According to updated estimates from the American multinational investment bank Citigroup, IT giants would invest more than $2.8 trillion in AI-related infrastructure by 2029. In addition to the initial estimate of $2.3 trillion, the modification was made.

The investment banking group believes that a growing enterprise appetite and aggressive early investments by hyperscalers will be the main drivers of this expansion.

Since OpenAI launched its ChatGPT in November 2022, the demand for AI systems has continued to soar worldwide, driving startling capital expenditures and data center construction. This comes in spite of the short-lived crisis of confidence brought on by China's less expensive DeepSeek model and persistent market apprehensions over US President Donald Trump's expansive tariff policy.

The Wall Street brokerage now projects that AI capital expenditures across hyperscalers might reach $490 billion by the end of 2026, up from an initial forecast of $420 billion.

Hyperscalers are anticipated to show this additional investment in their third-quarter earnings calls, according to Citigroup analysts, with guidance "building ahead of visible enterprise demand."

Also Read: Karnataka to Become Quantum Capital of Asia Soon

According to Citi, by 2030, the world's AI compute demand would require 55 gigawatts of additional energy capacity, or $2.8 trillion in additional spending.

Also Read: Jack Dorsey's BitChat Could be Signaling a New Shift in Messaging Paradigms

The United States alone will contribute $1.4 trillion of this amount. Citigroup also believes that AI infrastructure is no longer being financed solely by earnings for Big Tech companies. The companies are borrowing to stay up, and the expenses are extraordinarily high—roughly $50 billion for every 1GW of compute power.

Citigroup claims that Big Tech's financial reports already show this change, with expenditure beginning to eat away at free cash flows. Due to the shortcomings of old methods, investors are also starting to wonder how IT companies will finance this level of investment.

Also Read: Semicon India 2025: Designing A Self-Reliant Semiconductor Hub

Goldman Sachs, on the other hand, had previously predicted that the rise of AI capital expenditures would decelerate starting in the fourth quarter of 2026. In 2022, hyperscalers spent a total of $158 billion.