How Microsoft’s next-gen BitNet architecture is turbocharging LLM efficiency AI Least News How Microsoft’s next-gen BitNet architecture is turbocharging LLM efficiency