Big Tech to pour $650B into AI infrastructure in 2026
What's the story
Four of the biggest US tech companies, Alphabet, Amazon, Meta, and Microsoft, are projected to spend a whopping $650 billion on capital expenditures in 2026, Bloomberg reports. The massive investment is mainly for setting up new data centers and acquiring the necessary equipment such as AI chips, networking cables, and backup generators. The spending spree marks an unprecedented surge in the race for AI tool dominance.
Spending boom
Surge in spending reminiscent of telecom bubble
The projected capital expenditure by these tech giants for 2026 is a staggering 60% increase from last year. This unprecedented surge in spending is reminiscent of the telecom bubble of the 1990s and even the postwar federal investments in interstate highways. The massive outlay has sparked a global data center construction boom, with energy supply concerns and potential price hikes for other users.
Individual projections
Breakdown of projected capex for each tech giant
Meta has announced its full-year capex could reach as high as $135 billion, an almost 87% increase. Microsoft also reported a 66% increase in second-quarter capital spending, exceeding estimates. Analysts expect it will spend nearly $105 billion on capex for the fiscal year ending in June. Alphabet plans to spend up to $185 billion while Amazon has outdone them with a planned $200 billion capital expenditure for 2026.
Investment strategies
Tech giants' strategies to recover investments
Each of the tech giants has a unique strategy to recover their investments. However, they all share the same belief that tools like OpenAI's ChatGPT will become increasingly important in our daily lives. Building these advanced software models is an expensive process, involving thousands of chips costing tens of thousands of dollars each. The hope is that these end products will generate exponentially higher future revenue.
Execution challenges
Challenges and bottlenecks in the data center build-out process
Despite the massive investments, it's still unclear if these companies will be able to realize their ambitious plans. The data center build-out has already led to competition for limited resources such as electricians, cement trucks, and NVIDIA chips. "There are and will be bottlenecks," said Gil Luria, an analyst at DA Davidson.