Why soaring memory prices could be Nvidia's next big advantage (www.businessinsider.com)

Nvidia is poised to benefit significantly from rising memory prices, as leading companies like Meta and Microsoft attribute their increased AI spending to the escalating costs of chips and data centers. According to Bernstein, the anticipated Vera Rubin NVL72 system from Nvidia, crucial for training advanced AI models, is set to cost around $9.1 million per rack, with memory and storage expenses alone projected to reach $3.2 million. This surge represents over a third of the total cost, indicating that Nvidia's customers could face considerable financial pressure as they look to scale their operations. The implications of these soaring costs extend beyond memory, as elevated prices in networking, cooling, and power delivery are also contributing factors. Despite this, Nvidia appears well-positioned to leverage its market control, potentially implementing dynamic pricing strategies to pass these costs onto customers rather than adversely affecting its margins. CEO Jensen Huang has warned of prolonged memory shortages, further emphasizing the importance of Nvidia's new multiyear partnership with SK Hynix to develop next-generation memory solutions. As the AI infrastructure landscape evolves, Nvidia's strategic positioning could solidify its dominance in a competitive market.