4 ms·

Building and running LLMs is an energy intensive task. Those GPUs do not power themselves from nothing.