Artificial Intelligence Sets New Accents – From Gigantic Data Centers to Smart Edge AI

The rapid progress in generative AI has triggered an investment boom benefiting both startups and large tech companies. Cloud-based AI services, particularly models from OpenAI, are at the center of attention. Ambitions to build massive data centers are growing, as the need for stronger infrastructure increases. Leading companies could invest around $160 billion next year alone to acquire GPUs and other necessary equipment for AI training. Executives even predict that investments in data centers worldwide could rise to as much as $2 trillion in the coming years. However, the question remains whether the revenues from AI applications can truly offset the high investment costs. Parallel to this, a promising trend is emerging: Edge AI. By executing AI algorithms directly on end-user devices like smartphones and computers, Edge AI promises real-time responses and improved privacy, as data remains local on the device. According to analysts, by 2027, nearly 50% of all smartphones could possess generative AI capabilities—a significant increase from the current 4%. However, many current devices lack the computing power to run extensive models like GPT-4. Nonetheless, specialized, smaller AI models are gaining importance. These require less training data and are often available as open-source solutions, facilitating their implementation on consumer devices. Processor manufacturers are driving innovation by increasing the computing and storage capacities of smartphones and PCs. These developments could enable future devices to support larger AI models. For investors, Edge AI offers new opportunities, as the demand from users for more powerful devices to leverage new AI applications could drive the electronics market. Analysts expect that the combined revenue from smartphones and PCs could exceed $700 billion by 2027. Ultimately, the success of Edge AI depends on the development of compelling applications that provide users with sufficient added value.