NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) reports fiscal Q1 2027 results later today, and the bar has rarely looked higher. The traders on CNBC’s Fast Money segment “Yields Surge and Prepping for Nvidia Results” framed it around a single number: “Anything south of 75% gross margins, then we can start having a different conversation. 75% is a bogey.”
The setup heading into tonight
I have owned NVIDIA for over 15 years, and the math problem facing Jensen Huang’s team has never been quite this strange.
The company guided Q1 FY27 revenue to roughly $78.0 billion, excluding any Data Center compute revenue from China, while Street whispers have crept toward $80 billion or more. Non-GAAP gross margin was guided to 75.0%, plus or minus 50 basis points. Q4 FY26 came in at 75.2% non-GAAP, so the 75% line is real, and management has been telegraphing it for a year.
The stock sits at $220.61, putting the market cap near $5.7 trillion at the milestone Reddit was celebrating last week. NVDA is up 18% year to date and 63% over the past year. The implied move tonight is about 6%.
The 75% bogey and the growth math problem
One Fast Money trader laid out the law of large numbers cleanly: “If you’re looking for 70% earnings and sales growth year over year, that’s what it’s expected on like a $360 billion base. I mean, you’re getting to a point where it’s just going to be hard to really beat those sorts of expectations.” NVIDIA closed FY26 with $215.94 billion in revenue, up 65.5% year over year, and $96.58 billion in free cash flow. Q4 Data Center alone was $62.31 billion, up 75%, with Networking at $10.98 billion, up 263% on NVLink for GB200 and GB300.
Margin is where the thesis gets tested. Huang’s Q4 commentary called Grace Blackwell “the king of inference today, delivering an order-of-magnitude lower cost per token,” with Vera Rubin lined up behind it. If pricing power on Blackwell Ultra and the early Rubin ramp holds, 75% is defensible. If hyperscalers start squeezing on rack-level pricing, the bogey breaks.
China, infrastructure, and the buyback wildcard
There are three swing factors I am watching. First, China: reports that ByteDance, Alibaba, and Tencent now have clearance for H200 chips would reopen a market NVIDIA assumed at zero in its guide. Second, infrastructure. Power, cooling, and water demands for next-gen chips are becoming potential constraints. As Fast Money flagged, if Nvidia signals it can fill orders quickly, that could mean power constraints are easing, which previously hurt Microsoft’s stock.
Third, capital return. NVIDIA returned $41.1 billion to shareholders in FY26 and ended Q4 with $58.5 billion in remaining buyback authorization. One trader on the segment said the stock “has not been realizing its moves the last few quarters” and suggested “the incremental trade here might actually be the buybacks.” The pattern backs that up: NVIDIA has beaten profit estimates in 18 of the past 20 quarters, yet the stock has underperformed semi peers recently. The FY26 reaction table is brutal, with an average day-of change of -2% across four consecutive beats.
What I am actually watching
Polymarket is pricing the earnings beat at 97.9%, with analyst consensus at 58 buys, 2 holds, 1 sell and a $272.94 target. The detail that matters is buried in the filing footnotes: starting Q1 FY27, non-GAAP figures will include stock-based compensation, which runs about $1.9 billion this quarter. You can read the prior Q4 FY26 8-K on the SEC site to see what I’m seeing here.
If you believe Blackwell Ultra pricing holds, Rubin lands on time, and a Chinese reopening adds upside, the 75% bogey is a floor. If you think hyperscaler leverage is finally turning, or power siting starts gating demand, that same 75% becomes a ceiling. Tonight tells us which world we are in.
Top Gaining Stocks
Top Losing Stocks