AI Infrastructure Spending Hits $1.2T in 2027

Goldman Sachs initially projected that six hyperscalers would invest about $1.73 trillion in AI infrastructure in 2026 and 2027, with spending potentially reaching $4.14 trillion from 2028 to 2030. Its later forecast focuses on five companies—Amazon, Alphabet, Microsoft, Oracle and Meta—and estimates $800 billion in AI infrastructure capex in 2026 and $1.2 trillion in 2027. The 2027 figure is above Wall Street’s $1.1 trillion consensus, while an upside scenario reaches $1.4 trillion. Cumulative spending from 2026 to 2031 could total $7.6 trillion, with annual spending potentially reaching $2 trillion in 2028 despite slower growth. The scale of AI infrastructure spending implies that hyperscalers may need about $300 billion in annual AI revenue to break even. Goldman estimates that roughly $1 trillion in annual revenue may be required to generate attractive returns, creating monetisation, valuation and execution risks. Cloud backlogs for AWS, Azure and Google Cloud had reached $1.69 trillion in the second quarter of 2026, up 152% year on year. Energy supply is the main constraint, alongside labour shortages and limited high-bandwidth memory chips. Expanding power generation, data-centre capacity and semiconductor production could take years. For traders, the AI infrastructure theme supports semiconductor, data-centre and energy stocks, but also increases leverage and interest-rate sensitivity across the tech sector. Policy developments, including discussions involving Nvidia, Anthropic and Palantir executives, could add volatility through AI regulation, export controls and government contracts. The news has no direct cryptocurrency catalyst, but risk sentiment around technology and power demand could indirectly affect crypto markets.