UBS forecasts hyperscalers spend about 4.1 trillion on capex

UBS forecasts hyperscalers will ramp up capital spending to support artificial intelligence infrastructure, with total capex estimated at about 4.1 trillion from 2026 through 2028. The estimate reflects a shift in how cloud revenue is recycled into new capacity.

UBS projects hyperscaler capex of $492 billion in 2025, rising to $1.009 trillion in 2026, $1.447 trillion in 2027, and $1.619 trillion in 2028. Across 2026 to 2028, that would exceed three times the $1.292 trillion spent over the prior six years, according to UBS.

UBS also estimates Amazon, Alphabet, and Microsoft will collectively spend about 102% of their cloud revenue on capital expenditures in 2026. UBS said the ratio is expected to ease to roughly 99% in 2027 and 94% in 2028, while spending levels continue increasing.

UBS expects the industry buildout to be led by Amazon, Alphabet, Microsoft, and Meta Platforms, while SpaceX is expanding as well. UBS said Oracle, neocloud providers, and newer entrants are also contributing to the broader spending pool.

UBS’s company-level capex estimates for 2026 to 2028 include Alphabet at about $938 billion, Meta Platforms at about $683 billion, Microsoft at about $672 billion, and Amazon at about $628 billion.

UBS estimates SpaceX at about $335 billion, Oracle at about $276 billion, CoreWeave at about $130 billion, and Nebius Group at about $93 billion for the same 2026 to 2028 period.

UBS tied the investment risk to whether large-scale spending translates into sufficient AI revenue and cash flow. UBS noted that capacity utilization, customer payments, and AI services’ ability to cover depreciation, electricity, financing, and operating costs determine whether returns meet investors’ expectations.

UBS also highlighted timing risk, saying infrastructure could take years to reach full utilization. UBS warned that if AI demand grows more slowly than expected, depreciation expenses could outpace revenue and put pressure on margins and free cash flow.

UBS said the scale of commitments across multiple customers reduces the chance the buildup is only one company’s speculative bet. UBS framed the industry’s central question as whether AI demand stays large enough to keep new infrastructure productive for years.