Investment in artificial intelligence infrastructure by the five biggest US hyperscale cloud providers is projected to climb by over 50% in the coming year, reaching $1.2 trillion, according to analysts at Goldman Sachs Group Inc.

This projection surpasses the Wall Street consensus of $1.1 trillion in capital expenditure for 2027, as noted by the team headed by Ryan Hammond. They also indicated that to achieve breakeven on these investments, the companies would require annual AI revenues of roughly $300 billion in the years ahead.

"According to consensus forecasts, capital expenditure in 2027 is expected to account for a larger portion of GDP than any other technological investment cycle since the railway expansion of the late 1800s," Hammond and his associates stated.

Amazon.com Inc., Alphabet Inc., Microsoft Corp., Oracle Corp. and Meta Platforms Inc. are set to allocate $800 billion this year toward expanding their AI capabilities. Although next year's number marks a significant rise, the rate of increase in capital expenditure is anticipated to decelerate, according to the Goldman team.

"The pace of capital expenditure growth and the frequency of upside surprises relative to consensus estimates will decline compared to recent quarters," the analysts noted. Hyperscaler capital expenditure now surpasses operating cash flow, heightening the demand for debt and equity funding. Possible limitations include shortages in power, labor, and memory, along with constraints on data-center building.

They anticipate that hyperscaler capital expenditure growth will slow from nearly 100% this year to 54% in 2027 and to 12% in 2028, when it is expected to reach $1.4 trillion.

AI users would need to invest approximately $1 trillion annually in applications for hyperscalers to achieve robust returns and for application providers to secure healthy margins on their computing expenses, the analysts said. This is in contrast to around $1.5 trillion in global software spending projected for 2026.

Present estimated AI revenue is still under the threshold needed for hyperscalers to break even, even though sales are increasing rapidly, the Goldman team said. Cloud revenue growth at Amazon, Alphabet, Microsoft, and Oracle sped up to 48% in the second quarter of this year from 25% in 2024. Amazon, Alphabet, and Microsoft have also disclosed a combined $1.7 trillion in cloud revenue backlog.

Investor doubts about the longevity of the AI surge are evident in stock valuations. The median AI infrastructure stock currently trades at 22 times forward earnings, a decrease from 32 times in April, the analysts said. The hyperscalers have seen their multiples drop to the lowest in over ten years and now trade at the smallest premium to the median S&P 500 Index company ever recorded.