Key takeaways:

• We are in the beginning stages of generative artificial intelligence (GenAI): In the coming years, the application of constantly evolving GenAI models will bring revolutionary change to enable innovation, accelerate growth and boost the efficiency of the global economy.
• Investors underestimate the transformative nature of this technology and the opportunity it represents: Most investors do not yet fully understand GenAI and are significantly underestimating the value we believe it will deliver over the next decade. 
• Growth of GenAI will drive demand across inputs to scale: Large language models (LLMs) will drive demand for compute (raw processing power), electrical power, data and algorithmic enhancements.

Introduction

Despite some volatility in recent weeks, technology stocks have delivered strong returns year to date through July 31, 2024, outperforming all other sectors—often by a large margin.1 In our view, this outperformance represents a continuation of the trend that followed the launch of ChatGPT 3.5,2 the first widely adopted transformer-based artificial intelligence (AI) model, in November 2022. Despite witnessing the phenomenal growth of AI infrastructure businesses during this period, we are convinced that many investors do not fully comprehend the improvements in AI models that are yet to come. They also do not foresee how these improvements, once employed, will enable innovation, accelerate growth and boost the efficiency of the global economy in the years ahead.

As we explain in this paper, we believe we are at the beginning of a profound, revolutionary change. Recent capability improvements in transformer-based models are impressive;3 the world’s most advanced AI models are now able to complete complex coding tasks, ace the bar exam, outperform most humans on standardized tests, create videos and develop complex reasoning flows. And as more compute (raw processing power), data and algorithmic enhancements are applied to planned transformer models over the next few years, they will certainly continue to improve.

Read more here: Following the curves: AI trendlines and the case for long-term growth | Franklin Templeton