CXMT has produced HBM3E chips for the first time, putting a Chinese manufacturer into the memory generation that AI accelerators are currently built around. It remains one generation behind Samsung, SK Hynix and Micron, all of which are already mass-producing HBM4. Two domestic customers are lined up: T-Head, Alibaba's chip unit, and Cambricon, both testing HBM3E now and planning to use it in products from 2027.
High-bandwidth memory is several memory dies stacked on top of one another and placed next to the processor. Training and running large AI models depends on it, and American export restrictions make advanced HBM hard for Chinese companies to buy. That is why a domestic HBM3E line matters more than its position on the roadmap suggests: the constraint on Chinese AI silicon has not been the logic die so much as the memory sitting beside it.
The verb is doing a lot of work here, and it is worth separating from the headline. CXMT has produced HBM3E. Samsung, SK Hynix and Micron mass-produce HBM4. First silicon and volume manufacturing are different achievements, and on a product built by stacking several memory dies into one part, the distance between them is mostly yield. Nothing in this announcement says how many CXMT can make, at what cost, or with what share of the parts surviving.
The 2027 date is the other number to sit with. If T-Head and Cambricon put this memory into shipping products two years from now, the gap measured in deployed hardware is wider than the one generation the roadmap shows, because the incumbents will not be standing still at HBM4. On paper this reads like a company closing distance. In shipping product it reads more like a company holding distance while acquiring the ability to supply itself at all.
That is the trade being made, and I think it is the right one from Beijing's side. A domestic supplier one step behind, with captive customers who have nowhere else to buy, is worth more strategically than a faster part bought from a vendor that an export rule can switch off. The three incumbents compete for design wins; CXMT has been handed its demand by policy on both sides of the Pacific.
The question the announcement leaves alone is whether T-Head and Cambricon are testing this memory because it is good enough or because it is available. Those are not the same finding, and the 2027 timeline is long enough to hide either one. Export controls were written to deny China a component. What they have produced so far is a Chinese component supplier with a guaranteed order book and two years to get the yields right.