Groq, the AI chip startup, is seeking to raise $650 million in fresh capital from its existing investors as it pushes deeper into the inference neocloud market, according to sources cited by Axios. The company has developed its own AI chips and systems that allow developers and enterprises to host applications requiring intensive inference processing—the computational work that occurs after an AI model receives a prompt. Inference is currently in higher demand across the AI industry compared to model training, making this a strategically important area for growth. The funding request comes just months after Groq's landmark $20 billion agreement with Nvidia in December. That deal, characterized as a "not-acqui-hire," involved several top-level Groq executives departing for the chip giant along with Nvidia licensing Groq's proprietary hardware technology. If the arrangement had been structured as a full acquisition, it would have represented Nvidia's largest purchase ever. The agreement provided a favorable exit for Groq's investors, who received cash payouts from the transaction. Groq is now asking those same investors to support its expansion plans by participating in the new $650 million round. According to Axios, the company's major backers Disruptive and Infinitium have committed to covering any portions of the round that other existing investors choose not to fill with their pro-rata shares, effectively guaranteeing the raise. The company's current operations are being led by interim CEO Adam Winter and interim CFO Matt Eng. Groq's shift toward building out its inference cloud infrastructure represents a strategic pivot, positioning the startup to capitalize on the growing need for specialized computing resources to run AI applications in production environments rather than just training new models.