人工智能：下一個計算平台的未來 (只提供英文版)

In the evolution of AI, where are we today?

Matt Cioppa: In 2024, we have primarily been in the buildout phase. In many ways, it is reminiscent of the early era of early cloud computing, when major technology businesses invested heavily in cloud infrastructure, initially pressuring margins, but paving the way for substantial profitability. We could see a similar trajectory for AI, with “Big Tech” investing over US$200 billion this year alone.¹

AI is a transformational shift in computing platforms, akin to the internet or mobile revolutions. We believe the significant capital expenditure in foundational technology reflects hyperscalers’ confidence in the long-term demand for AI solutions and the potential for new business models. While only a handful of tech giants can afford massive foundational investments, smaller firms can also thrive by building successful businesses on top of new infrastructure, much like Salesforce and ServiceNow did during the cloud computing boom. As AI infrastructure evolves, we will continue to edge from buildout into applications.

What are you seeing as AI applications begin to develop? 

Matt Cioppa: Demand for generative AI applications is accelerating, with companies exploring use cases in areas such as customer support, code generation and marketing. Businesses are already reporting tangible benefits; for example, JP Morgan expects to save around US$1.5 billion² through AI-enhanced customer interactions and fraud prevention. We think these initial savings are just the beginning, with AI adoption expected to drive productivity gains that could create trillions in value.

However, investors should be prepared to be patient, as major tech shifts take time. For instance, it took seven years after the iPhone launch for smartphones to achieve widespread adoption. We believe AI has similar potential, with the launch of ChatGPT considered AI’s “iPhone moment.” Generative AI could reshape existing industries and enable entirely new ones, much like how Uber, Airbnb, and Netflix emerged in the mobile era.

AI is incredibly energy intensive. How will we meet this demand? 

Matt Cioppa: As generative AI adoption accelerates, the energy needed to support its growth becomes a critical consideration. While we don’t anticipate immediate power constraints, long-term energy requirements will call for innovative solutions. In some cases, this may involve repurposing existing technologies, such as nuclear power. Leading technology companies, including Microsoft, Amazon and Google, are already partnering with or acquiring nuclear facilities to secure stable energy supplies.

Beyond nuclear, a diverse energy mix is essential. Recent deals, such as Microsoft’s record power purchase agreement (PPA) with Brookfield Infrastructure, highlight the trend toward large-scale clean energy projects. We expect more agreements like this, combining advanced infrastructure upgrades with expanded clean energy investments to meet AI’s growing power needs effectively. We could also see both chip and model architecture enhancements that increase efficiency and reduce power demands.

What do you anticipate for tech post-US election? 

Matt Cioppa: We expect a release of pent-up technology spending. Donald Trump’s victory could bring lighter regulation, favouring large technology firms and potentially accelerating mergers and acquisitions among smaller companies. Additionally, if Trump, as promised, reverses the Biden administration’s executive order on AI safety, it could accelerate AI innovation as the US competes to maintain its technological edge globally.

Do you think current technology sector valuations are warranted?

Matt Cioppa: As of October 31, 2024, the IT sector traded at a 33% premium to the S&P 500 Index, well above its historical 15% average.³ However, excluding mega-cap companies like Microsoft, Apple, and Nvidia, the premium drops to just 7%,⁴ suggesting attractive opportunities in smaller and mid-sized technology stocks.

As AI adoption progresses, we expect a broader range of companies to benefit, shifting investor focus from a few dominant players to a wider array of technology businesses poised to capitalise on AI. On a growth adjusted basis, technology sector valuations are in line with the five-year average, which we think reflects the sector’s growth potential.⁵

What is your outlook for AI in 2025? 

Matt Cioppa: We believe that the technology sector, underpinned by the transformative power of AI, offers a compelling investment case. As AI evolves and drives productivity gains, we expect new industries and market leaders to emerge, in a similar way to past technological shifts. For investors, this is a rare opportunity to participate in the early stages of a new growth cycle, where patience could be rewarded as the full impact of AI unfolds.