Nvidia's Q2 FY2026 Results: Record-Breaking $46.7B Revenue Driven by AI Boom and Gaming Success

Nvidia posts $46 billion revenue in another record quarter -- Data center and gaming GPU sales break records

Nvidia on Wednesday published its financial results for the second quarter of its fiscal year 2026, posting record revenue of $46.743 billion and marking its best quarter ever. During the quarter the company managed to tangibly increase sales across all product categories both sequentially and year-over-year. Yet, AI GPUs accounted for nearly 88% of its revenue, asserting Nvidia's dominance in the AI hardware segment. But perhaps the biggest thing that will please Nvidia's investors is its outlook for the third quarter. For the second quarter of fiscal year 2026, Nvidia posted GAAP revenue of $46.7 billion, up 56% year-over-year (YoY) and 6% quarter-over-quarter (QoQ). The company's net income surged to $26.4 billion, a 59% increase from the same period last year, despite the fact that the company's gross margin dropped 72.4% from 75.1% a year ago, to some degree due to a $180 million charge related to banned sales of its H20 AI accelerators to Chinese entities. However, Nvidia's gross margin increased sharply from 60.5% in the prior quarter. Data center AI platforms is by far Nvidia's biggest business with revenue reaching $41.1 billion, up 56% year-over-year and a 5% sequential gain. Large cloud service providers (CSPs) accounted for approximately 50% of total data center revenue. Compute revenue was $33.8 billion, up 50% YoY, but declined 1% sequentially, primarily due to a $4.0 billion drop in H20 GPU sales to Chinese entities. Networking revenue surged to $7.3 billion, reflecting 98% YoY growth and 46% QoQ growth, driven by strong adoption of Nvidia's GB200 NVL72 and GB300 NVL72 systems that exclusively rely on NVLink fabric, Ethernet, and Infiniband solutions developed by Nvidia. Perhaps the biggest highlight of Nvidia's data center business during the quarter was the ramp-up of Blackwell GPUs, including Blackwell Ultra, which grew 17% quarter-over-quarter and contributed revenue across all customer segments. Meanwhile, Nvidia still sold plenty of Hopper-based processors and HGX servers to various parties that still deploy such machines. "Blackwell is the AI platform the world has been waiting for, delivering an exceptional generational leap -- production of Blackwell Ultra is ramping at full speed, and demand is extraordinary," said Jensen Huang, co-founder and chief executive of Nvidia. "Nvidia NVLink rack-scale computing is revolutionary, arriving just in time as reasoning AI models drive orders-of-magnitude increases in training and inference performance. The AI race is on, and Blackwell is the platform at its center." As Nvidia sells a great deal of AI GPUs for data centers, some people outside of the PC gaming world probably begin to forget that the company's origins are indeed in PC gaming. This business segment not only continues to bring Nvidia a lot of money, but it thrives. In the second quarter of fiscal 2026, Nvidia sold a whopping $4.3 billion worth of gaming GPUs, reflecting a 49% increase year-over-year and a 14% rise from the previous quarter. The company attributed strong performance to robust demand and improved supply of its Blackwell-based gaming GPUs, which continued to gain traction in the consumer market. Another reason is of course the launch of highly-successful performance-mainstream GeForce RTX 5060 and RTX 5060 Ti graphics cards. Additionally, the OEM and Other segment posted $173 million in revenue, up 97% year-over-year and 56% sequentially, which marks increasing demand even for Nvidia's low-end GPUs. Nvidia's professional visualization (ProViz) business generated $601 million in revenue in Q2 FY2026, reflecting a 32% YoY increase and an 18% QoQ rise. What is a little bit unexpected, is that Nvidia claims that the growth was primarily driven by the accelerated adoption of Blackwell-based notebook GPUs by 'AI-powered workflows, real-time graphics rendering, and data simulation,' which somewhat implies that not all ProViz GPUs were set to be used for graphics. Last but not least, Nvidia's Automotive segment achieved $586 million in revenue in Q2 FY2026, its best quarter ever. As adoption of Nvidia's autonomous driving and robotics platforms across global OEMs increase, sales of appropriate products increased 69% compared to Q2 FY2025 and 3% compared to the previous quarter. Nvidia projects revenue of approximately $54 billion ± 2% for the third quarter of its ongoing fiscal year. The outlook does not include any H20 shipments to China during the quarter, which is driven by the ongoing political uncertainity due to tensions between the U.S. and PRC. Nvidia anticipates GAAP gross margin of 73.3% as the company set to benefit from product mix optimization and further Blackwell volume ramp. The company reaffirmed its goal to exit the fiscal year with a non-GAAP gross margin in the mid-70% range.

