JPMorgan forecasts $5.5T AI capex growth through 2030, signaling infrastructure supercycle

JPMorgan forecasts $5.5T AI capex growth through 2030, signaling infrastructure supercycle

The bank's midyear outlook projects hyperscalers will spend $342 billion in 2025 alone, with ripple effects reaching crypto miners and semiconductor supply chains

JPMorgan Global Research just put a number on the AI buildout, and it’s staggering. The firm’s midyear 2026 outlook projects global AI and data center capital expenditure will reach at least $5 trillion by 2030, with a potential ceiling of $7 trillion.

The numbers behind the buildout

Hyperscalers, think Microsoft, Amazon, and Alphabet, are forecast to spend $342 billion in capital expenditure in 2025. That represents a 62% year-over-year increase. Cumulatively, hyperscaler spending is expected to reach between $650 billion and $697 billion through 2026.

On the physical infrastructure side, 122 gigawatts of new data center capacity is planned for development between 2026 and 2030. Data center construction has already hit a $40 billion annualized run rate as of mid-2025, up 30% year-over-year.

Advertisement

JPMorgan estimates roughly $150 billion in leveraged finance tied to AI and data centers over the next five years. Investment-grade bonds linked to these sectors could reach $1.5 trillion in the same timeframe.

Where the money flows

The US commands about 85% of AI and machine learning venture capital, according to JPMorgan’s analysis. South Korea, Taiwan, and China are expected to capture downstream demand given their critical roles in semiconductor supply chains.

The bank also frames AI as a long-term disinflationary force in the economy, suggesting AI could eventually reduce costs across industries fast enough to put downward pressure on prices.

The crypto angle: miners pivot to AI

JPMorgan’s research highlights a potential pivot by Bitcoin miners toward hosting AI and high-performance computing operations. Miners already operate facilities with massive power capacity, cooling infrastructure, and rack space. AI compute contracts tend to offer more predictable revenue streams than Bitcoin mining, where margins fluctuate with hash rate difficulty and BTC price.

The 122 GW of planned data center capacity represents an enormous appetite for electricity. Power constraints are already one of the primary risks JPMorgan flags in the outlook. In regions where Bitcoin mining operations sit on favorable power purchase agreements, those contracts become strategic assets worth more than their original mining economics justified.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.

JPMorgan forecasts $5.5T AI capex growth through 2030, signaling infrastructure supercycle

JPMorgan forecasts $5.5T AI capex growth through 2030, signaling infrastructure supercycle

The bank's midyear outlook projects hyperscalers will spend $342 billion in 2025 alone, with ripple effects reaching crypto miners and semiconductor supply chains

JPMorgan Global Research just put a number on the AI buildout, and it’s staggering. The firm’s midyear 2026 outlook projects global AI and data center capital expenditure will reach at least $5 trillion by 2030, with a potential ceiling of $7 trillion.

The numbers behind the buildout

Hyperscalers, think Microsoft, Amazon, and Alphabet, are forecast to spend $342 billion in capital expenditure in 2025. That represents a 62% year-over-year increase. Cumulatively, hyperscaler spending is expected to reach between $650 billion and $697 billion through 2026.

On the physical infrastructure side, 122 gigawatts of new data center capacity is planned for development between 2026 and 2030. Data center construction has already hit a $40 billion annualized run rate as of mid-2025, up 30% year-over-year.

Advertisement

JPMorgan estimates roughly $150 billion in leveraged finance tied to AI and data centers over the next five years. Investment-grade bonds linked to these sectors could reach $1.5 trillion in the same timeframe.

Where the money flows

The US commands about 85% of AI and machine learning venture capital, according to JPMorgan’s analysis. South Korea, Taiwan, and China are expected to capture downstream demand given their critical roles in semiconductor supply chains.

The bank also frames AI as a long-term disinflationary force in the economy, suggesting AI could eventually reduce costs across industries fast enough to put downward pressure on prices.

The crypto angle: miners pivot to AI

JPMorgan’s research highlights a potential pivot by Bitcoin miners toward hosting AI and high-performance computing operations. Miners already operate facilities with massive power capacity, cooling infrastructure, and rack space. AI compute contracts tend to offer more predictable revenue streams than Bitcoin mining, where margins fluctuate with hash rate difficulty and BTC price.

The 122 GW of planned data center capacity represents an enormous appetite for electricity. Power constraints are already one of the primary risks JPMorgan flags in the outlook. In regions where Bitcoin mining operations sit on favorable power purchase agreements, those contracts become strategic assets worth more than their original mining economics justified.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.