Nvidia's 3-Year-Old Chip Still Beats Huawei's Best

Nvidia's lead over domestic Chinese rivals like Huawei is not just about raw performance—it is a generational gap. While Huawei's Ascend 910C delivers less than half the compute of Nvidia's H200, Nvidia has already moved through Blackwell and Blackwell Ultra into full production on Vera Rubin, which quadruples the H200's performance. This widening moat is reflected in Nvidia's financials: Q2 FY27 revenue of $96.22 billion, up 105.85% year over year, with data center revenue up 117%. By contrast, AMD, the closest US-listed peer, posted data center revenue of only $6.72 billion, making Nvidia's data center segment more than thirteen times larger.

The company's capital returns—approximately $26 billion in Q2, with ~$99.0 billion left on its buyback authorization—further bolster its investment case. The main risk, China export controls, appears priced out: Hopper shipments to China were less than 1% of total Data Center revenue in Q2, and Q3 guidance assumes zero from China. As CEO Jensen Huang put it, "AI has reached its inflection point... compute is revenue."

The stakes hinge on whether Nvidia can sustain its growth trajectory as it ramps Vera Rubin, especially given its assumption of zero China revenue. For long-duration holders, the compounding effect of Vera Rubin's performance gains and continued buybacks will be the key story to watch, as any slowdown in the AI infrastructure buildout could pressure the premium valuation.