NVIDIA Corp (NVDA) (Q2 2027) Earnings Call Highlights: Record Revenue of $96 Billion, Fiscal ...

Yahoo Finance ·

NVIDIA Corp (NVDA) (Q2 2027) Earnings Call Highlights: Record Revenue of $96 Billion, Fiscal ... GuruFocus News Thu, August 27, 2026 at 1:01 AM EDT 9 min read NVDA This article first appeared on GuruFocus . Revenue: Total revenue of $96 billion, more than doubled year-over-year. Data Center Revenue: $89 billion, up 18% quarter-over-quarter. Hyperscale Revenue: $49 billion, up 13% sequentially. ACIE Revenue: $40 billion, up 25% sequentially and 138% year-over-year. Gross Margin: GAAP and non-GAAP gross margins were both 75%, largely unchanged from last quarter. Operating Expenses: GAAP and non-GAAP operating expenses were up 10% and 11% sequentially. Inventory: Increased to $32 billion, preparing for the Vera Rubin launch. Days Sales Outstanding: Increased to 60 days, reflecting extended payment terms for large purchases. Shareholder Returns: Returned a record $26 billion to shareholders, including $20 billion through share repurchases and $6 billion through dividends. Networking Revenue: Grew 18% sequentially, with Spectrum-X Ethernet growing 2.6 times year-over-year. Grace CPU Revenue: Trailing 12-month revenue exceeded $5 billion. On-Prem Revenue (Automotive): Trailing 12-month revenue reached $8 billion. On-Prem Revenue (Financial Services, Manufacturing, Health Care): Combined $7 billion in revenue. Sovereign AI Revenue: Grew 35% sequentially and more than tripled year-over-year in Q2. Q3 Revenue Outlook: Expected to be $108 billion, plus or minus 2%. Q3 Gross Margin Outlook: GAAP and non-GAAP gross margins expected to be 74% plus or minus 50 basis points. Q4 Gross Margin Outlook: Expected to bottom in the 71% to 72% range. Fiscal 2028 Gross Margin Outlook: Expected to settle at 72% to 73%. Q3 Operating Expenses Outlook: GAAP and non-GAAP operating expenses expected to be approximately $9.2 billion and $9.0 billion, respectively. Warning! GuruFocus has detected 4 Warning Signs with NVDA. Is NVDA fairly valued? Test your thesis with our free DCF calculator. For the complete transcript of the earnings call, please refer to the full earnings call transcript . Record revenue of $96 billion, more than doubling year-over-year, with growth accelerating for the fourth consecutive quarter. Strong demand across all segments, including hyperscalers, AI labs, enterprises, and sovereign customers, with ACIE revenue growing 138% year-over-year. Expansion of partnership with AWS, deploying an additional 2 million GPUs and adopting NVIDIA's full physical AI stack for warehouse robots. Vera Rubin, the next-generation platform, is in production with purchase orders from all major customers, expected to be the fastest product ramp in NVIDIA's history. Networking business achieved record revenue, with Spectrum-X Ethernet growing 2.6 times year-over-year, positioning NVIDIA as a leading network company. New revenue-sharing model with NeoClouds creates recurring revenue streams, potentially driving billions in revenue over the medium to long term. Investments in Frontier AI labs and partnerships with infrastructure capital providers to raise over $500 billion, supporting long-term growth. Gross margin remains strong at 75%, with expectations to settle at 72-73% in fiscal 2028 despite memory cost pressures. Record shareholder returns of $26 billion in Q2, with a commitment to return excess free cash flow. Guidance for fiscal 2028 revenue growth of approximately 70%, driven by robust demand and supply chain expansion. Supply constraints limit growth to 70% for fiscal 2028, despite demand being significantly higher, potentially disappointing customers. Gross margins are expected to decline to 71-72% in Q4 due to extreme memory pricing conditions, with a reset in expectations. Memory scarcity is a significant challenge, with prices expected to rise further into next year, impacting profitability. China data center revenue is excluded from forward outlook due to geopolitical uncertainty, with Hopper shipments to China being dilutive to gross margins. Inventory increased to $32 billion, and days sales outstanding rose to 60 days, reflecting extended payment terms and potential cash flow impacts. Operating expenses are expected to grow in the low 50s for the full year, driven by portfolio broadening and AI tool usage. The company faces challenges in securing land, power, and shell capacity, which are critical for scaling AI infrastructure. Investments in Frontier AI labs and credit enhancements carry risks, with some labeling it as circular financing, though NVIDIA sees it differently. The company's growth is heavily dependent on the AI build-out, and any slowdown in AI adoption could impact future results. Despite strong demand, the company is unable to fully meet it, leading to potential lost revenue opportunities. Q : What gives you the confidence to guide a full year out to 70% growth for fiscal 2028, and what is the key constraint separating that from the higher demand you see? A : Jensen Huang (President and CEO): AI has become useful, and agentic AI consumes 15 to 100 times more compute than human-prompted use. We are the only company offering a full-stack AI factory platform, and we see demand from sovereign AI, neo clouds, AI startups, and enterprisesrepresenting about half our businessgrowing 100% a