American multinational investment bank Citigroup revised its projections for AI-related infrastructure investment by tech giants upward to exceed $2.8 trillion through 2029. The revision comes on top of the initial projection of $2.3 trillion estimated earlier. The investment banking group opines this growth will be driven by aggressive early investments by hyperscalers and a growing enterprise appetite. Since OpenAI unveiled its ChatGPT in November 2022, AI growth has continued to drive staggering capital outlays and data center expansion as demand for AI systems continues to skyrocket globally. This is despite the brief crisis of confidence sparked by China’s cheaper DeepSeek model and lingering market concerns over US President Donald Trump’s sweeping tariff regime. Now, AI capex across hyperscalers is seen reaching as much as $490 billion by end of 2026, according to the Wall Street brokerage’s projections, an increase from an earlier estimate of $420 billion. Citigroup says compute demand will need new energy capacity According to a Reuters article, data center operators or hyperscalers, including Amazon, Alphabet, and Microsoft, have already poured billions of dollars into investments to ease capacity constraints that have limited their ability to meet the ballooning demand for AI services. A Barrons report shows that over the past four quarters, Alphabet, Amazon, Meta and other tech firms that operate global networks of data centers – data hyperscalers – have invested over $300 billion in AI capital expenditures. This month alone, American Big Tech firms made vast commitments for AI infrastructure investments in the UK, during a second state visit by President Trump in that country. Nvidia, Google, and Microsoft pledged billions of pounds towards data center and other AI-related developments, as was previously reported by Cryptopolitan . Analysts at Citigroup have said hyperscalers are likely to reflect this incremental spend in their third quarter earnings calls, with guidance expected to be “building ahead of visible enterprise demand.” Citi estimates global AI compute demand would need 55 gigawatt of new energy capacity by 2030, translating to $2.8 trillion in incremental spend. Of this total, the US alone will account for $1.4 trillion. Citigroup is also of the view that Big Tech firms are no longer relying only on profits to fund AI infrastructure. The costs are extremely high – at about $50 billion for every 1GW of compute capacity and the firms are also resorting to borrowing to keep up. According to Citigroup, this shift is already exhibited in Big Tech’s financial reports, with spending starting to chew into free cash flows. Investors are also beginning to ask how the tech firms will fund this scale of investment, especially as traditional models fall short. “Enterprises have provided a clear external validation of value,” Citi said. However, Goldman Sachs earlier projected otherwise, saying AI capex growth will slowdown beginning in the fourth quarter into 2026. Spending by hyperscalers in 2022 totalled at $158 billion. “This extreme reduction in capex would likely be accompanied by a deterioration in the outlook for long-term AI-driven earnings growth, weighing on valuations as well,” Goldman analysts wrote in their note earlier this month. Join a premium crypto trading community free for 30 days - normally $100/mo.
