
Mirendil, an AI lab specializing in self-improving AI, has entered into a multiyear partnership with Google Cloud, valued at over $100 million. This deal grants Mirendil access to Google's TPUs and Nvidia GPUs, crucial for their compute-intensive AI research. The partnership reflects a broader industry trend where cloud providers are increasingly supporting AI startups with significant infrastructure commitments. Mirendil aims to use this collaboration to advance its AI technology, which could automate scientific research and improve various scientific fields.
Read originalAlibaba is testing a new business model for its upcoming Qwen open-weight AI model, introducing revenue-sharing terms for commercial users. This move targets larger companies that profit from offering the model as a service, requiring them to enter a commercial agreement with Alibaba. The approach mirrors the licensing model used by Moonshot for its Kimi K3 model, which includes revenue-sharing for companies exceeding certain revenue thresholds. This shift signifies Alibaba's strategy to monetize its open-weight models while maintaining their open-source nature, potentially setting a precedent for other AI developers.
© GitHub ChangelogGitHub has rolled out a new feature for enterprise users, allowing them to manage which Model Context Protocol (MCP) servers their Copilot clients can access. This update empowers enterprise owners to approve trusted MCP servers and block those that are untrusted or non-compliant, enhancing security and compliance across their organizations. The feature is integrated into the enterprise managed settings, enabling teams to customize their own allow and deny lists. This development marks a significant step in giving enterprises more control over their AI tool usage, ensuring that only vetted servers are utilized.