Memory now dominates AI chip costs as GPU prices plateau

As GPU prices have stabilized, memory has become the largest cost driver in AI chips, accounting for nearly two-thirds of component expenses. For indie makers and small teams building AI products, this shift means memory optimization—not raw compute—is becoming the real bottleneck for cost-effective inference and fine-tuning.