Nvidia Warns: AI Memory Supply Bottleneck May Persist Until End of Fiscal Year 2028
Nvidia reported Q2 fiscal year 2027 revenue of $96.2 billion, doubling year-over-year, but executives warned that memory supply bottlenecks may persist until the end of fiscal year 2028. Data center business revenue reached $89 billion, with hyperscale customers contributing $49 billion. CFO Colette Kress stated that the company is collaborating with three major memory suppliers to expand capacity, and CEO Jensen Huang expects fiscal year 2028 revenue to grow 70% year-over-year. Analysts noted that enterprises should focus on AI investment returns rather than merely purchasing GPUs.
Key Takeaways
- Nvidia reported revenue for the second quarter of fiscal 2027 (ending July 2026) of$96.2 billion, more than doubling year over year, driven by continued investment in AI infrastructure, executives said on Wednesday's earnings call.
- The tech giant'sdata center revenuereached$89 billion, up 117% year over year, despite memory shortages. Of that, hyperscale cloud providers contributed$49 billion, while enterprise, AI cloud, and industrial customers contributed$40 billion。
- Hyperscale customers remain the primary growth engine, CFO Colette Kress said on the call. Nvidia announced Thursday an expanded partnership with AWS to deployan additional 2 million Nvidia GPUson AWS infrastructure to meet surging demand.
Deeper Insights
Nvidia's sustained high growth has not been hindered by tight memory supply—a bottleneck caused by AI buildout that the company expects to last untilthe end of fiscal 2028。
"The tightening memory supply is a reflection of the same surge in demand driving our growth," Kress told investors. "We have long-standing, deep relationships with the three major memory suppliers and are working closely with them to further increase the capacity needed for our roadmap."
Nvidia acknowledged that supply constraints will impact itsfiscal 2028revenue outlook—which calls for 70% year-over-year growth. President and CEO Jensen Huang told investors on the earnings call. The company is working to secure infrastructure up and down the supply chain to support elevated demand.
As hyperscale customerssnap up available resourcesto support AI workloads, global semiconductor revenue is expected toexceed $1.3 trillion in 2026, with DRAM and NAND flash chip prices expected to rise125% and 243%, according to Gartner.
Forrester Principal Analyst Naveen Chhabra said in an emailed statement that enterprises should expect AI compute demand and pricing pressure to remain high, with no significant GPU price declines in the next one to two years.
Chhabra noted that Nvidia's results highlight that CIOs should extract measurable business value from AI, rather than simply pursuing more GPUs.
"For enterprise technology buyers, the key takeaway is that AI infrastructure demand is still accelerating, but supply constraints, ecosystem lock-in, and ROI pressure are becoming dominant strategic issues," Chhabra said.