Introduction
Citigroup raised its AI capex forecast for hyperscalers’ AI capital spending to about $490 billion by the end of 2026, up from a previous $420B outlook, and now projects cumulative AI infrastructure spending above $2.8 trillion by 2029. This surge will meaningfully lift demand for GPUs, accelerators and data-center builds.
[related url=”https://tekznology.com/iphone-fold-inches-closer-samsung-foldable-oled/”]
Why this matters:
- Massive capex by a handful of hyperscalers will shape the semiconductor and cloud markets, driving demand for advanced chips and power infrastructure.
- The forecast raises investor, policy and energy-use questions — from chip shortages to grid stress and regional data-center siting.
The mechanics behind the number
Citigroup’s model factors in compute-intensive LLM training and inference demand; it estimates AI compute needs that will require tens of gigawatts of extra electricity and billions in server and chip investments through 2029. The bank points to early aggressive investment by major cloud providers and hardware suppliers.
[related url=”https://tekznology.com/record-breaking-great-white-spotted-near-popular-us/”]
Market and policy implications
- Chips & supply: Strong orders for GPUs and custom accelerators imply sustained semiconductor demand and possible shortfalls unless capacity expands.
- Energy & location: Data-center growth raises power and cooling needs; countries with supportive grid and tax incentives will attract more builds.
Author note: I use Citigroup and Reuters reporting to explain how AI capex affects chips, data centers, and geopolitical tech races. Watch spending flows — they shape product roadmaps for smartphones and cloud services alike.
[related url=”https://tekznology.com/microsoft-adds-anthropic-models-to-copilot-studio/”]