Flash forward: how China’s tech strategy Is reshaping AI

The recent presentation of the world’s fastest flash memory by researchers at Shanghai’s Fudan University is a major milestone in Chinese technological development. PoX, still in the prototype stage, is capable of writing data in as little as 400 picoseconds, which is equivalent to 25 billion operations per second.

Far from being an isolated event, the achievement reflects China’s technology strategy to advance AI by increasing traditional computing power in response to US embargoes through memory and processing integration. As I argued recently, Washington’s misguided policy is encouraging China to come up with evermore ingenious and efficient solutions, drawing on its huge pool of highly qualified engineers.

Fudan’s flash memory breakthrough is critical to overcoming von Neumann architecture bottlenecks as a result of memory and processor working separately. In contrast, integrating both functions increases speed, reduces power consumption and enables parallel data processing directly into storage, which has huge potential for AI applications.

Moreover, this approach to innovation is not limited to universities or research centers: companies like Huawei are developing proprietary chips such as the Ascend 910C and assembling them into configurations capable of maximizing performance in…