AI infrastructure emerges as a new macro cycle

Artificial intelligence (AI) is no longer simply a technology theme. It is increasingly becoming a macroeconomic force that is reshaping global investment, inflation dynamics and capital markets. The scale of capital being committed to AI infrastructure is unprecedented, creating a reinforcing cycle of semiconductor demand, cloud investment and energy consumption.

While this investment boom is contributing to near-term inflationary pressures and periods of market volatility, this insight explores how it is also laying the foundations for longer-term productivity gains that could influence global economic growth.

• AI infrastructure investment is becoming a meaningful driver of global capital expenditure
• Rising demand for semiconductors, data centres and electricity is beginning to influence inflation dynamics
• Productivity gains may ultimately offset some of these inflationary pressures over the longer term
• South Korea has emerged as a significant beneficiary due to its role in advanced semiconductor manufacturing

The magnitude of investment illustrates the scale of the opportunity. During 2025, leading AI companies, including Amazon, Anthropic, Google, Meta, Microsoft, OpenAI and Oracle, committed an estimated US$300 billion to AI infrastructure spanning semiconductors, data centres, power infrastructure and specialised labour.

Looking ahead, LSEG data indicates that the five largest US hyperscalers are projected to spend approximately US$720 billion on capital expenditure during 2026, supported by rapidly expanding AI-related revenues. Microsoft reported an annualised AI revenue run rate of US$37 billion, representing 123% year-on-year growth, while Amazon Web Services (AWS) recorded its fastest revenue growth in more than three years as AI-related workloads accelerated.

This investment wave is beginning to influence inflation across multiple sectors. Prices for computer software and accessories have risen by almost 14% over the past 12 months, reversing decades of persistent price declines. Wholesale prices for electronic components increased 28% over the same period.

The demand shock is also evident in global trade. US imports of computers more than doubled during the first quarter of 2026 to approximately US$93 billion, while semiconductor imports increased 40% and computer accessory imports rose 37% year on year. Rather than reflecting temporary supply disruption, these trends suggest that AI infrastructure has become an increasingly important source of global demand for advanced technology products.

Energy is emerging as another key constraint. Following more than a decade of relatively stagnant electricity demand, US electricity production increased 2.5% in 2024, 2.4% in 2025, and 3.0% year-on-year in March 2026, reflecting the rapid expansion of AI data centres.

Consumer electricity prices increased 4.6% year-on-year over the same period, highlighting how AI-related investment may be extending inflationary pressures beyond semiconductors into utilities and broader infrastructure. AI therefore represents not only a digital transformation but also a physical investment cycle requiring substantial expansion of electricity generation and transmission capacity.

Despite these inflationary effects, AI also offers the prospect of becoming structurally disinflationary over the longer term. Greater automation, improved supply chain optimisation, enhanced decision-making and productivity improvements could allow economies to generate stronger output growth without proportionate increases in labour or capital costs.

The transition, however, is unlikely to be immediate. Large-scale spending on semiconductors, data centres and energy infrastructure may continue to support upward pressure on selected prices before productivity gains become sufficiently widespread to offset these effects. The relationship between AI-related investment and inflation may therefore be sequential rather than simultaneous: inflationary during the investment phase and potentially disinflationary during broader adoption.

This dynamic presents a complex challenge for central banks. Policymakers must balance inflation generated by exceptionally strong capital expenditure against the possibility that AI ultimately raises an economy's potential growth rate.

As a result, interest rate decisions are likely to remain sensitive to incoming inflation data, even as policymakers increasingly recognise the structural nature of AI-related investment.

The impact on equity markets has been substantial. Prior to its recent correction, the Korean Composite Stock Price Index (KOSPI) was among the strongest-performing major equity indices globally, supported largely by gains in Samsung Electronics and SK Hynix as AI demand accelerated.

At one stage, however, the index experienced a correction of more than 8%, illustrating how concentrated investor positioning in AI-related technology stocks has increased market sensitivity to shifts in global sentiment. Rather than indicating weaker corporate fundamentals, the decline was widely attributed to profit-taking following a strong rally and uncertainty surrounding US monetary policy.

Industry fundamentals remain robust. Demand from hyperscale cloud providers, AI developers and enterprise customers remains strong and continues to outpace available semiconductor supply, allowing manufacturers of advanced memory products to maintain significant pricing power.

This supply-demand imbalance has become one of the defining features of the current semiconductor cycle and differentiates it from previous technology upcycles that were driven primarily by consumer electronics demand.

South Korea's macroeconomic backdrop nevertheless highlights some of the risks accompanying the AI investment cycle.

The Bank of Korea has reduced its policy rate to 2.50% in an effort to support domestic demand while remaining cautious about additional easing given persistent inflation risks. The Korean won has weakened towards multi-year lows as higher US interest rates, foreign investor repositioning and periods of stronger US dollar demand have weighed on the currency.

Taken together, these developments suggest that AI is evolving into a global capital expenditure supercycle comparable in scale to previous periods of industrial transformation. The investment theme now extends far beyond software, encompassing semiconductor manufacturing, cloud infrastructure, electricity generation, advanced manufacturing and global supply chains.

Although elevated valuations, tighter monetary policy and geopolitical uncertainty are likely to generate periods of volatility, the structural drivers remain intact. Demand for computing power continues to outpace supply, corporate investment plans continue to expand, and countries with strategic positions within the semiconductor ecosystem, particularly South Korea, appear well positioned to benefit from this evolving investment cycle.