A：Nvidia's second quarter fiscal 2027 was marked by record revenue, operating income, and earnings per share (EPS). Total revenue more than doubled year over year, and the surge in AI demand drove a global infrastructure build-out across various segments including hyperscalers, AI labs, AI natives, enterprises, and sovereign customers. The company expects to grow revenue by approximately 70% in fiscal 2028.

A：Q2 data center revenue increased 18% quarter over quarter to $89 billion, with strong contributions from hyperscale and AI/Enterprise (ACI) segments. Hyperscale revenue grew 13% sequentially to $49 billion, driven by sustained strength in Blackwell, while ACI revenue increased 25% sequentially and 138% year over year to $40 billion.

A：The partnership expansion between Nvidia and AWS is significant as it will see an additional deployment of 2 million GPUs starting in the current quarter through the second quarter of fiscal 29, along with Vera CPUs. This will support Nvidia's Neumann family of open models on Amazon Bedrock and SageMaker, as well as Nvidia's full physical AI stack. It is a part of the ongoing build-out of AI infrastructure and is expected to further drive growth in the data center segment.

A：Nvidia's AI computing technology is fully utilized across every cloud and is seen as the most capable and economic choice for hyperscale, Neo Cloud, and AI lab partners. Nvidia supports both closed and open models and is known for its performance, fungibility, and durability, making it the preferred platform for AI computing infrastructure.

A：Nvidia's unique capabilities driving growth include its architecture, which supports every model and is expanding its share in both closed and open model adoption. The company's full-stack AI factory platform is another key driver, with an expanding share of the data center market and significant revenue growth across its AI computing stack.

A：Nvidia highlighted performance gains and product launches such as the Hopper series, which increased the revenue opportunity, and Vera Rubin, a product that offers extreme performance gains and is expected to mark the fastest product ramp in Nvidia's history. The company also mentioned the success of the Grey CPU, the introduction of the next generation Vera CPU, and the grok 3 lpx system, which demonstrates significant performance improvements in data processing.

A：Recent milestones include global VC funding in AI exceeding $400 billion in the first half of 2026, surpassing the previous year's total, with nearly 20 companies in the AI ecosystem now logging over $1 billion in annualized run rate revenue. Additionally, vertical enterprise software experienced the fastest growth, and revenue contributions from various sectors like automotive, financial services, manufacturing, and healthcare reached $15 billion in the trailing 12 months.

A：Nvidia's business primarily through regional Neo Clouds grew 35% sequentially and more than tripled year over year in Q2. Growth is attributed to countries allocating land and power directly to regional cloud partners. We have helped build entire infrastructure businesses for Neo Clouds worldwide, and there is a strong demand pipeline for diverse off-takers.

A：Nvidia has introduced a revenue sharing model with a minimum revenue guarantee that gives lenders confidence to underwrite projects. This model involves Nvidia providing a take-or-pay commitment on a portion of the facility's capacity, and sharing in the revenue earned above that floor, in exchange for a portion of the Neo Cloud's revenue. This creates a reoccurring usage linked revenue stream alongside core platform revenue.

A：Nvidia's three unique capabilities—running every model on its platform, capturing a larger share of the data center TAM, and extending AI into markets through CUDA libraries—reinforce one another and drive the company's growth.

A：The growth and compute needs of the Frontier AI Labs are being addressed by a meaningful investment of nearly $50 billion by Nvidia, and by establishing partnerships with leading infrastructure capital providers to raise over $500 billion of third-party capital. These partnerships will help the AI labs build and fund their infrastructure using long-term institutional capital at relatively attractive rates.

A：Nvidia has secured land, power, and shell capacity through a partnership with SoftBank Energy to exclusively host Nvidia compute at their Portsmouth site. The initial deployment is expected to support 4.25 GW of AI factory capacity, with potential for multiple upgrade cycles. OpenAI has committed to substantially deploying Nvidia AI infrastructure through 2030, with Nvidia providing selective credit enhancement for nearly 2 GW of compute.

A：Given ongoing geopolitical uncertainty, Nvidia is managing its business exposure in China by aligning with U.S. government licensing requirements. Next year's guidance anticipates no China data center compute revenue due to the current situation.

A：The GAAP and non-GAAP gross margins were both 75%, largely unchanged from the previous quarter due to a similar product mix. The sequential increase in GAAP and non-GAAP operating expenses by 10% and 11%, respectively, was primarily due to high compute infrastructure costs and compensation and benefits costs.

A：The expected total revenue for the third quarter is $108 billion plus or minus 2%. Sequential growth is expected to be driven primarily by AC and E with data centers, with growth in hyperscale L anticipated to reaccelerate in the fourth quarter and into fiscal year 28. Vera Rubin is expected to account for about 20% of data center revenue in the third quarter.

A：The company now expects GAAP and non-GAAP operating expenses to be approximately $9.2 billion and $9.0 billion respectively for the full year. The expenses are expected to grow in the low 50s, driven by a broadening of the product portfolio and an increase in the usage of AI tools that enhance engineering productivity. The company is working on GAAP and non-GAAP tax rates to be between 16 and 18%, excluding any discrete items.

A：Component costs have risen significantly, leading to extreme pricing conditions in memory. The magnitude of price increases has exceeded expectations and is anticipated to continue into the next year. As a result, the company has reset expectations for Q3, with GAAP and non-GAAP gross margins expected to be 74% plus or minus 50 basis points. Margins are expected to bottom in Q4 in the 71% to 72% range and settle at 72% to 73% in fiscal year 28 as price increases take effect in Q1.

A：The AI build-out is a major factor contributing to memory scarcity, unlike a component that merely raises costs without providing offset benefits. Tightly-supplied memory is a symptom of the same demand surge that's driving the company's growth. The company has deep relationships with all three major memory suppliers and is working closely with them to increase capacity. The roadmap requires a focus on addressing this issue.

