Nvidia posted a strong Q1 and marginally missed estimates in Q2 due to Blackwell revenue that exceeded expectations. The larger Blackwell systems are in full production and are shipping in volume now, which sets up a strong second half of the year.
According to management commentary, the ramp is happening very quickly: “On average, major hyperscalers are each deploying nearly 1,000 NVL72 racks or 72,000 Blackwell GPUs per week and are on track to further ramp output this quarter.” The rough math here implies hyperscalers are deploying $3 billion every week right now since each rack goes for $3 million. Furthermore, the run rate of this comment implies data center revenue will be above and beyond analyst consensus for Q2, Q3 and Q4 – thus, either analyst consensus comes up or these systems will become further supply constrained somewhere down the line and analysts are being conservative for now.
Among the many reasons that Blackwell is an improvement compared to the Hopper architecture, management focused on inference stating: “Compared to Hopper, Grace Blackwell is some 40 times higher speed and throughput compared.”
The loss of China revenue in Q1 was $2.5 billion yet inventory charges were higher at $4.5 billion from orders placed prior to April 9. For Q2, the loss of China revenue was $8 billion –or about $2 to $3B higher than the typical $5.5B in China revenue. The readthrough is that Blackwell came in $2-$3B above analyst expectations to absorb that impact from China, since guidance marginally missed.
You can view an interview on Fox where I discussed the puts and takes going into the earnings report plus the new price target I/O Fund published here. If you’re brief on time, the takeaway is that Blackwell has enough ammo to push the stock into the mid-to-high $200s or a $6+ trillion market cap. I discuss this and more below.
Slight Revenue Beat in Q1, Marginal Miss in Q2
Nvidia reported a slight revenue beat in Q1, reporting 69.2% YoY growth to $44.06 billion in revenue, just ahead of the $43.25 billion consensus.
For Q2, Nvidia guided $45 billion, +/- 2%, representing a deceleration to 49.8% YoY growth and a marginal miss at midpoint versus consensus at $45.66 billion. At the low end of the guide, revenue growth would be up only 2.3% sequentially, reflecting how large of an impact the H20 ban is having on growth.
Nvidia also quantified more of the H20 impact, providing details on the revenue impact to both Q1 and Q2 – in total, both quarters are seeing a combined impact of just over $15 billion. Nvidia added that the inventory charge of $4.54 billion was less than the $5.5 billion anticipated as it was able to re-use certain materials.
For Q1, Nvidia said that it recorded $4.6 billion in H20 revenue, or about 10.4% of revenue, while it was unable to ship an additional $2.5 billion of H20 revenue due to export restriction. In total, this implies $7.1 billion in H20 revenue in Q1. This would represent around 16% of total revenue or 18.2% of data center revenue in the quarter.
For Q2, Nvidia said that its guidance reflects the loss of approximately $8 billion in H20 sales, implying nearly 13% QoQ growth was expected to fill extremely high Chinese demand for the chip. However, based off management’s commentary, its $45 billion guide for Q2 suggest that other Blackwell SKUs are ramping rapidly and filling much of the H20 void.
Key Segments
Data Center
Nvidia reported 73.3% growth in data center revenue to $39.11 billion in Q1, marginally higher than analyst expectations from Visible Alpha of $39.08 billion. This marked the end of Nvidia’s seven-quarter streak of $1 billion-plus beats in the segment – based on the Visible Alpha estimate, Nvidia beat by just $33 million, its lowest in the past nine quarters.
Compute revenue rose 76% YoY but just 5% QoQ to $34.16 billion, impacted by the H20 ban, while Networking revenue rebounded swiftly, rising 56% YoY and 65% QoQ to $4.96 billion. Nvidia said Networking’s performance was “driven by the growth of NVLink compute fabric in our GB200 systems and continued adoption of Ethernet for AI solutions at cloud service providers and consumer internet companies.”
