NVDA
SOXX
Nvidia’s not done flexing
Goldman Sachs just raised its revenue and earnings estimates for Nvidia and, in polite Wall Street language, said: maybe this AI story is bigger than the $1 trillion headline already plastered everywhere.
The bank thinks Nvidia could beat first-quarter expectations by about $2 billion in revenue and then keep the momentum rolling into the next quarter. That matters because the stock has been playing catch-up while the broader semiconductor crew, including the iShares Semiconductor ETF, has been having a very good time.
The real question: is $1 trillion already too small?
Nvidia has told investors to expect $1 trillion in cumulative revenue from Blackwell, Blackwell Ultra, and Rubin across calendar 2025 through 2027. Goldman’s point is simple: that number may not include the whole buffet.
What’s left off the menu?
- Rubin Ultra, which is expected in 2027
- Vera CPU-only racks, timed for the second half of 2026 and beyond
- Inference-focused setups like Rubin-CPX and Groq 3 LPX
In other words, the market is trying to price a rocket ship, and Goldman is asking whether we’re only looking at the first stage.
What investors will be listening for
When Nvidia reports on May 20, traders won’t just care about whether the numbers beat. They’ll be listening for clues on:
- whether that $1 trillion framework gets upgraded
- demand from non-hyperscaler buyers like OpenAI and Anthropic
- how agentic AI is changing the mix toward more general-purpose compute
- whether memory costs and Rubin ramping pressures start nibbling at those juicy gross margins
Big picture: Nvidia’s AI story is already huge. The annoying part for bears is that it may still not be big enough.