The giant Intel has made its boldest strategic move in decades. The American company has officially announced its return to the production of DRAM memory - a sector it left back in 1985. The partnership with Japanese conglomerate SoftBank and its subsidiary SAIMEMORY has one main goal: to create the next generation of high-speed modules capable of powering resource-hungry data centers of artificial intelligence.
At the heart of this collaboration is the Z-Angle Memory (ZAM) technology, which promises to rewrite the rules of the game. While the current gold standard HBM relies on complex vertical connections, the ZAM architecture uses an innovative stepped topology. The results of the first tests are impressive: the new memory consumes between 40% and 50% less power and offers a capacity of up to 512 GB on a single chip. This is not just an improvement, but a technological leap that could provide Intel and SoftBank with a decisive advantage over the current market leaders.
The project builds on Intel's developments under the Next Generation DRAM Bonding (NGDB) initiative, funded by the US Department of Energy. Although the ambitions are huge, the road to market implementation will take time. The first working prototypes are expected to be presented in the spring of 2028, and mass commercial production is planned for the fiscal year 2029-2030. For investors, the news was received with euphoria, with the shares of both companies immediately rising.
This move is a clear signal that Intel does not intend to remain on the periphery of the AI revolution. With ZAM, the company is not only returning to its roots as a memory manufacturer, but also trying to offer an alternative to the expensive and difficult-to-access solutions that are currently slowing down the development of global computing infrastructure. If the promises of lower cost and twice the efficiency are realized, Bavarian precision in chips could soon have serious competition in the face of Munich and Japanese engineers.