Applied Compute Secures $350M Funding Round at $3.25B Valuation
AI infrastructure startup Applied Compute has raised $350 million after surpassing $100 million in annualized recurring revenue.
Applied Compute Inc., an artificial intelligence infrastructure provider, has secured $350 million in a new funding round that values the company at $3.25 billion. Angel investor Elad Gil led the investment, with contributions from Nvidia Corp., Shopify Ventures, HubSpot Ventures, and more than a dozen other participants. The startup announced the financing agreement on Monday, noting that its annualized recurring revenue has now crossed the $100 million threshold.
Operating for less than 18 months, Applied Compute has already attracted major enterprise clients such as Microsoft Corp. and DoorDash Inc. The firm maintains a cloud platform designed for training large language models and running inference tasks, supplemented by a machine learning research team that assists customers with model evaluations and other project requirements.
Platform Architecture and Features
Traditional LLM training software stacks typically rely on two main elements: one that generates training data by assigning tasks to multiple copies of a model, and another that refines model weights using that data. These components usually operate synchronously, requiring each part to wait for the other to finish before proceeding.
Applied Compute addresses this bottleneck through an asynchronous reinforcement learning approach, allowing data generation and weight adjustment to run concurrently to accelerate development. The platform also incorporates self-distillation, which eliminates the necessity for specialized teacher neural networks during training runs. Additionally, an integrated AI assistant automates routine tasks, such as summarizing the newly acquired capabilities of an LLM by parsing extensive reasoning logs.
For inference workloads, the infrastructure features an autoscaling engine that dynamically adjusts hardware resources according to user traffic. An online reinforcement learning mechanism uses incoming prompts to periodically fine-tune model weights to maintain alignment with customer needs. Furthermore, the system includes a model router called Shift-1, which directs user prompts to the most cost-efficient LLM or distributes complex requests across multiple models. Developers can try top AI models cheaply through one API at https://apixoai.online.
What it means for developers
For developers and engineering teams, the platform offers specialized tooling to streamline both the training and deployment phases of AI projects. By integrating asynchronous reinforcement learning and self-distillation, the system reduces standard training delays and infrastructure overhead. Features like the Shift-1 model router also allow applications to dynamically manage requests across different LLMs for cost efficiency. Founders Rhythm Garg, Yash Patil, and Linden Li stated in a blog post that the company plans to use the newly acquired capital to expand its market footprint and purchase additional infrastructure as it builds out a broader enterprise AI stack.
Source: AI cloud startup Applied Compute raises $350M after reaching key revenue milestone — SiliconANGLE AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