year. We have greater visibility upstream and downstream, including land, power, and shell. While demand is much greater than 70%, our supply allows us to confidently deliver that number, and we will work to close the gap. Q : Can you discuss the evolving inference workloads and how you see your market share evolving, especially with the growing value of the TAM per generation and the inclusion of Groq 3 LPX? A : Jensen Huang (President and CEO): The AI life cycle is more complex, spanning data preparation, pretraining, post-training, and agentic inference. Our NVLink72 rack-scale architecture creates one fungible system that transitions across all phases, preserving customer investment. Our revenue opportunity per gigawatt has grown from $18 billion with Hopper to $25 billion with Grace Blackwell and $40 billion with Vera Rubin, which includes CPU, multiple networking types, and Groq. This full-stack approach captures more of the data center TAM. We are excited about Groq 3 LPX for high-interactivity services, but the vast majority of data centers will use Vera Rubin NVLink72. Q : Can you break down the contributors to the 70% growth in fiscal 2028, and what would the growth be if it were not supply-constrained? A : Jensen Huang (President and CEO): The unconstrained number would be significantly higher than 70%. The growth is driven by both hyperscalers, who see backlogs of $2 trillion and profitable compute, and the less visible ACIE segment (enterprise, neo clouds, sovereign), which is growing rapidly. Customers want to race to the next generation because each generation increases revenue per gigawatt and productivity. We are going through a platform shift affecting every industry, and everyone needs to build AI infrastructure. Q : The sum of your ecosystem investments appears to be around $500 billion. Is that the total, and how are you balancing investing in Frontier labs like OpenAI and Anthropic, which are also designing their own custom chips? A : Jensen Huang (President and CEO): We are building a platform, not a single chip. NVIDIA's AI factory platform spans the entire AI life cycle and runs in every cloud. These companies will need to scale globally and will run on NVIDIA. I have 100% confidence our technology will remain extraordinary for them. Investing in these AI labs is a once-in-a-generation opportunity, and my only regret is not investing more and sooner. They will be some of the most consequential technology companies in history, and they will be customers and partners for a long time. Colette Kress (CFO) added that the supply commitments are essential for raising Vera Rubin and give confidence in revenue growth. Q : Do you see the rise of open-source models as good or negative for NVIDIA, given that much end demand is driven by frontier model companies? A : Jensen Huang (President and CEO): The world needs both closed and open models, and both are skyrocketing in use. Nearly all open models run on NVIDIA because our architecture is the most fungible and has the largest footprint. Open models are vital for enterprises and countries to build proprietary AI, and they have reached frontier levels. Both are extraordinarily successful and simultaneously drive our sales. We are the only platform that runs every frontier model, whether closed or open, and we are delighted by any model's success. Q : With developments in recursive self-improvement and AGI, what happens to industry demand, and what does it mean for NVIDIA? A : Jensen Huang (President and CEO): Demand will inflect further. Most AI is becoming agentic, and in the future, every company will have millions of agents running continuously in the background. This is a form of coarse-grained self-improvement. The most important things are that AI is doing productive work, generating profitable tokens, and that more compute leads to more profitable tokens. This is why everyone is leaning in, and we are in a phase where compute is the limiting factor. Q : Can you rank order the most acute supply constraints, such as power, shell, DRAM, or foundry availability? A : Jensen Huang (President and CEO): Our entire supply chain is challenged and running flat out. More capacity is coming online every day, and we are working on yield improvements with all suppliers. We currently have supply for about 70% growth, but demand is much higher. We are being transparent with our supply chain about our needs, and we will work hard to close the gap to avoid disappointing customers. Q : As you think about the path beyond Vera Rubin, should we conceptualize the revenue per gigawatt scaling from $40 billion to $60 billion or $80 billion, and how does that impact your ability to scale capacity? A : Jensen Huang (President and CEO): Our goal is to put as much compute as possible on a plot of land. We started at roughly $3-5 billion per gigawatt with general-purpose computing, grew to $18 billion with Hopper, $25 billion with Grace Blackwell, and $40 billion with Vera Rubin. This will continue to increase. This is fantastic for customers as long as productivity, durability, and fungibility grow. Return on invested capital for these data centers is now less than a year, which speaks to the productivity and rentability of NVIDIA's technology.

DYAX Investor Sentiment

Bullish (Long) 55% · Bearish (Short) 45%

510 participants

Related News

원문 보기 — Yahoo Finance