In the Tech.eu podcast, Fractile founder Walter Goodwin discusses Fractile's AI inference chips which he claims can run LLMs faster and more energy efficient than Nvidia's GPUs.