What Nvidia's CFO Actually Said On The Call
Colette Kress told analysts that Nvidia is experiencing extreme pricing conditions in memory. The Nvidia CFO said the magnitude of the price increase has exceeded the company's own prior expectations and is headed even higher into next year. That sentence sat inside a results call that otherwise read as unambiguous good news: $96.2 billion in quarterly revenue, up 106 percent year on year, and a Data Center segment that grew 117 percent to $89.0 billion.
This is the same shortage every hyperscaler and memory maker has already flagged this year. The difference is who is now pricing it: Nvidia's own CFO, live on an earnings call, translating that shortage directly into a lower gross margin figure for the next quarter.
The Numbers Behind The Guidance
Four figures from the call carry the signal a buyer actually needs.
| Metric | Q2 FY2027 | Forward guidance |
|---|---|---|
| Total revenue | $96.2B (+106% YoY) | - |
| Data Center revenue | $89.0B (+117% YoY) | - |
| Gross margin | 75.0% | 71-72% (Q4) |
| FY2028 revenue growth | - | about 70% (was about 44%) |
Revenue and Data Center growth describe the quarter that already happened. The margin guidance and the FY2028 revenue guidance describe what Nvidia itself expects to happen next, and those are the two numbers a procurement team should be tracking.
Why The Margin Cut Is The Real Signal
Nvidia's own gross margin fell from 75.0 percent in Q2 to a guided 71-72 percent for Q4, and the company named the cause. Kress tied that decline directly to memory cost, not to pricing pressure on GPUs themselves or to competition, which is the detail a buyer should sit with.
Revenue is not slowing down; margin is. That gap matters because it shows the shortage has moved past Nvidia's suppliers and into Nvidia's own cost base, which means the same pressure is already sitting inside every quote a reseller or cloud provider sends out this quarter.
The FY2028 Guidance Hike Nobody Priced In
Nvidia raised its own revenue growth guidance for FY2028 to roughly 70 percent, up from about 44 percent expected before the call. That increase landed in the same breath as the margin cut, and both numbers came from the same set of underlying costs.
Nvidia also confirmed that its outlook assumes zero Data Center compute revenue from China. The higher growth guidance is therefore a bet on volume and pricing power outside China alone, at a moment when the company is simultaneously guiding its own margin down because of memory cost.
What This Means For A European Buyer Negotiating 2027 Capacity
A European enterprise negotiating GPU or cloud capacity for 2027 now has a hard, dated reference point. Account teams and resellers have spent months describing memory cost increases as unavoidable pass-through, but that argument previously rested on their own word.
Nvidia's CFO put a number on it, in a public filing, on the same day analysts and press could check the transcript against the numbers. For a buyer building a 2027 budget in EUR or GBP, that call is now the reference point to hold a vendor's account team against.
Read next: AWS's Anti-Nvidia Chip Adopts Nvidia's Memory | Why OpenAI's New Chip Won't Replace Nvidia
