\n\n
Major chipmakers are feeling the pressure of a rapidly expanding AI‑driven memory market. Samsung Electronics is expected to see an "18‑fold jump in profit on surging AI demand for memory," highlighting how AI workloads are turning memory capacity into a revenue engine. At the same time, both Samsung and SK Hynix saw their shares tumble—Samsung down 4.99% and SK Hynix down 5.98%—after China’s Changxin Memory Technologies announced it will start mass‑producing a next‑generation DRAM with smaller cell sizes. This new Chinese supplier threatens the Korean duopoly that currently controls about 83% of global DRAM, intensifying competition and prompting a broader market slide of 4.4‑4.5% on AI‑memory concerns.
The impact goes beyond DRAM. Intel has introduced a new AI‑accelerator that ships with 160 GB of memory, with 480 GB versions planned for 2027, positioned as a cheaper, denser alternative to Nvidia’s and AMD’s flagship GPUs. By offering higher memory density at lower cost, Intel aims to re‑enter the AI data‑center arena after previous setbacks.
Why it matters: AI models are becoming ever larger and more data‑intensive, making high‑capacity, low‑latency memory a critical bottleneck. New DRAM technologies from China and scaling‑focused chip designs from Intel could democratize access to AI compute, lower operating costs, and reshape the competitive landscape among memory‑centric firms.
What to watch next:
1. How quickly Changxin’s next‑gen DRAM can achieve volume and price parity with Korean offerings.
2. Whether Samsung’s projected profit surge materializes as AI workloads continue to scale.
3. Intel’s rollout timeline for its 480 GB AI accelerators and customer adoption rates.
4. Potential policy responses or supply‑chain adjustments as the memory market becomes increasingly contested. [[1, 2]]