NVIDIA is (really) profiting from the AI boom

NVIDIA has revealed that its revenue for the second quarter ending on July 27 rose 56 percent compared to the same period last year, and that's without shipping any H20 chips to China. It reported a revenue of $46.7 billion and a net income of $26.4 billion. As CNBC has noted, the company previously said that H20 shipments to China could have added $8 billion in sales to its second quarter figures. Instead, the company was only able to release $180 million worth of the chips to a customer outside of China. NVIDIA clearly still did very well in the second quarter, however, just like in the previous ones. This is the ninth straight quarter, ever since AI skyrocketed in popularity back in 2023, wherein the company's year-on-year revenue increase has exceeded 50 percent. The company designed the H20 specifically for the Chinese market, but the US government blocked its shipment earlier this year over concerns it could aid the nation's military. In mid-July, the government told NVIDIA that it will be able to ship the H20 to China again, but Chinese authorities discouraged local tech firms from using the chips. Now, the company is believed to be developing a new chip for China that's more advanced than the H20 and is based on the company's Blackwell architecture. Both the H20 and this upcoming more advanced chip could lead to another big leap in revenue and sales for the company. Speaking of Blackwell, NVIDIA has also announced that sales for its Blackwell-based chips rose 17 percent from the first quarter. The new product line, which is designed for advanced AI, reached $27 billion in sales and represented 50 percent of the company's data center revenue.

Nvidia makes almost ten times more from AI than gaming but 'margins are already eroding as rivals push into the space'

Nvidia has caught the wave of the AI boom thanks to its selection of GPUs and data centers, and that appears to be paying off big time, but with any boom comes competition. Some analysts worry Nvidia can't keep this level of profit margin long term. Nvidia just had its Q2 2025 earnings call. In it, Nvidia reports a total revenue of $46.7 billion, which is 6% higher than last quarter and 56% higher than this time last year. Data center revenue for its latest Blackwell architecture grew by 17% too. Nvidia has been all in on AI for some time now. Just a few months ago, Nvidia CEO Jensen Huang called Trump "America's unique advantage" because his ramping up of energy meant more for AI factories. Last year, he claimed future customers might want three computers for AI: one to create AI, one to simulate AI, and one to run AI. That's, well, a lot of AI. This AI-led approach comes through in the earnings call, which is largely focused on both it and Nvidia's prospects in China. In the call, Nvidia's executive vice president and chief financial officer, Colette Kress, reports, "We are focusing on meeting the soaring global demand. This growth is fueled by capital expenditures from the cloud to enterprises, which are on track to invest $600B in data center infrastructure and compute this calendar year alone." Gaming got a comparably small amount of the call, with Nvidia reporting a "49% jump year on year" with a revenue of $4.3 billion. This growth is attributed to continued sales from RTX Blackwell, which launched early this year, as well as the recent launch of the RTX 5060. Seeking Alpha reports that the total revenue is slightly below some estimates, and share prices for the computing company dipped by 2.7% afterwards. These revenue figures do not represent any income made from the H20 chip, which was originally intended to be sold as an AI chip in China. This was first banned by the US, then the US allowed it, and China banned it instead. Despite major growth, there are a few worries behind this AI expansion. As reported by The National News, deVere Group's chief executive Nigel Green says, "The shift is that the company's moved from hyper growth to high growth", and he says, "This matters because markets have priced Nvidia as if its rate of expansion could continue indefinitely, and that level of outperformance was never sustainable." According to Capacity, Green also warned that a large chunk of Nvidia's revenue is reliant on just two customers (believed to be Microsoft and Meta). This high level of concentration can be a worry, and "Markets don't like that level of exposure." Green argues, "Nvidia has been the undisputed champion of the AI boom, but margins are already eroding as rivals push into the space. Growth in AI demand is relentless, but Nvidia's share of that growth is being squeezed. The story is no longer about one company dominating; it's about an entire industry expanding at pace." Operating expenses are higher than last year, but so too is operating income. Gross margin has slipped from 75.1% this time last year to 72.4% and diluted earnings per share have grown from $0.67 to $1.08. So shareholders can't be too annoyed about the performance. There's a gold rush going on right now with AI development, one that many companies are in on, and for the likes of Meta and Microsoft, Nvidia is selling the shovels. These reports wonder how long Nvidia can be the sole provider here before someone else comes along with more for cheaper. Whether this is a sustainable industry or a bubble is anyone's guess, but Nvidia is certainly making a lot of money right now. And it's not primarily from gamers like me and you.

The RTX 50-series has delivered a record-breaking $4.28B in gaming revenue for Nvidia... no matter what you think about VRAM levels, launch pricing, and availability

Although AI is absolutely the main focus of Nvidia right now, with Jen-Hsun now referring to it as "an AI infrastructure company", the green team hasn't forgotten about gaming. The launch of the RTX 50-series graphics cards this year has certainly been a boon for the company and accounts for a huge increase in gaming revenue. The second quarter of Nvidia's fiscal year 2026 (or 2025 if you're not Nvidia) has seen a record $4.28 billion of revenue, which is 49% higher than the same period last year. The start of 2024 didn't see any major launches for GPUs, with the RTX 40-series cards rolling out from the latter half of 2022 to the middle of 2023. For a longer term picture of what this stat means, Nvidia reported a gaming revenue of $1.83 billion in the final quarter of 2022 (Nvidia's Fiscal 2023. Yeah, it's a confusing system). This same quarter saw the launch of RTX 40-series laptops, as well as the RTX 4070 Ti. Prior to that, in Q3 of 2022, we'd only just seen the launch of the RTX 4090, and Nvidia reported a gaming revenue of $1.57 billion. Nvidia followed up with gaming revenue of $2.24 billion in Q1 fiscal 2024, as well as the launch of the RTX 4070. Finally, Q2 of fiscal 2024 saw a revenue of $2.49 billion, with the launch of both the RTX 4060 and the RTX 4060 Ti. Where the RTX 30-series saw plenty of extra demand thanks to cryptominers, that was less viable, and less profitable by the time the RTX 40-series came around. By the start of 2022, crypto GPU sales had dropped off significantly. Though AI development had certainly long begun by this point, it was far less relevant to Nvidia's business model by this point. How times have changed. The RTX 50-series launch, however, has seen the highest revenue for its gaming segment that Nvidia has ever had. AI still outweighs it from a data centre/enterprise perspective, with it almost being ten times the revenue in the latest earnings report, though. It is worth noting, however, that where this section in Nvidia's reports previously just said 'Gaming', it has said 'Gaming and AI PC' since the Q1 report for its fiscal year of 2024 instead. Nvidia can't separately account for sales made on gaming tech that is not used for gaming, so obviously has to pool the two together. Those looking to run AI servers or rigs (like individual professionals, schools, etc) will likely buy the consumer-grade cards and put them into their own systems. In the gaming and AI PC section of the report, Nvidia cites its partnership with OpenAI to launch open-weight models for use in local AI on RTX GPUs, and local LLM usage is certainly growing. It is worth noting that all of the RTX 50-series cards launched very close together, far closer than previous generations, and stock even finally appears to be stabilising. The first few months seemed like a desert in terms of graphics cards you could buy for a reasonable price, but GPUs, almost across the board, appear to be available for around MSRP. We've even seen the RTX 5090 drop down to its MSRP recently, and that's the first time that's happened in the US since launch. Nvidia has certainly managed to eventually deliver a good level of stock after launch, even if the launch itself was a frustrating mess.

Nvidia beats its own outlook with $46.7bn Q2 revenue

Jensen Huang, co-founder and CEO, Nvidia. Image: Fortune Global Forum/ Flickr (CC BY 2.0) The company did not sell any H20 processors to China-based customers it said. Nvidia made $46.7bn in revenue in the second quarter past, up 56pc year-on-year, beating its own outlook by a around $1.5bn. However, while Nvidia's year-on-year revenue has exceeded 50pc every quarter since mid 2023, Q2 FY 2026 marked the company's slowest growth in that period. Last quarter, the company's revenue growth neared 70pc, bringing in just more than $44bn. While revenue is growing at a considerable pace, Nvidia stocks fell slightly following the earnings call after the data centre unit made marginally less than what estimates expected. The data centre unit, which houses its AI offerings, brought in $41.1bn, falling short of estimates by around $240m. The company said that Blackwell sales rose by 17pc from the first quarter. In May, its new product line had reached $27bn in sales, the company said, accounting for around 70pc of data centre revenue. Nvidia has, for a number of years, found itself in the midst of the US-China advanced chip conflict. The company estimated a loss of around $8bn this quarter if it didn't receive an export license to sell its H20 processors to China, And while it didn't sell any H20 chips to China-based customers in the second quarter, the company said it benefitted from the release of $180m worth of H20 inventory to a customer outside China. After weeks of negotiations and multiple visits to the White House, Nvidia CEO and co-founder Jensen Huang managed to land an agreement with the US Trump administration to export their AI chips to China in a never-before-seen 15pc revenue sharing deal with the government. Though that has not yet been finalised, according to the company chief financial office Colette Kress. "[US government] officials have expressed an expectation that the USG will receive 15pc of the revenue generated from licensed H20 sales." "But to date, the USG has not published a regulation codifying such a requirement." Estimates suggest that without the chip curbs in place, Nvidia would have generated around $23bn in H20 chip sales to China in 2025. Although, demand for the company's AI chips are expected to drop the longer it stays out of the country, analysts suggested. Outside of the data centre unit, Nvidia's gaming division brought in $4.3bn in sales, up 49pc from a year ago. During the quarter, the company said that its GPUs intended for gaming would be able to run certain OpenAI models on personal computers. In July, Nvidia made history by becoming the first company to briefly touch a $4trn valuation, highlighting just how important of a part it plays in the global AI race. Don't miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic's digest of need-to-know sci-tech news. Jensen Huang, co-founder and CEO, Nvidia. Image: Fortune Global Forum/ Flickr (CC BY 2.0)

NVIDIA posts $46.7 billion revenue for Q2 2025, up 56% from 2024 thanks to Blackwell AI GPUs

TL;DR: NVIDIA reported a record $46.7 billion revenue in Q2 2025, up 56% year-over-year, driven by strong AI GPU sales and data center growth. The company projects $54 billion for Q3, fueled by Blackwell AI platform adoption and expanding enterprise, gaming, automotive, and robotics markets worldwide. NVIDIA has just posted its Q2 2025 financial report, with $46.7 billion in revenue over the three-month period, breaking its previous sales record from Q1 2025 with $44.1 billion. The company posted a press release on its website noting that its quarterly earnings are up 56% from a year ago, as well as no H20 AI GPUs sold to China-based customers in Q2 2025. NVIDIA benefited from a $180 million release of previously reserved H20 chips sitting in inventory, from around $650 million in unrestricted H20 sales to customers outside of China. NVIDIA expects the next quarter to be even bigger again, once again breaking (even the new Q2 record) in Q3 2025, with the AI GPU leader projecting $54 billion in revenue. That would mean $44B for Q1, $46.7B for Q2, and a big step up to $54B in Q3... GB200 is mass-produced now, and the new Blackwell Ultra GB300 AI servers are now hitting the market... Q3 is going to be massive for NVIDIA. Jensen Huang, founder and CEO of NVIDIA, said: "Blackwell is the AI platform the world has been waiting for, delivering an exceptional generational leap - production of Blackwell Ultra is ramping at full speed, and demand is extraordinary. NVIDIA NVLink rack-scale computing is revolutionary, arriving just in time as reasoning AI models drive orders-of-magnitude increases in training and inference performance. The AI race is on, and Blackwell is the platform at its center". NVIDIA notes that its new RTX PRO 6000 Blackwell Server Edition GPU is powering the world's most popular enterprise servers, with companies including Disney, Foxconn, Hitachi, Hyundai, and even TSMC being the first to use the new servers. The company also highlights more data center progress since Q1 2025, collaborating with partners across the planet to build and accelerating advanced AI supercomputers, with Doudna in the US, JUPITER and Blue Lion in Germany, Isambard in the UK, and FugakuNEXT in Japan. NVIDIA says that it was recently revealed that NVIDIA was working with nations across Europe including France, Germany, Italy, Spain, and the UK, as well as technology industry leaders to build NVIDIA Blackwell AI infrastructure, including the world's first industrial AI cloud for European manufacturers, to fuel the region's next industrial transformation. On the gaming side of things, Q2 2025 revenues for its Gaming division (GeForce, etc) reached $4.3 billion, which is a 14% increase over Q1 2025, and a 49% increase over a year ago. This shows that a lot of gamers are upgrading, system integrators are flooding their systems with GeForce RTX 50 series GPUs, especially up 49% from 12 months ago. NVIDIA included highlights for its Automotive and Robotics divisions, with the announcement of its new NVIDIA Jetson AGX Thor developer kit and production modules, with powerful new AI supercomputers designed to power millions of robots across multiple industries.

GeForce RTX 5060 is NVIDIA's 'fastest-ramping x60-class GPU ever'

TL;DR: NVIDIA reported $46.7 billion revenue in fiscal Q2 2026, driven by a 56% Data Center increase from Blackwell hardware. Gaming revenue rose 49% year-over-year, boosted by the successful GeForce RTX 5060 launch. DLSS 4 expanded to 175+ games, and RTX Blackwell debuted on GeForce NOW cloud gaming. NVIDIA posted its latest financial results for fiscal Q2 2026, with $46.7 billion in revenue, a 56% year-over-year increase in Data Center revenue driven by Blackwell hardware. However, it's also good news for NVIDIA's Gaming and AI PC segment that includes the company's latest RTX Blackwell generation of GeForce RTX hardware. According to the report, fiscal Q2 2026 Gaming revenue reached $4.3 billion, representing a 14% increase from Q1 2026 and a 49% rise from the same period last year. During this time, NVIDIA launched the mainstream-focused GeForce RTX 5060 GPU on May 19, which NVIDIA says has become the "fastest-ramping x60-class GPU ever" for the company. Although we're not sure if "fastest-ramping" translates to "fastest-selling," it's clear that the GeForce RTX 5060 has been a massive success in recent months. This also means that we wouldn't be surprised if it starts to make big moves up the Steam Hardware & Software Survey results for August 2025. Critically, the GeForce RTX 5060 has received a mixed response due to sticking with 8GB of fast GDDR7 memory, which many see as a bottleneck for pushing image quality and higher resolutions in some titles. Many were hoping that the latest x60-class GPU from NVIDIA would arrive with 12GB or 16GB of VRAM. This is something we noted in our review of the MSI GeForce RTX 5060 8GB GAMING OC; however, we were impressed by its 28% faster 1080p gaming performance than the previous generation's GeForce RTX 4060. Other highlights for NVIDIA's Gaming sector performance in fiscal Q2 2026 include DLSS 4 becoming available in over 175 games and apps, covering all the biggest PC releases, and the arrival of RTX Blackwell on the company's GeForce NOW cloud gaming service.

Nvidia net income jumps 59% to $26.4B in second quarter results

CEO Jensen Huang touted Blackwell as the centerpiece of the AI race, projecting $3-4T in infrastructure spending by decade's end. Nvidia, recognized as the world's most valuable entity, disclosed its latest financial performance, revealing substantial sales growth. The company's earnings statement, released Wednesday, indicated revenue of $46.7 billion, marking a 56% surge compared to the corresponding period of the previous year. This expansion was primarily propelled by its data center business, which focuses on artificial intelligence and experienced a comparable 56% year-over-year revenue increase. The corporation's net income also demonstrated considerable expansion compared to the preceding year. Nvidia documented a net income of $26.4 billion for the second quarter, demonstrating a 59% surge relative to the equivalent timeframe in the prior year. These figures highlight the company's robust financial health and its capacity to generate substantial profits amid evolving market dynamics. Data center sales accounted for $41.1 billion in revenue during the quarter, indicating sustained demand from artificial intelligence companies for advanced graphics processing units. This substantial revenue stream underscores Nvidia's pivotal role in supplying the computational resources required by organizations engaged in cutting-edge artificial intelligence research and deployment. The demand indicates that Nvidia's products are indispensable to organizations pushing the boundaries of artificial intelligence development. The most recent generation of Nvidia chips, known as Blackwell, accounted for $27 billion of the data center sales. This figure illustrates the market's strong adoption of Nvidia's advanced technology. The Blackwell series represents a significant advancement in GPU architecture, offering enhanced performance and capabilities that are highly sought after in the artificial intelligence and data analytics sectors. Nvidia CEO Jensen Huang emphasized the significance of the Blackwell platform. "Blackwell is the AI platform the world has been waiting for," Huang stated. He added, "The AI race is on, and Blackwell is the platform at its center." Huang's statement underscores Nvidia's vision of Blackwell as a central component of the ongoing advancements in artificial intelligence, positioning the company as a driver of innovation in the field. Huang projected significant investment in artificial intelligence infrastructure. He anticipates that there will be $3 to 4 trillion in AI infrastructure spending by the conclusion of the decade. During an exchange with an analyst, Huang remarked, "$3 to 4 trillion is fairly sensible for the next five years." This forecast highlights the anticipated scale of investment in artificial intelligence. Nvidia highlighted its involvement in the launch of OpenAI's open source gpt-oss models. The company reported that it processed "1.5 million tokens per second on a single Nvidia Blackwell GB200 NVL72 rack-scale system." This illustrates the capacity of Nvidia's technology to handle intensive computational workloads. The collaboration with OpenAI demonstrates Nvidia's commitment to supporting the open-source community. The financial results provided insight into Nvidia's attempts to market its products in China. Nvidia reported no sales of its H20 chip, specifically designed for the Chinese market, to customers within China during the past quarter. Despite this, Nvidia disclosed that $650 million worth of H20 chips were sold to a client located outside of China, indicating a viable international market for the product. Regulations imposed by the United States have restricted sales of advanced GPUs to Chinese customers. However, Nvidia is permitted to sell its chips to China under certain conditions. A 15% export tax is levied on these sales, payable to the U.S. Treasury. Legal scholars have described this arrangement as a potentially unconstitutional use of power. Nvidia's CFO, Colette Kress, clarified that the lack of H20 shipments was due to uncertainty surrounding the export arrangement. "While a select number of our China-based customers have received licenses over the past few weeks," Kress stated, "we have not shipped any H20 devices based on those licenses." This statement indicates that regulatory ambiguities have affected Nvidia's ability to fulfill orders in the Chinese market. The Chinese government has reportedly discouraged the use of Nvidia chips by local businesses. This has led Nvidia to reportedly halt production of the H20 chip earlier in the month. This policy shift in China has implications for Nvidia's market strategy and revenue projections in the region. Nvidia anticipates $54 billion in revenue for the third quarter. This projection does not include any H20 shipments to China. The company has noted a potential fluctuation of 2% in either direction. This outlook reflects the ongoing uncertainties in the Chinese market and their impact on Nvidia's financial forecasts.

Nvidia Q2 revenue jumps 56 per cent to $46.7 billion, but shares slip as AI chip segment shows shortfall

Nvidia announced robust financial results for the second quarter of fiscal 2026, ending July 27, 2025. The company reported revenue of $46.7 billion, up 6 per cent from the previous quarter and 56 per cent from a year ago. However, its shares declined in after-hours trading following a slight revenue shortfall in a key segment. Data Center revenue reached $41.1 billion, up 5 per cent sequentially and 56 per cent year-on-year, with its Blackwell Data Center revenue growing 17 per cent sequentially. Revenue expected for the 3rd quarter is likely to be $54.0 billion, plus or minus 2 per cent. The company has not assumed any H20 shipments to China in the outlook. CEO Jensen Huang said, "Blackwell is the AI platform the world has been waiting for, delivering an exceptional generational leap -- production of Blackwell Ultra is ramping at full speed, and demand is extraordinary." "NVIDIA NVLink rack-scale computing is revolutionary, arriving just in time as reasoning AI models drive orders-of-magnitude increases in training and inference performance. The AI race is on, and Blackwell is the platform at its center." During the first half of fiscal 2026, Nvidia returned $24.3 billion to shareholders through share repurchases and cash dividends. At the end of Q2, $14.7 billion remained under the existing repurchase authorization. On August 26, 2025, the Board approved an additional $60 billion in share repurchase authorization, with no expiration. The next quarterly cash dividend of $0.01 per share will be paid on October 2, 2025, to shareholders of record on September 11, 2025. NVIDIA expects strong growth in the third quarter of fiscal 2026, with revenue projected at around $54 billion, plus or minus 2 per cent, and no H20 shipments to China assumed in that. The company anticipates GAAP and non-GAAP gross margins of 73.3 per cent and 73.5 per cent, respectively, and expects non-GAAP margins to remain in the mid-70 per cent range by the end of the year. Operating expenses are expected to be about $5.9 billion (GAAP) and $4.2 billion (non-GAAP), with full-year expense growth in the high-30 per cent range. Other income and expenses are projected to be approximately $500 million, excluding gains or losses from certain equity securities. GAAP and non-GAAP tax rates are expected to be around 16.5 per cent, plus or minus 1 per cent, not including any one-time items. Although the quarterly earnings exceeded expectations, the company's shares fell due to concerns about a potential AI chip spending bubble and its stagnated business in China. According to CNBC, the stock has risen 35 per cent this year after nearly tripling in 2024. Nvidia is the leader in producing powerful chips used worldwide for training artificial-intelligence models. These are revolutionizing the workflow across industries and leading to innovations at lightning speed, but this key segment failed to meet market expectations in the second-quarter revenue report. Market reaction included a roughly 3-4 per cent decline in NVIDIA's shares in after-hours trading as some investors took profits after the stock had nearly doubled since April. Nvidia addressed concerns about its business in China in its latest earnings report, stating that sales of its H20 processor had not occurred in the country. "There were no H20 sales to customers based in China during the second quarter," the company said. The H20 processor is a weaker version of Nvidia's H100 chip, specifically designed for the Chinese market while adhering to current US regulations regarding the export of high-performance computer chips used in AI applications. In January, the Trump administration announced that it was seeking to tighten export controls on Nvidia H20 sales to China due to concerns regarding national security. The Trump administration initially imposed strict export controls on H20 chips to China, with licenses and fees totaling about $5.5 billion. However, in August, the administration reversed its stance, allowing sales to resume if Nvidia pays the US government 15 per cent of the revenue from those sales.

Nvidia: Q2 EPS up 54%

On Wednesday evening Artificial intelligence (AI) chip giant Nvidia reported adjusted (non-GAAP) EPS of $1.05 for Q2 2025-26, up 54% y-o-y. Its non-GAAP gross margin contracted by three points to 72.7%, with revenue growing 56% to over $46.7bn, despite no sales of H20 products to China-based customers over the period. Blackwell is the AI platform the world has been waiting for, offering an exceptional generational leap: Blackwell Ultra production is accelerating at full speed and demand is extraordinary, management said. On August 26, its board of directors authorized an additional $60bn in share repurchases, with no expiration date, adding to the $14.7bn remaining at the end of Q2. For the current quarter, Nvidia expects a non-GAAP gross margin of 73.5%, plus or minus 50bp, and revenue of $54bn, plus or minus 2% (still excluding H20 shipments to China).