Back to News
Industry updatev1Tom's Hardware

CXMT LPDDR6 Mass-Production Claim Conflicts With Its Sampling Disclosure

A report says CXMT is producing LPDDR6 for a Xiaomi phone, while the company's half-year disclosure describes customer sampling and a production ramp still under way.

Original title: China's CXMT beats Western chipmakers to announcement of LPDDR6 mass production — Xiaomi smartphones to debut industry’s first LPDDR6 chips

John Wood

Public reporting currently gives two different levels of progress for CXMT’s LPDDR6. Tom’s Hardware reported on August 31, citing Reuters, that CXMT had started mass production for a Xiaomi smartphone and that Xiaomi would debut the first device using the memory. The article did not identify a handset model, a chip part number, shipment volume, or a public Xiaomi confirmation.

Two days earlier, IT Home described a more cautious milestone from CXMT’s 2026 half-year report. According to that report, CXMT’s internally developed LPDDR6 reaches a peak rate of 12,800Mbps and a maximum capacity of 16GB per chip. It had been sent to key customers for sample validation, and the company was accelerating work toward mass production. “Mass production has started” and “customer sampling while mass production is being advanced” are not the same production stage. The latter still includes customer qualification and manufacturing ramp work.

Tom’s Hardware also summarized the broader LPDDR6 generation, including PAM3 signaling, a standard transfer-rate ceiling of 43.2 GT/s, and a reduction in I/O width from 32 bits to 24 bits. Those are architectural characteristics of the generation. They do not show that CXMT’s disclosed 12.8Gbps product reaches the standard ceiling, nor do they replace device-level evidence for power, stability, and repeatable volume supply.

The defensible public position is therefore narrower than the headline claim. CXMT has disclosed LPDDR6 parameters and customer sampling progress through its half-year reporting, while the stronger Xiaomi debut and mass-production statement remains attributed media reporting. Without a named phone, a formal bill of materials, supply volume, yield data, or announcements from both companies, the available evidence cannot distinguish limited introduction or engineering validation from sustained commercial supply across multiple products.

The useful next evidence would be a CXMT or Xiaomi announcement naming the product, speed, and capacity; an independent teardown after the handset ships; and CXMT updates on production yield, customer coverage, and shipment scale. If those appear, LPDDR6 can move from a sampling milestone to verifiable market supply. If the public company language remains at customer validation and production preparation, the event should not be described as completed large-scale mass production.