Chinese firms plan to route 46% of AI chip spend to local silicon

The Chinese buyers are moving toward domestic AI accelerators. Everyone already knew that. What is worth pausing on is how fast the intent has shifted in a single planning cycle.

Bloomberg Intelligence's latest survey, released Tuesday, has executives in China saying they will allocate 46% of their AI-accelerator budget to domestic products over the next 12 months, up from 30% today. That is a 16-percentage-point swing in one year, and Bloomberg frames it as Chinese companies ditching Nvidia's advanced accelerators in favor of local silicon as US tensions reshape the AI infrastructure buildout.

The second number in the survey is the one worth sitting with. 80% of executives said their total infrastructure spending is running over budget this year, mostly because of the high cost of AI-related projects. So the shift toward local chips is not landing in a comfortable spend environment, it is landing in one where buyers are already hunting for cost relief. That reframes the swing as more than a compliance response. If domestic parts are cheaper as well as politically safer, the intent number is more likely to harden into actual orders.

The honest caveat is what the reporting does not give you. The summary does not name which domestic vendors are absorbing the shift, does not disclose the survey's sample size or industry mix, and does not say how respondents ranked performance against Nvidia's high end. So take the 46% as directional intent, not a purchase-order forecast. Surveys of this kind tend to overstate near-term substitution when a policy tailwind is this obvious.

For Nvidia the read is that its China line is being budgeted away rather than merely restricted, which is a slower and more permanent kind of pressure. For the domestic accelerator vendors, the window to prove their silicon can actually carry the workloads Chinese buyers are already planning around is now, before that intent softens.