Huawei Horinger data center trails SpaceX Colossus 2 in compute
The compute-per-watt figure is the takeaway: even as chip export controls hold China back, its grid capacity means the bottleneck could flip to energy on the US side within a few years. Enterprise planners and policymakers tracking data-center investment, especially in power-constrained ASEAN markets choosing between US and Chinese AI infrastructure, should watch both constraint curves. The concrete benchmark to compare is 141,000 versus 1.8 million H100-equivalents by early 2028.
An Epoch AI analysis of satellite imagery, cited in METR's latest Substack, finds Huawei's flagship AI data center at Horinger holds roughly 105,000 Ascend 910C chips, equivalent to about 81,000 H100s — around 7% of the compute of SpaceX's Colossus 2, which is roughly 14 times larger.
A new building under construction could lift Horinger to about 141,000 H100-equivalents by early 2028, but Colossus 2 is expected to reach 1.8 million H100-equivalents or more by the same date. The gap stems largely from chip constraints: US data centers extract about 2.5 times more AI compute per watt than their Chinese counterparts.
The analysis frames the asymmetry between the two AI ecosystems. China's main bottleneck remains access to advanced chips, while the United States is approaching an energy bottleneck as compute scale grows. The author notes that what Chinese labs have shipped in the past two years is notable given this hardware disadvantage.
METR reports are always worth reading. Just finished reading their latest Substack. If you want to understand how much ahead the US is regarding data center, read those numbers:
Huawei's flagship AI data center has roughly 7% of the compute of SpaceX's Colossus 2, according to a new Epoch AI analysis of satellite imagery.
That's around 105,000 Ascend 910C chips, or roughly 81,000 H100-equivalents. Colossus 2 has about 14x as much.
A new building could push Horinger to about 141,000 H100e by early 2028. Colossus 2 should be at 1.8M or more by then.
However (something I point to repeatedly) China has the power. The chips are the bottleneck: US data centers get 2.5x more AI compute per watt. Makes what Chinese labs shipped in the last two years even more impressive.
The biggest bottleneck for the US will be energy in the very near future, for China the bottleneck stays Chips.
This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Huawei, Ascend 910C, SpaceX, Colossus 2.
Source attribution: curated from kimmonismus (@kimmonismus). Cross-source corroboration: 7 independent sources have covered this development (including The Information, theverge, rohanpaul_ai), confirming industry-wide signal strength.
