/banner-2023/HSBC%20Global%20Private%20Banking-Entrepreneurship-May%202023-How%20video%20games%20tech%20can%20power-up%20your%20business-banner.jpg)
CIO Academy: The World’s Aye to AI: Impact and Opportunities in 2025 & Beyond
Feb 4, 2025
- The global AI race has entered a new competitive landscape. The emergence of a Chinese startup - DeepSeek’s R1 AI model - that was apparently developed on a shoestring budget in comparison to OpenAI’s ChatGPT hit some big chip manufacturers like an unanticipated asteroid, piercing a hole in their sky-high valuations and triggering a huge tech sell-off. This incident, almost overnight, upended a common assumption in the Silicon Valley that having access to advanced chips and building large clusters of GPUs was at the heart of the AI revolution. While questions over the accuracy of DeepSeek’s R1 model have emerged since, and some of the most affected stocks have recovered slightly, the incident has indeed created questions in investors’ mind over the need and viability of tech mega caps’ high capex and multiples. Could US AI spending be optimised and monetised better? How should investors navigate this treacherous investment landscape?
- We think that the emergence of competition is good. It kicks out complacency. If indeed, the cost of AI training and inferencing can go down, more low-cost AI models will emerge, triggering the Javon’s paradox i.e. lower costs accelerating widespread adoption. As such, lower AI costs will democratise its deployment and will make applications for different use cases cheaper. AI’s adoption will grow from AI enablers (Big Tech) to AI adopters (other sectors). This is also a direct positive for software companies and tech platforms that have AI embedded in their P&L. With lower implementation costs and better efficiency, they would be able to monetise their AI models faster, for example, by using AI to target advertisements to customers. Wider AI adoption will reshape corporate operations and boost productivity through time and cost efficiencies delivered by automation, and implementation of smart unified processes - driving real and broader economic impact. This is when winners will emerge from beyond just the tech sector, in other sectors like financials, industrials, and healthcare. As broad productivity gains materialise and transform existing business models at scale, new business models and industries should also emerge. At this stage, AI almost becomes like a fifth factor of production - aiding and enhancing the output of the four conventional factors of production in economics – land, labour, capital, and organisation. Such AI adoption also means continued demand for GPUs
- Furthermore, with America’s AI leadership emerging as one of the key policy priorities, the focus remains firmly on establishing an AI infrastructure, which may now be done much more efficiently. More data centers will be built. They need GPU chips, power, cybersecurity, HVAC cooling systems and a lot more. As such, the need for adequate infrastructure preparedness calls for a fit for purpose energy infrastructure, especially as the US grid remains underinvested and unfit to meet future power needs coming from both reshoring of manufacturing and AI adoption. This will create structural opportunities in investment themes like Digital Infrastructure & Energy Transition. And here, both public and private markets present attractive investment opportunities. Lastly, while 2024 saw the adoption of Generative AI - a small subset of machine learning, 2025 should see a transformative phase of AI ‘agentification’ – which is AI’s evolution from task specific tools like chatbots to specialised interconnected virtual AI agents that are capable of logical reasoning and decision making. Such AI advancements will continue to deliver incremental productivity gains