SoftBank subsidiary Saimemory just inked a deal with Intel to commercialize next-generation memory technology designed for AI's exploding demands. The Z-Angle Memory program targets prototypes by fiscal 2028 and commercial rollout in 2029, addressing critical shortages as AI workloads outpace traditional DRAM capabilities. Intel stock jumped 5% in overnight trading while SoftBank shares climbed 3.13% following Tuesday's announcement, signaling investor confidence that this partnership could reshape the AI infrastructure landscape.
SoftBank isn't backing away from the AI infrastructure race. The Japanese conglomerate's brand-new subsidiary Saimemory just announced a collaboration with Intel to commercialize memory technology specifically engineered for artificial intelligence's brutal performance demands. The partnership, unveiled Tuesday, centers on the Z-Angle Memory program - an effort to build DRAM that doesn't buckle under the weight of large language models and high-performance computing workloads.
The market responded immediately. Intel shares popped 5% in overnight Robinhood trading, while SoftBank climbed 3.13% as investors digested the implications. According to SoftBank's press release, prototypes should arrive by the fiscal year ending March 31, 2028, with commercial production slated for fiscal 2029.
Timing matters here. The semiconductor industry is wrestling with memory shortages that have rippled across the entire supply chain, driven by AI applications that devour bandwidth faster than traditional architectures can deliver. "Standard memory architectures aren't meeting AI needs," Dr. Joshua Fryman, Intel Fellow and CTO of Intel Government Technologies, said in a statement. His team has been developing what they're calling a fundamentally different memory architecture - one that improves DRAM performance while slashing power consumption and costs.
Intel's bringing serious technical firepower to the table. The chipmaker participated in the U.S. Department of Energy's Advanced Memory Technology program, which focused specifically on boosting performance and power efficiency for next-generation DRAM used in servers and computing infrastructure. That government-funded research now becomes the foundation for Saimemory's commercial ambitions.
Saimemory itself is barely two months old, established in December 2024 as SoftBank doubled down on AI infrastructure investments. The subsidiary's entire purpose revolves around tackling the memory bottleneck that's throttling AI advancement. By partnering with Intel rather than building from scratch, SoftBank gains immediate access to years of advanced research and manufacturing expertise.
The emphasis on energy efficiency isn't just marketing talk. Data centers running AI models are consuming staggering amounts of electricity, raising costs and sustainability concerns across the industry. Memory chips that cut power draw in half while maintaining or improving performance could fundamentally change the economics of AI deployment at scale. Fryman positioned the technology for "broader adoption over the next decade," suggesting Intel sees this as a platform play rather than a niche product.
This isn't the first whisper of the collaboration. Nikkei Asia reported on SoftBank-Intel memory development last year, and Japanese IT giant Fujitsu is also reportedly involved in the project, adding another layer of technical and manufacturing capacity to the effort.
The competitive landscape is heating up fast. Memory manufacturers like SK Hynix and Samsung are racing to deliver high-bandwidth memory solutions for AI accelerators, particularly for training massive models. Intel's approach with ZAM appears focused on a different angle - improving the fundamental DRAM architecture rather than just stacking more bandwidth onto existing designs.
What makes this partnership particularly interesting is the strategic alignment. SoftBank has been aggressively positioning itself in AI through investments in companies like Arm and its Vision Fund portfolio. Intel, meanwhile, is fighting to regain relevance in a chip market increasingly dominated by Nvidia's AI accelerators and TSMC's manufacturing prowess. A breakthrough in memory technology could give both companies a crucial foothold in the AI infrastructure stack that doesn't rely on competing directly with entrenched players.
The 2029 commercialization timeline puts ZAM squarely in the window when current AI memory solutions will likely hit their scaling limits. If prototypes deliver on the performance and efficiency promises by 2028, Intel and SoftBank could be positioned to capture significant market share just as demand for next-generation memory reaches critical mass.
The SoftBank-Intel memory partnership represents a calculated bet that AI's infrastructure demands will require fundamentally new approaches, not just incremental improvements to existing technology. With prototypes expected in two years and commercial launch in three, the ZAM program sits at the intersection of SoftBank's AI investment strategy and Intel's fight for semiconductor relevance. If the technology delivers on its promise of halving power consumption while boosting performance, it could reshape data center economics and give both companies a critical piece of the AI infrastructure puzzle. The real test comes in 2028 when prototypes either validate the hype or reveal whether standard DRAM architectures still have more runway than the industry thinks.
