OpenAI aggressively expands its AI capabilities, aiming for 1 million GPUs by year-end and an astonishing 100 million GPUs worth $3 trillion, fueling concerns about AI supremacy as its Texas data center becomes the world's largest at 1 gigawatt by 2026.
DeepSeek V3, OLMo 2, and Gemma 3 unleash groundbreaking transformer models, harnessing techniques like Multi-Head Latent Attention, QK-norm, and sliding window attention for improved efficiency, stability, and reduced memory footprint.
Meta unveils a groundbreaking AI framework, MTTT, that enables models to learn how to learn at test time, removing the need for human-designed self-supervised tasks and outperforming linear attention on long sequences while maintaining linear complexity.
OpenAI researchers Jason Wei and Hyung Won Chung depart for Meta's superintelligence lab, amid an AI talent war where Meta offers up to $300 million over four years to top AI talent.
Meta intensifies the race for artificial general intelligence by poaching two prominent researchers, Jason Wei and Hyung Won Chung, from OpenAI, offering lucrative compensation to bolster its AI capabilities under CEO Mark Zuckerberg's leadership.
NVIDIA introduces Dynamo, an open-source framework optimizing performance and scalability for large language models and generative AI applications on the cloud, featuring innovations like disaggregated prefill and decode phases, dynamic GPU management, efficient caching, and accelerated data transfer, showcasing deployment with DeepSeek-R1-Distill-8b on Amazon EKS.
AI startup Thinking Machines Lab, founded by former OpenAI executive Mira Murati, secures $2 billion at $12 billion valuation in funding round led by Andreessen Horowitz, aiming to build safer, more reliable AI systems for broader applications than rivals.
In a potential industry-shaking move, Meta, the tech giant behind Facebook, weighs developing closed AI models, risking backlash from open source advocates as it seeks to monetize AI beyond advertising and retain top talent with high salaries.
Meta CEO Mark Zuckerberg announces hundreds of billions in AI infrastructure investment, launching Superintelligence Labs and massive hiring spree while planning Prometheus supercluster for 2026 alongside additional multi-gigawatt facilities including Hyperion, plus a $14 billion Scale AI investment.
Moonshot AI unveils Kimi K2, a groundbreaking 1 trillion parameter open-source language model that surpasses GPT-4 in coding, autonomous tasks, and math reasoning while being offered for free, potentially disrupting big tech's profit centers with its cost-efficient MuonClip optimizer.