India launches urgent $1.25 billion AI mission securing 19,000 GPUs to build six large-scale models by 2025, selecting Sarvam AI for a 70-billion-parameter model optimized for the country's 22 official languages after DeepSeek's breakthrough sparks global AI competition.
Nvidia-backed startup Starcloud makes history by successfully training the first AI model in space using an H100 GPU satellite 100 times more powerful than previous space computers, paving the way for massive orbital data centers that could solve Earth's energy crisis.
OpenAI acquires AI training startup Neptune for under $400 million in stock as the ChatGPT maker prepares for a potential trillion-dollar IPO that could rank among the largest public offerings in history.
Meta secretly develops 'Avocado' AI model to replace Llama while spending $14.3 billion to hire Scale AI founder Alexandr Wang as chief AI officer, abandoning open-source strategy after Llama 4's failure and implementing grueling 70-hour workweeks to compete with OpenAI and Google by 2026.
The US Commerce Department prepares to authorize Nvidia H200 AI chip exports to China, dramatically reversing previous restrictions on advanced processors, though Beijing may reject the offering as it shifts toward domestic AI hardware development.
Moore Threads Technology shares explode 425% after raising $1.13 billion in Shanghai IPO as Chinese AI chip companies surge to fill gaps left by U.S. export restrictions, with Nvidia's CEO warning China's semiconductor industry doubles annually while Western growth lags at just 20-30%.
Poetiq's AI reasoning system shatters records on prestigious AGI benchmarks while slashing operational costs by 50%, as the company releases open-source code enabling researchers worldwide to replicate their breakthrough results using existing models like Gemini and OpenAI.
Johns Hopkins researchers discover that biologically inspired AI systems can mimic human brain activity patterns without any training data, challenging the industry's reliance on massive datasets and suggesting neural architecture design may be more important than extensive learning processes.
Google unveils Gemini 3 Deep Think mode exclusively for Ultra subscribers, delivering groundbreaking AI performance with 41.0% on Humanity's Last Exam and 45.1% on ARC-AGI-2 benchmarks while offering enhanced reasoning for complex mathematical, scientific, and logical problems.
NVIDIA launches DRIVE Alpamayo-R1, the world's first open reasoning AI model for autonomous driving that combines chain-of-thought reasoning with path planning for level 4 autonomy, while also releasing new digital AI tools including speech recognition and content safety models.