Mar 13, 2026 · 24-Hour AI Briefing: Samsung and NVIDIA Bet on Next-Gen NAND, While WeRide Brings Robotaxis Into WeChat
Two updates in the last 24 hours, one at each end of the AI stack. Samsung and NVIDIA are pushing to speed up next-gen NAND development, a sign that storage is quietly becoming one of the bigger bottlenecks in AI infrastructure. WeRide, meanwhile, just brought its Robotaxi service into WeChat — a reminder that getting people to actually use new technology usually comes down to convenience, not capability.
1. Samsung × NVIDIA: Using AI to build better chips, 10,000x faster
Samsung and NVIDIA are working together to accelerate next-gen NAND flash development, with a new "physics-informed neural operator" model that analyzes ferroelectric NAND performance at speeds reportedly over 10,000 times faster than existing methods.
The real story isn't the partnership itself — it's that they're using AI to change how chips get developed in the first place. FeNAND research has always been slow going: materials are hard to model, optimization takes endless trial and error, and each test cycle can drag on for months. If this model can genuinely compress that timeline to days, the downstream effect isn't just faster research — it's a shorter path from lab design to actual production.
The timing matters. NAND prices jumped 90% quarter-over-quarter in Q1, and NVIDIA's upcoming Vera Rubin accelerator is expected to use new NAND — potentially consuming around 9.3% of global production capacity on its own. If supply can't keep up, the bottleneck won't just hit customers; it could start slowing NVIDIA's own roadmap. SK Hynix is also moving fast on 3D FeNAND for in-memory computing. Ferroelectric storage is no longer a research curiosity — it's becoming a serious competitive front.
That said, faster simulation doesn't guarantee better yields. Whether any of this translates into stable mass production is still the real question.
2. WeRide × Tencent Cloud: Robotaxi is now inside WeChat
WeRide has deepened its partnership with Tencent Cloud, integrating Robotaxi directly into WeChat Mini Programs. Users in Guangzhou can now hail a ride, take it, and pay — all without leaving WeChat.
On paper it's a simple integration. In practice it's a meaningful shift: Robotaxi is starting to feel less like a tech demo and more like an actual consumer service. The autonomous driving industry tends to talk in miles driven and safety metrics, but most people just want something that's easy to use. Removing the friction of a separate app download — and plugging into an ecosystem people already live inside — goes a long way toward making that happen.
There's also a scale argument. More users means more trips, more trips means more data, more data means faster iteration. WeChat isn't just a distribution channel here; it's a flywheel. Guangzhou helps too — it's one of the few cities globally where Robotaxi can operate point-to-point across core urban areas, which gives WeRide the route density it needs to actually build that data advantage.
Zooming out, WeRide's strategy is becoming clearer: Uber, Grab, and TXAI overseas; Tencent at home. The goal seems to be less "great Robotaxi company" and more "mobility layer that plugs into wherever people are already booking rides." Ambitious framing. Whether the operational numbers in 2026 back it up is another matter.
Most important AI events from the past 72 hours
Meta expands its in-house chip strategy while Cursor moves closer to a $50B valuation