By Deborah Mary Sophia and Greg Bensinger

Feb 5 (Reuters) – Amazon.com on Thursday projected a jump of more than 50% in capital expenditures this year, joining its Big Tech peers in adding new expenses ​as they race to build out artificial-intelligence infrastructure.

It is the latest sign that tech companies will not ‌be hitting the brakes on their hefty AI investments anytime soon. The market reacted sharply, sending shares of Amazon down as much as 11% ‌after-hours before settling at around 7% down.

The company said it expects to invest about $200 billion in capital expenditures across Amazon in 2026, compared with about $131 billion in 2025.

Amazon’s forecast for first-quarter operating income of $16.5 billion to $21.5 billion fell below analysts’ estimate of $22.04 billion.

This guidance includes roughly $1 billion in higher costs related to its high-speed internet business Leo, as well as investment in ⁠quick commerce and sharper prices in its ‌international stores business.

Big Tech is spending enormous amounts of money on processors, data centers and networking equipment as the companies rush to build out their AI infrastructure. The top four hyperscalers – ‍Amazon, Microsoft, Alphabet’s Google and Meta – are expected to collectively spend more than $500 billion this year.

But tech earnings over the past few days have shown that Wall Street has a clear message for tech companies: Soaring AI spending can continue only if companies show ​commensurate operational or financial returns.

Google’s eye-popping capex forecast of $175 billion to $185 billion for the year got a pass from ‌investors on Wednesday as the company delivered stellar growth in its cloud revenue, as did Meta’s plan to spend between $115 billion and $135 billion.

But investors punished Microsoft’s stock last week after its cloud unit growth just squeaked past estimates.

For Amazon, the largest cloud-services provider in the world, enterprise demand for both AI infrastructure and core digital migration workloads has been strong, even as industrywide capacity constraints limit its ability to fully meet the demand.

The company invested heavily in the fourth ⁠quarter to ease those constraints. It launched its AI infrastructure project “Rainier,” ​bringing nearly half a million of its in-house Trainium2 chips online, primarily ​for use by Claude chatbot-maker Anthropic.

Although a smaller unit for Amazon, contributing just 15% to 20% of overall sales, cloud platform Amazon Web Services generates over 60% of the company’s operating profit.