In contrast to the prior two quarters, Nvidia did not give a number for Blackwell revenue in the quarter, stating only that its Blackwell ramp expanded to all customer categories and that large CSPs remained its largest customers at just under 50% of data center revenue.
Nvidia also said that its hyperscaler customers “are each deploying nearly 1,000 NVL 72 racks or 72,000 Blackwell GPUs per week,” with this output level on track to ramp further this quarter.
- Gaming revenue rebounded sharply, rising 48% QoQ and accelerating 53 points sequentially to 42% YoY with revenue of $3.76 billion in Q1. Nvidia said that this was driven by its Blackwell architecture and the fastest ramp in company history.
- Automotive revenue rose 72% YoY but declined (1%) QoQ to $567 million.
- Pro Viz revenue rose 19% YoY and was approximately flat QoQ at $509 million.
- OEM and Other revenue rose 42% YoY but declined (12%) QoQ to $111 million.
Margins Take Large Hit from H20, Though Q2 Points to Swift Rebound
Nvidia’s margins took a rather large hit from the H20-related inventory write-down, with gross margin and operating margins contracting significantly. However, management’s guidance for Q2 points to a rapid recovery in margins as Blackwell ramps, likely aided by its pricing power.
- GAAP gross margin was 60.5% and adjusted gross margin was 61%, around 10 points below management’s initial guidance for 70.6% and 71% due to the $4.54 billion charge related to the H20 ban. Management noted that excluding the charges associated with the ban, adjusted gross margin would’ve been 71.3%, at the upper end of the guided range of 71% +/- 0.5%.
- For Q2, management guided for 71.8% GAAP gross margins and 72% adjusted gross margins, a rebound of approx. 11 points sequentially.
- GAAP operating margin was 49.1%, well below guidance for 58.5% and a sequential contraction of 12 points. Adjusted gross margin was 52.8%, nearly 10 points below the guide for 62.6% and a sequential contraction of more than 12 points.
- For Q2, management’s guidance implies operating margins will rebound with gross margins, projecting approximately a 10 point sequential expansion to a 59.1% GAAP and 63.1% adjusted operating margin.
- GAAP net margin was 42.6%, while adjusted net margin was 45.2%. The broad-based margin recovery in Q2 is expected to mostly transfer through to the bottom line, with management guiding for a 7.6 point recovery to a 50.2% GAAP net margin.
EPS Beats, Growth Expected to Rebound
Nvidia reported a slight EPS beat despite the margin contractions, with adjusted EPS of $0.81 coming in ahead of the $0.75 estimate. GAAP EPS of $0.76 missed estimates for $0.81.
Adjusted EPS growth slowed quite dramatically, decelerating more than 38 points sequentially, in part due to the H20 ban; Nvidia noted that excluding the ban, adjusted EPS would be $0.96. This would represent YoY growth of 57.4% versus the 32.8% reported.
Looking ahead, adjusted EPS growth is expected to rebound and remain in the low to mid-40% range as margins recover. However, given that Q1’s EPS excluding the ban showed growth in the high-50% range, estimates may move higher as Q2’s margin outlook shows almost no persisting impact.
Cash Flows and Balance Sheet
Cash flows were surprisingly strong as Nvidia’s cash flow margins expanded approximately 20 points sequentially, while it added more than $10 billion in cash to its balance sheet.
- Operating cash flow was $27.41 billion, up nearly 79% YoY on higher revenue, timing of its cash collections, and lower cash taxes. OCF margin was 62.2%, up 20 points QoQ and more than 3 points YoY. Nvidia said it expects a substantial increase in cash taxes in Q2, which will weigh on OCF.
- Free cash flow was $26.14 billion, up 75% YoY. FCF margin was 59.3%, up nearly 20 points QoQ and just 2 points YoY.
- Inventories were $11.33 billion, rising more than 12% QoQ. However, days sales of inventory decreased from 86 days in Q4 to 59 days in Q1, due to the sharp increase in COGS from the H20 inventory charges.
