Spanish Startup Raises $290M to Shrink AI Models, Cuts Energy Use by 84% While Challenging Big Tech's Power-Hungry Approach

Spanish startup Multiverse Computing raises $290M to revolutionize AI with small language models that slash energy consumption by 84% while running on regular CPUs instead of power-hungry GPUs, challenging Big Tech's wasteful approach as data centers devour increasing amounts of electricity.

• Multiverse Computing develops small AI models (SLMs) that compress large language models to run on regular CPUs instead of expensive GPUs, reducing energy consumption by 84% with only 2-3% accuracy loss
• The Spanish startup has raised $290 million for a $500+ million valuation and serves clients like Bosch, Moody's, and Bank of Canada with applications ranging from military drones to smart appliances
• Founders argue the AI industry's focus on massive models is fundamentally wrong given environmental costs, as US data centers now consume 4.4% of electricity and will more than double consumption by 2030