OpenRouter, a New York-based AI (Artificial Intelligence) model exchange, announced on May 26, 2026 that it had closed a $113 million Series B funding round. <cite index="1-1">The round was led by Alphabet's independent growth fund CapitalG, with participation from NVentures (NVIDIA's venture capital arm), ServiceNow Ventures, MongoDB Ventures, Snowflake Ventures, Databricks Ventures, alongside existing investors Andreessen Horowitz and Menlo Ventures.</cite>
While the company did not officially disclose the valuation, <cite index="3-3,3-4">The New York Times reported that the round closed at approximately $1.3 billion post-money, more than double the estimated $547 million post-money valuation it reached a year earlier, per PitchBook, after raising $40 million in Series A funding in June 2025.</cite>
Business model and scale
OpenRouter operates as a unified API (Application Programming Interface) gateway that allows developers to access and route requests across hundreds of LLMs (Large Language Models) from multiple providers. <cite index="8-5,8-6">The platform manages and optimizes inference across more than 400 models from AI providers including Anthropic, Google, OpenAI, xAI, and DeepSeek, serving over 8 million global users that include AI-native startups and large enterprises.</cite>
Usage growth has accelerated sharply. <cite index="1-2,1-3">OpenRouter's volume has surged to 25 trillion tokens per week (100 trillion tokens per month), representing a 5X increase from the 5 trillion tokens processed per week just six months ago, illustrating how quickly enterprises are deploying agents and scaling AI across multiple models and providers.</cite>
<cite index="6-5,6-6,6-7">Founded in 2023 and led by CEO Alex Atallah, OpenRouter enables developers and enterprises to access, route, and optimize across hundreds of AI models through a single API, standardizing access to providers through a single interface, with controls like per-request data handling policies, team-level access and routing permissions, spend visibility, and audit-friendly usage reporting.</cite>
Investor thesis
CapitalG framed the investment as a bet on inference infrastructure rather than model development. <cite index="1-8,1-9,1-10">"OpenRouter is solving the infrastructure gap for inference in the AI era," said Mo Jomaa, partner at CapitalG, while partner Jane Alexander added that as companies shift toward a multi-model paradigm, OpenRouter is positioned to become "the data clearinghouse and unified intelligence layer for AI models."</cite>
The company cited enterprise demand as a driver. <cite index="1-17,1-18,1-19">OpenRouter said the new capital will expand its routing, governance, and optimization capabilities as enterprises increasingly deploy AI into production, citing a 2026 Deloitte study that found 67% of enterprises are already consuming over one billion tokens per month, and noting a shift toward multi-model strategies where companies route across models and providers to optimize for cost, latency, and capability.</cite>
CEO Alex Atallah framed routing as a structural requirement of large-scale deployment. <cite index="1-20,1-21,1-22,1-23">"Running inference at scale is fundamentally a multi-model problem. The era of picking a single model is over," Atallah said, adding that success now depends on continuously routing across a changing market and that because OpenRouter sits in the flow of production traffic, it can optimize every request for cost, performance, and reliability in real time.</cite>
Market context
The round reflects a pattern of capital flowing toward orchestration and infrastructure providers situated between application developers and foundation model labs. Participation from the venture arms of Databricks, Snowflake, MongoDB, and ServiceNow — all enterprise data platforms — alongside NVIDIA's NVentures suggests strategic interest in standardizing the inference layer. <cite index="9-9">CapitalG partner Mo Jomaa compared OpenRouter's role in AI infrastructure to foundational platform companies such as Cloudflare, Stripe, and Databricks.</cite>