China's hyperscalers trail US AI capex roughly 8-to-1, Goldman Sachs estimates
China's largest internet groups are ramping up artificial-intelligence capital expenditure, but remain far behind their US counterparts in absolute terms, according to estimates from Goldman Sachs Global Investment Research.
The bank's internet team groups Alibaba, Tencent, ByteDance and Baidu as China's leading AI hyperscalers. Their combined AI capex is estimated to rise from US$8bn in 2022 to US$57bn in 2025, then to US$102bn in 2026 and US$123bn in 2027.
Over the same period, Goldman estimates US hyperscalers — Amazon Web Services, Microsoft, Google, Meta and Oracle — will scale spending from US$156bn in 2022 to US$443bn in 2025, US$764bn in 2026 and US$1,018bn in 2027. That leaves the US group spending roughly eight times more than the Chinese group at the 2027 estimate, a gap that persists even after adjusting for purchasing-power-parity differences.
Goldman notes that, because ByteDance is privately held and not formally covered, its figures are extrapolated from the operating metrics of comparable listed peers.
The estimates appear as Exhibit 9 in Goldman Sachs Global Investment Research and were circulated by Goldman senior economics writer Callum Williams.
Read more: Callum Williams (X) · Goldman Sachs