DeepSeek Triggers Buying Frenzy While Huawei Muscles Into Nvidia Territory

Huawei just doubled its AI chip yields to 40 percent. Quite the achievement for a company Washington tried to knee-cap with export controls. Their Ascend chip production line turned profitable for the first time—champagne corks likely popped in Shenzhen that day.

Meanwhile, Chinese tech giants hoard Nvidia H20 chips like apocalypse preppers stocking canned beans. Tencent, Alibaba, and ByteDance can't get enough of them to power DeepSeek's suddenly popular AI models. Even smaller companies jump into the AI game now. The velvet rope has fallen.

Huawei aims to crank out 100,000 of its latest 910C processors this year. They'll make 300,000 of the older 910B chips too. Not bad for technology America thought it had contained.

The company partnered with SMIC to create a clever workaround. They developed a production process that doesn't need those banned Dutch EUV machines. Necessity mothers invention, as they say.

Nvidia still dominates, though. The American firm sold a million H20 chips to Chinese customers last year. That's $12 billion in revenue—enough to make even the wealthiest tech executive weep with joy.

Technical headaches plague Huawei's efforts. Their 910B chips stumble during large-scale AI training. Memory issues and connection problems haunt them like ghosts in ancient hardware. They're working to fix these bugs in the newer 910C models.

Donald Trump's administration now threatens to cut off the H20 lifeline. Chinese firms scramble to place orders before potential new restrictions hit. Nothing motivates procurement departments like impending scarcity.

Beijing pushes local companies to buy Huawei chips instead. They want tech independence. Huawei's founder recently told President Xi that China's "lack of core and soul" worries have eased. Translation: "We're figuring this chip stuff out, boss."

Why this matters:

  • The technological cold war intensifies—with China proving sanctions create innovation just as often as they create barriers
  • AI hardware democratization marches forward, pulling sectors like healthcare and education into computing's big leagues
  • Geopolitical chess moves increasingly determine who can build what technology, creating a patchwork global innovation landscape

Read on, my dear: