The AI CapEx: U.S. hyperscalers vs China’s big four This…
0 reactions
The AI CapEx: U.S. hyperscalers vs China’s big four
This chart compares estimated AI/cloud infrastructure CapEx for two baskets of major hyperscalers:
**U.S.:** AWS, Microsoft, Google, Meta, Oracle
**China:** Alibaba, Tencent, ByteDance, Baidu
The main takeaway is not only that the U.S. group is much larger in absolute dollars. It is that the estimates themselves have been repriced upward quickly.
An earlier Goldman Sachs Research estimate shown in an AllianzGI deck had 2027E CapEx at roughly **$710B** for the U.S. group and **$67B** for the China group. Updated estimates put the figures closer to **$1.0T** and **$123B**, respectively.
That means the AI buildout is not just getting bigger. It is getting more expensive, as demand for GPUs, HBM/DRAM, power, cooling and data-center capacity collides with supply bottlenecks.
Important caveat: this is **selected hyperscaler CapEx**, not total national AI investment.
#education
source
This chart compares estimated AI/cloud infrastructure CapEx for two baskets of major hyperscalers:
**U.S.:** AWS, Microsoft, Google, Meta, Oracle
**China:** Alibaba, Tencent, ByteDance, Baidu
The main takeaway is not only that the U.S. group is much larger in absolute dollars. It is that the estimates themselves have been repriced upward quickly.
An earlier Goldman Sachs Research estimate shown in an AllianzGI deck had 2027E CapEx at roughly **$710B** for the U.S. group and **$67B** for the China group. Updated estimates put the figures closer to **$1.0T** and **$123B**, respectively.
That means the AI buildout is not just getting bigger. It is getting more expensive, as demand for GPUs, HBM/DRAM, power, cooling and data-center capacity collides with supply bottlenecks.
Important caveat: this is **selected hyperscaler CapEx**, not total national AI investment.
#education
source
| 3,000 mlx | |
| 4 | |
| 0 |