things didn’t slow down in 2025. Instead, the company further evolved its platform by illuminating innovations in reasoning clusters and agentic AI capabilities. Together AI has impressively captivated the allegiance of over 450,000 registered developers and corporate clout counting Krea AI, Captions, and Pika Labs among its clientele. CEO Vipul Prakash highlights just how widely DeepSeek-R1 has impacted AI infrastructure demand. The company’s growth has not just survived, but thrived, in the face of the disruptive entrance of DeepSeek-R1 into the market. While there was an assumption that such an advanced, open-source reasoning model could necessitate less infrastructure, the reality proved far different. The company has amplified its infrastructure to support the skyrocketing demand for workloads related to DeepSeek-R1.