OpenAI to develop custom AI chip, reducing reliance on Nvidia

OpenAI has announced plans to develop its own custom AI chip, reducing reliance on Nvidia. In a policy shift, OpenAI relaxed censorship rules, embracing ‘free expression’ under the new Trump administration. The updated policy says its AI models should “empower people to explore, debate, and create without arbitrary restrictions-no matter how challenging or controversial a topic may be.”

Meta’s $200-billion AI data centre
Meta is reportedly planning a massive AI-focused data centre campus, demonstrating its commitment to long-term AI infrastructure investments that could redefine the digital landscape. The project aligns with Meta’s broader AI ambitions, including the rumoured expansion of Meta AI services and new monetisation models for its chatbot. Reuters reported citing unnamed sources that it plans to start a trial paid subscription service for Meta AI, which will provide users with access to advanced versions of their chatbot. Meta also unveiled Project Waterworth, an AI-driven subsea cable infrastructure to enhance global connectivity. It claimed to improve cable deployment and maintenance, ensuring reliable and efficient internet access worldwide, while also investing in India, one of its key markets. It will bring the advanced subsea cable project to connect India. “Driven by India’s growing demand for digital services, this investment reaffirms Meta’s commitment to economic growth, resilient infrastructure, and digital inclusion, supporting India’s thriving digital landscape and fostering technological innovation,” it said.

Perplexity’s AI-powered web browser ‘Comet’
Perplexity introduced ‘Comet,’ an AI-driven web browser designed to enhance internet browsing through agentic AI capabilities, promising a smarter, more intuitive browsing experience. The browser aims to integrate deep learning models for real-time information retrieval and personalised browsing experiences. “Perplexity will be launching Comet, a new agentic browser, very soon! This is serious engineering undertaking. Please join us to help build the future of internet browsing, with AIs doing deep research and tasks for us!” its co-founder & CEO Aravind Srinivas wrote in a LinkedIn post.

Grok 3 ‘world’s smartest AI’
Elon Musk’s xAI claims Grok 3 surpasses all existing chatbots in intelligence and capability. The latest iteration of the model is touted as the ‘world’s smartest AI.’ Musk’s AI ambitions have fuelled competition, especially with OpenAI and Google, as xAI seeks to carve out a niche in the generative AI space. Musk also claimed that it successfully predicted all the winners at this year’s Oscars, including best picture, best director, and best actress. According to Musk, Grok predicted Anora winning the best picture award, with Sean Baker winning best director and Mikey Madison taking home best actress.

AI geopolitics
South Korea suspended new downloads of Chinese AI app DeepSeek over privacy concerns. Adding to the tensions, OpenAI said it had uncovered evidence that a Chinese security operation had built an artificial intelligence-powered surveillance tool to gather real-time reports about anti-Chinese posts on social media services in western countries. There have been growing concerns that AI can be used for surveillance, hacking and spreading disinformation. Financial Express