MediaTek secures second AI chip customer for 2028 ramp

My assessment is that the important signal is potential customer diversification, not an immediate increase in available compute. A second design relationship could give MediaTek another opportunity to build a durable accelerator business. But the reported production date is 2028, and the customer remains unnamed, so the announcement does not establish capacity available to frontier labs today. Those limitations are central to interpreting the story. ([ithome.com](https://www.ithome.com/1/020/400.htm))

For the AGI race, the competitive question is whether another custom-chip supplier can eventually help buyers obtain useful compute at better economics. That outcome would depend on execution across chip design, software support, manufacturing, and deployment. A customer win alone cannot answer those questions. Investors should therefore distinguish a promising commercial pipeline from delivered infrastructure, while researchers should avoid treating a future production target as evidence of an imminent training breakthrough. If the project reaches production on schedule and performs competitively, it could broaden hardware options. Until then, its significance is strategic positioning: a possible future alternative whose practical contribution remains unmeasured.