\n\n
Key developments: Chinese firms have bought over 1 million Nvidia gaming and workstation GPUs this year to fill a domestic AI compute gap, while Nvidia’s own AI‑accelerator shipments to China fell 44% in the first half of 2026, ceding market share to Huawei. Industry estimates put China’s AI‑chip supply at about 3.5 million units in 2026 against a demand of more than 5 million, prompting Alibaba, ByteDance and Tencent to double their AI‑related capex to roughly $104 billion. The surge in spending is driving a broader wave of AI‑focused data‑center construction, a trend highlighted by former UK prime minister Rishi Sunak who warned that technology companies will become the largest issuers of long‑term debt by 2027, outpacing US government borrowing. In Japan, NTT Data is close to finalising a deal to build Anthropic’s first dedicated data centre, a 1 trillion‑yen (US$6.3 billion) project with 500 MW power capacity slated for partial operation by 2028.
Why it matters: The aggressive procurement of Nvidia hardware and the rapid expansion of domestic chip production signal China’s push to secure the compute backbone essential for next‑generation AI models. Shrinking Nvidia shipments and rising Huawei output underline a shift toward home‑grown supply chains, reducing reliance on US‑origin technology. The associated financing pressures—illustrated by Sunak’s debt‑issuance forecast—highlight how AI‑capex is reshaping global capital markets and could strain sovereign debt levels. Large‑scale data‑center projects like the planned Osaka facility further cement the physical infrastructure needed for AI dominance.
What to watch next: • Whether Nvidia’s market share continues to slide in China as Huawei and other local vendors scale production. • The pace and financing of new data‑center builds in China and Japan, especially the Osaka Anthropic hub slated for 2028. • International policy responses, including potential export controls or sovereign AI initiatives, as governments grapple with the funding and supply‑chain implications of the AI compute race. [[1]]