
18 DEC, 2024
By LIONTRUST

Author: Storm Uru, Head of Liontrust Global Innovation team
AI is no longer a future concept; it is here, and it is reshaping industries at an unprecedented pace. From assisting doctors with diagnostics to enabling autonomous vehicles, AI is integrated into real-world applications that improve efficiency and decision-making. This transformation is comparable to the electrification of industries in the 19th and 20th centuries. Just as electricity revolutionised production, communication, and daily life, AI is set to revolutionise the modern economy. Over the next decade, every pound of GDP will feel the touch of AI, driving a fundamental shift in economic dynamics.
The journey to AI’s current capabilities is rooted in the evolution of computing. For decades, Moore’s Law predicted the doubling of computational power every two years, a trend that persisted for over 50 years. However, this rate of progress could not sustain itself indefinitely. The industry transitioned to accelerated computing, a shift pioneered by Nvidia, which achieved a 100,000x increase in computational performance over a decade.
Accelerated computing is a game-changer, delivering 100x faster processing at 98% lower costs compared to traditional systems. This leap in computational power has unlocked the capabilities of AI, enabling breakthroughs that were unimaginable just a few years ago.
AI’s journey from rudimentary outputs to professional-grade applications highlights the importance of scaling compute power. In 2016, asking an AI model to generate an image of “an elephant in a room” resulted in distorted, unrecognisable outputs. By 2022, improvements were evident, but the results still fell short of professional use. By 2023, however, AI was capable of producing stunningly accurate and professional-quality images within seconds.
This progress was driven by scaling both data and compute power. While early models relied on single GPUs, the current generation leverages clusters of 30,000 or more AI accelerators. Companies are now building clusters with up to a million accelerators, paving the way for even more advanced capabilities. At this scale, AI’s ability to generate content will expand from images to videos and beyond, unlocking new use cases and applications.
Each iteration of AI brings smarter, faster, and more cost-effective systems. The rate of change in AI technology is unprecedented, outpacing advancements in any other field in living memory. Early models, such as OpenAI’s GPT-4, already outperform humans in standardised tests like the SATs and the Bar exam. Yet, they initially lagged in common-sense reasoning − a gap now being closed with next-generation models like OpenAI’s O-1.
Economically, AI’s cost dynamics are just as revolutionary. The expense of inference − the process of generating outputs from AI models − has fallen by 90% in the past year and continues to decrease. This deflationary trend makes AI solutions more accessible and scalable, marking a turning point in their adoption across industries. Sophisticated AI systems are no longer limited to experimental use; they are now viable at scale, ready to transform business operations globally.
One of the most striking aspects of AI adoption is its unparalleled return on investment (ROI). Traditional software implementations, such as Salesforce, often have payback periods of several years. In contrast, AI offers extraordinary returns, with early applications delivering a payback in as little as 1.5 days and ROI of over 2000%. This level of efficiency is fuelling an arms race among companies to adopt AI technologies.
AI-driven disruption is already evident across industries. Startups like Perplexity and Sierra are challenging giants like Google and Salesforce by offering superior products at a fraction of the cost. Perplexity, for example, is redefining search with a streamlined workforce of 160 employees, compared to Google’s 160,000. Similarly, Sierra is disrupting customer relationship management with AI-powered solutions that cost 90% less than traditional offerings. Legal tech startup Harvey exemplifies this trend by reducing contract review times by an astounding 99.97%.
AI is at the forefront of a new innovation cycle, offering exceptional opportunities for growth and disruption. As it becomes smarter, more affordable, and widely adopted, AI will reshape industries, redefine competition, and create new markets Pioneering companies are already demonstrating tangible results. For instance, Meta has leveraged AI-powered content recommendation and ad tools to drive a 23% year-on-year increase in advertising revenue. Similarly, ServiceNow's Now Assist Generative AI assistant has delivered over $10 million in annualised enterprise savings in just 90 days of internal deployment. These examples highlight the substantial economic impact AI can generate for early adopters, presenting a generational opportunity for investors. For companies, the message is clear: adopt and embed AI or risk irrelevance. The greatest investment opportunities today lie predominantly in the AI infrastructure layer of the technology stack, which is essential for building the foundation for broader AI adoption. Over time, the baton will pass to the AI application layer, a transition likely to happen faster than in previous technology cycles due to the immense potential rewards. For instance, Microsoft achieved $3.5 billion in annualised revenues from generative AI within 18 months – much faster than its seven-year cloud transition.
The winners of this AI-driven revolution will not only transform their industries but also set the stage for the next decade of innovation. By focusing on first-mover advantage, scale, and a founder-driven mindset, these companies are positioned to lead the charge in redefining the future. As history has shown, those who embrace and invest in transformative innovation stand to achieve unparalleled success.