- Accounts receivable were $22.1 billion, declining just over (4%) QoQ. Days sales outstanding decreased sequentially from 53 days to 46 days due to improved shipment linearity (shipments more evenly distributed throughout the quarter) and timing of collections.
- Cash and equivalents rose more than $10 billion sequentially to $53.69 billion, despite Nvidia returning more than $14.3 billion to shareholders in the quarter with $14.1 billion in share repurchases.
- Debt remained steady at $8.46 billion.
Earnings Q&A:
Inference Demand is Skyrocketing
In the opening remarks, management stated they are seeing “a sharp jump in inference demand.” Our firm recently covered the 5X increase in tokens quoted by Microsoft to 50T tokens per month and 100T per quarter stated in their most recent earnings report. In that analysis, we pointed toward up to $18 billion in annualized revenue for API usage in high-end models. Google recently stated at their I/O Developer event they are processing 450T tokens per month up 50X from a year ago.
In the opening remarks the following was shared about the NVL72 inferencing capabilities: “Inference serving startups are now serving models using B200, tripling their token generation rate and corresponding revenues for high-value reasoning models such as DeepSeek-R1 as reported by artificial analysis. NVIDIA Dynamo on Blackwell NVL72 turbocharges AI inference throughput by 30x for the new reasoning models, sweeping the industry […] In the latest MLPerf Inference results, we submitted our first results using GB200 NVL72, delivering up to 30x higher inference throughput compared to our 8-GPU H200 submission on the challenging Llama 3.1 benchmark.”
Later, in the Q&A session, Jensen Huang stated inference is reaching an inflection point, stating “we've reached an extraordinary milestone with AIs that are reasoning, are thinking, what people call inference time scaling. Of course, it created a whole new — we've entered an era where inference is going to be a significant part of the compute workload.”
It was then re-emphasized again in the Q&A with Huang stating:
“Yeah, thanks. Thanks, Ben. I would say compared to the beginning of the year, compared to GTC timeframe, there are four positive surprises. The first positive surprise is the step function demand increase of reasoning AI, I think it is fairly clear now that AI is going through an exponential growth, and reasoning AI really busted through [… So, number one is inference reasoning and the exponential growth there, demand growth.”
Note on Valuation:
I've written a substantial amount on Blackwell — perhaps the most important being my $10T prediction came out about two months after Blackwell was announced at GTC. If you read-between-the-lines on our new price target, then I’m saying Nvidia can reach more thn a $6T market cap as soon as next year.
This relies on two assumptions. The first is that we see Nvidia reach the valuation it saw in 2024 when the forward PE Ratio was at 50 forward a handful of times. That implies a move of up to 51% as it stands today.
On the top line, NVDA has traded as high as 28 forward PS, implying room of up to 75%.
The second assumption is that Nvidia beats estimates, forcing the valuation higher. We’ve seen a small glimpse of this in Q2 with $2 to $3B in Blackwell revenue absorbing China losses. However, I’ve been crystal clear that it’s the August call and November call that will be fireworks. I expect to see Nvidia grand slam beats in these quarters especially and/or analyst consensus moving up into those quarters, creating more room in the valuation then the 50% to 75% we see currently.
Conclusion:
The marginal miss in Q2 would have been more pronounced if Blackwell were not ramping. Pay attention, as this is the cyclical bottom for Nvidia as the Hopper generation fades out and yet its earnings reports have been unscathed. The future is bright and the fireworks are locked and loaded for H2.
Please note: The I/O Fund conducts research and draws conclusions for the company’s portfolio. We then share that information with our readers and offer real-time trade notifications. This is not a guarantee of a stock’s performance and it is not financial advice. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis. Beth Kindig and the I/O Fund own shares in NVDA at the time of writing and may own stocks pictured in the charts.
Recommended Reading: