AI Breakthroughs & Future: IMO Gold, AGI Race, & Daily Prompts

Step into the dynamic world of AI where innovation never sleeps. This week, we saw Google's Gemini Deep Think achieve a historic IMO Gold, solving complex mathematical problems in natural language, marking a significant shift in AI capability. However, not all ambitious ventures soar smoothly; the $500 billion "Project Stargate" from SoftBank and OpenAI faces significant hurdles, struggling to secure data centre deals, even as OpenAI independently secures a massive deal with Oracle. Meanwhile, the race for Artificial General Intelligence (AGI) intensifies, with analysis suggesting the first nation to achieve AGI could gain a decisive military advantage, akin to nuclear weapons. Aggressive investment is advocated to secure this "fourth offset". On the technical front, Alibaba's Qwen team released an updated Qwen3-235B model with major improvements. Groundbreaking research from Kimi K2 introduced MuonClip, a technique solving training crashes in trillion-parameter models. Prima Mente also unveiled Pleiades, epigenetic foundation models showing promising results in neurodegenerative disease detection, including Alzheimer's and Parkinson's. Beyond the lab, Apple shared technical insights into how their new AI models are trained and optimised, while OpenAI revealed that ChatGPT now handles an astonishing 2.5 billion prompts daily, showcasing its rapid global adoption. Fidji Simo, OpenAI's incoming CEO of Applications, passionately articulated AI's potential as the "greatest source of empowerment," envisioning widespread benefits across society. Looking to foster further growth, Meta and AWS have partnered to launch a program supporting U.S.-based startups building generative AI applications with Llama models, providing essential resources for the next wave of innovation. This week truly highlights the dynamic landscape of AI, from groundbreaking achievements to strategic challenges and inspiring visions for the future.