Nvidia may have signed multi-year HBM (high-bandwidth memory) and DRAM (dynamic random-access memory) supply agreements with SK hynix and Micron Technology, according to Edgewater Research, as AI infrastructure demand continues to outpace memory production. SpaceX CEO Elon Musk separately said the company is targeting 15 to 20 gigawatts of AI data-center capacity by the end of next year and identified memory semiconductors as its biggest bottleneck. Musk said memory demand is rising about 200% annually, while manufacturers are increasing unit volumes by roughly 20%. The shortage could persist until 2028, supporting higher prices, longer-term contracts and further capacity investment. Musk also said SpaceX would use Nvidia hardware exclusively going forward, citing the Vera Rubin architecture. Nvidia shares rose 3.4%, while AMD fell more than 8% despite record quarterly revenue of $11.5 billion. AMD’s data-center revenue doubled to $6.7 billion, net income reached $2.3 billion, and CEO Lisa Su reaffirmed guidance for server revenue growth of more than 80% in the second half of 2026. Micron is committing $10 billion to research over the next decade and is positioned to benefit from sustained memory demand, although investors continue to weigh AI spending durability, valuation and competition.