Forward-looking: Fusion energy is often regarded as the holy grail of power generation because it harnesses the same atomic process that powers the sun. The concept involves forcing atomic nuclei to fuse together, unleashing immense energy. If scientists can crack the code using lasers, it could lead to a virtually limitless supply of safe, sustainable energy without any carbon emissions.
In context: Demand for GPUs and other AI accelerators isn't expected to decrease anytime soon, and the cost of DRAM will be directly influenced by the strong "generative bubble" we are experiencing right now. Market intelligence company TrendForce has made some estimations for the third quarter of the year, and, unsurprisingly, prices are going up.
Cutting corners: Researchers from the University of California, Santa Cruz, have devised a way to run a billion-parameter-scale large language model using just 13 watts of power – about as much as a modern LED light bulb. For comparison, a data center-grade GPU used for LLM tasks requires around 700 watts.