No market trend can climb in a single direction indefinitely, and the massive spending underway in artificial intelligence is no exception. Speaking at the Goldman Sachs Communications and Technology Conference, Chief Economist Jan Hatzius addressed the surging wave of AI expenditure, cautioning that "this boom will not last forever. There is also a possibility that a substantial portion of AI investment ultimately proves to be low-yield. We cannot rule out this downside risk. Even if the overall outcome of this infrastructure expansion is positive, the pace of investment growth will inevitably slow down in the end."

Projections for AI infrastructure spending continue to climb to striking heights. According to the latest Global Data Center Outlook from PwC, cumulative global investment in AI infrastructure is expected to reach a record $31.6 trillion by 2050. On an annual basis, capital expenditure on data centers is projected to rise from approximately $800 billion in 2026 to $1.8 trillion per year by 2050. Major technology players like Meta, Google, and Microsoft are each pouring billions of dollars into supporting their AI ambitions. PwC notes that because chips and various connected devices require upgrades every few years, the pace of AI infrastructure investment is set to accelerate even further.

Clara Cutajar, PwC Australia's Global Infrastructure Leader, commented: "AI infrastructure has become one of the core challenges of next-generation capital allocation. It spans technology, energy, real estate, supply chains, regulation, and finance, which forces infrastructure investors to reassess the relationship between capital needs, risk, and returns."

Hatzius' baseline scenario is optimistic, assuming that "AI investment is sustainable, can generate real benefits, and can drive significant productivity gains in the future." However, he added: "Even so, new technologies follow their own objective patterns of expansion. Even if most projects are of reasonable quality, there will be a construction phase characterized by a massive surge in investment scale. After that, we enter a harvest phase where the technology is applied across the economy, and investment levels will actually decline during that period. This will bring a host of challenges and will have negative consequences for market participants who assume the AI boom will last indefinitely."