Bank of America: AI Data Center Systems Could Reach $1.7 Trillion by 2030

This progression reflects a growing consensus that AI infrastructure spending will be both larger and longer‑lasting than many early forecasts assumed.

What’s Driving the Upgrade: Hyperscaler Capex

The biggest driver behind the revised outlook is a surge in capital spending by the world’s largest cloud platforms.

Microsoft, Amazon, Alphabet (Google), and Meta are investing heavily in data centers, GPUs, networking hardware, and power infrastructure to support AI workloads. Combined, these hyperscalers are expected to spend roughly $700 billion to $715 billion on capital expenditures in 2026, a massive jump from about $410 billion in 2025.

Much of that spending is directed toward:

• AI training clusters and GPU farms
• Data‑center construction and expansion
• High‑speed networking and memory systems
• Power generation and cooling systems

Because these companies dominate global cloud computing, their investment cycles effectively determine the pace of the entire AI infrastructure market.

Semiconductor Demand at the Center of the Boom

AI data centers require vast quantities of specialized chips, making semiconductors one of the primary beneficiaries of the infrastructure surge.

Within the projected $1.7 trillion market, a substantial share is expected to come from AI accelerators—GPUs and custom chips used for training and inference workloads.

This demand is helping drive strong outlooks for chipmakers and suppliers across the AI stack. Analysts have highlighted companies such as:

• Nvidia
• AMD
• Broadcom
• Micron
• Marvell

These firms supply key components including GPUs, networking chips, high‑bandwidth memory, and custom AI processors required for hyperscale data centers.

Power and Grid Infrastructure: The Hidden Constraint

While chips and servers dominate headlines, BofA and other analysts argue that the real bottleneck for AI expansion may be electricity and grid capacity.

AI data centers consume enormous amounts of power, and the rapid increase in deployments is stressing energy systems and physical infrastructure. Bank of America has warned that expanding AI facilities will place growing pressure on:

• electrical grids
• water resources
• critical metals used in electronics and power equipment.

The scale is striking: projections suggest that global data‑center electricity consumption could exceed Japan’s total national power use by 2030 if current growth trends continue.

Because of these constraints, energy generation, grid equipment, cooling systems, and power‑distribution infrastructure are becoming key “picks‑and‑shovels” industries in the AI era.

The AI Infrastructure Investment Cycle

Taken together, these trends form a reinforcing cycle:

1. Hyperscalers increase capital spending to compete in AI.
2. That spending drives massive demand for semiconductors and computing hardware.
3. Data‑center expansion increases power demand and strains energy infrastructure.
4. New investment flows into electricity generation, cooling, and grid capacity.

Bank of America’s upgraded forecast reflects the view that this cycle is still in its early stages, with several years of large‑scale infrastructure buildout ahead.

What the $1.7 Trillion Forecast Really Means

The revised projection highlights how quickly the economics of AI infrastructure are expanding. A market approaching $1.7 trillion by 2030 implies one of the largest technology capital‑expenditure cycles in modern history.

But the forecast also comes with uncertainty. The pace of expansion will depend on factors including semiconductor supply, hyperscaler investment discipline, and whether energy and grid infrastructure can keep up with AI’s growing electricity demands.

What is increasingly clear is that the AI boom is no longer just about software models—it is fundamentally reshaping the global buildout of data centers, chips, and energy systems that power them.