MemryX AI announced the MX3+ inference accelerator on November 10, 2026, setting design targets of up to 10 times more peak AI compute and nearly 10 times more on-chip memory than MX3 while remaining within the same low-power operating envelope. The fabless semiconductor company is aiming the device at edge hardware, industrial systems and high-density inference infrastructure running larger or more complex models. Its MX3+ product page adds that the design is intended to retain MX3’s silicon…