A：The projected demand growth is driven by the increasing use of AI, particularly large language models that require extraordinary amounts of compute. The company's unique position as the only entity offering a complete AI factory platform is a factor. There is an entire market segment, including AI, Neo Clouds, AI startups, and enterprises, which is experiencing 100% annual growth and represents half of the company's business. The company is working to secure infrastructure for future computing needs, involving itself in and securing infrastructure far upstream and downstream in the supply chain.

A：The company has never forecasted or guided to a year in advance. While the demand is much greater than 70%, the company can confidently deliver 70% with the current supply. The company aims to be consistent with its customers, shareholders, and supply chain by providing the same view. The upcoming year is expected to be extraordinary, and the company is working with its supply chain to increase capacity. Consistency in information is crucial for resource allocation and planning.

A：The four phases of the AI life cycle include preparing all the necessary data, which involves synthetic and real data and human labor; pre-training the models; post-training; and agentic inference, which is considered extremely complicated.

A：Nvidia's architecture has evolved significantly to support AI with the development of the first, second, and third generation of its Envy Link 72 rack scale systems. The architecture was built with NVIDIA NVlink and has supported various stages of the AI life cycle from data creation to AIC inference. The design has also supported multiple types of networking and has significantly increased Nvidia's revenue exposure and productivity.

A：The Hopper, Vera Rubin, and Grok series represent different aspects of Nvidia's product line. Hopper and Vera Rubin are advancements in the architecture, representing the 3rd generation of Envy Link 72 systems, with significant revenue growth. Vera Rubin is a core engine for high interactivity and speed, while Grok is a high-performance computing platform with upcoming advancements in token interact rates and interactivity. The vast majority of data centers will likely use Vera Rubin, while certain high-demand services might benefit from Grok accelerators.

A：The benefits to customers include incredible productivity and performance. The architecture allows for a $60 billion investment in data centers to be preserved and useful for a longer time across multiple phases of the AI life cycle, supporting every type of model. This versatility and longevity of investment contribute to a revenue opportunity per gigawatt of $40 billion, up from $30 billion just 5 years ago.

A：The 70% growth in fiscal 28 indicates a substantial increase in revenue, potentially adding up to a $200 billion uptick over the prior outlook. The prior outlook suggested a trillion dollars over three years, so this growth likely adds about 200 billion to that projection. The constraints are primarily around capacity; despite year-over-year growth of 100%, more capacity needs to be secured as the demand for Nvidia's products is high. The company has made considerable progress in supply chain and partnership agreements, but additional supply needs to be obtained and integrated.

A：The other half of the picture that contributes to Nvidia's value proposition is the AC segment, which encompasses all the enterprise, edge, and government clouds that are not visible in the market because they do not buy custom chips one at a time. These entities require entire factory platforms tailored to their needs, which Nvidia significantly adds value to.

A：The specific cash part of the investment for fiscal 28 is not explicitly mentioned in the transcript.

A：The investments are balancing the need to support the ecosystem and competitive solutions by developing a platform, the AI factory platform, which is not inference-specific but spans the entire AI lifecycle and can be used in any cloud. The speaker is confident that the economics of using Nvidia technology, which is utilized in AI services around the world, will be extraordinary, thus supporting the ecosystem while allowing for competitive solutions.

A：Nvidia's AI factory platform is differentiated from other inference-specific chips as it spans the entire AI lifecycle, can be used in any cloud, and is built to run anywhere, helping to set it up anywhere.

A：Investing in AI labs like OpenAI and Anthropic is seen as a significant opportunity because these companies have the potential to become some of the most consequential technology companies in history, and they are building an ecosystem on top of Nvidia architecture. The speaker expresses delight at the opportunity to partner with them.

A：The importance of supply commitments for the growth and revenue of Nvidia is essential, particularly for the raising of Vera rubbin today and all next year. The commitments provide the confidence in terms of growth and revenue, given the substantial amount of supply already aligned and committed.

A：The rise of open source models is seen as beneficial for Nvidia's business as both closed and open models are gaining rapid use and Nvidia's architecture is the most fungible, being found everywhere from PCs to edge devices, and is foundational to many AI startups and enterprise companies. Open models are enabling companies to build their proprietary AI and are also crucial in areas like cybersecurity. Nvidia's position in open models is strong due to the ubiquitous presence of their CUDA libraries, and the company benefits from both closed and open models succeeding.

A：Recursive self-improvement in AI, as demonstrated by developments at Anthropic and OpenAI, is expected to drive industry demand. This development could lead to further inflection in demand, potentially meaning greater demand for Nvidia's products. The company is looking at this as a significant demand catalyst.

A：The speaker believes that in the future, every company will have a large number of AI agents working continuously in the background to improve the company and our lives.

A：AI agents improve through recursive processes where each run updates the skills document based on how the agent could do better next time, resulting in a form of coarse-grained self-improvement.

A：The speaker asserts that AI has achieved AGI, is now doing productive work, generating profitable tokens, and that more profit can be generated with additional compute power.

A：The speaker mentions that the entire supply chain is challenged and running flat out, with the availability of data centers, power, and other resources like DRAM wafer foundry being some of the constraints.

A：There is more supply than the 70% that is forecasted to be needed, but demand is much higher. The company is working hard to ensure they do not disappoint customers and is transparent about the need for help from the entire supply chain.

A：Nvidia's goal is to put as much compute as possible on a plot of land, with an ideal outcome being a trillion dollars of compute per GW and one piece of land. The speaker indicates that with each new generation, such as Hopper, Grace Blackwell, and beyond, the productivity and value of compute are increasing, which is positive for the industry, customers, and investors.