Analyst(s): Nick Patience
Publication Date: May 22, 2026
Dell Q2 FY 2025 earnings report, highlighting how growing demand for AI solutions and Dell’s partnership with NVIDIA are already producing record results for Dell’s Infrastructure Solutions Group, and why softness in Dell’s Client Solutions Group numbers for the quarter suggests that the PC refresh supercycle, which kicked off just a few weeks ago, may take a few more quarters to gain momentum.
What is Covered in This Article:
- NVIDIA Q1 FY2027 financial results and headline metrics
- Blackwell architecture adoption across hyperscalers and frontier model builders
- New Hyperscale / ACIE reporting segmentation and what it signals
- Vera CPU platform as an incremental revenue opportunity for agentic AI
- NVIDIA’s characterization of SRAM-based (LPX) accelerators as niche
- Sovereign AI revenue growth and how NVIDIA defines the opportunity
- China export controls: explicit exclusion from Q2 guidance
- Q2 FY2027 outlook and capital return program
The News: NVIDIA Corporation (NASDAQ: NVDA) reported financial results for the first quarter of fiscal year 2027 on May 20, 2026. Total revenue was $82 billion, up 85% year over year and 20% sequentially – the company’s third consecutive quarter of year-over-year revenue acceleration and fourteenth straight quarter of sequential growth. The $13.5 billion sequential increase was a company record. GAAP earnings per share were $1.87, ahead of the consensus of $1.77.
Data center revenue of $75 billion was up 92% year over year and 21% sequentially, driven by continued demand for Blackwell architecture systems. Data center computing revenue was $60 billion (up 77% YoY) and data center networking revenue was $15 billion (nearly tripling YoY). Free cash flow reached $49 billion, up from $35 billion in the prior quarter. NVIDIA also announced an $80 billion new share repurchase authorization and raised its quarterly dividend from $0.01 to $0.25 per share. The Q227 guidance (see below) explicitly excludes any China data center compute revenue, reflecting continued export licensing uncertainty around H200 shipments
NVIDIA Q1 FY2027: Data Center Diversification, Blackwell Scale, CPU Upside
Analyst Take: NVIDIA’s Q1 FY2027 results are notable not just for their scale – $82 billion in revenue, $75 billion in data center, $49 billion in free cash flow – but for what management said on the earnings call about the structural drivers behind the numbers. The earnings call sketched a company that is systematically expanding its addressable market while using the current demand cycle to pre-position in product categories it did not previously occupy.
Blackwell Is Now the Baseline, Not the Leading Edge
Blackwell has moved from ramp to standard deployment. Every major hyperscaler, cloud provider, and frontier model builder has adopted it. On the call, CFO Colette Kress noted that frontier model builders and hyperscalers have cumulatively deployed hundreds of thousands of Blackwell GPUs, describing it as the fastest product ramp in NVIDIA’s history. Microsoft’s Farweave facility – described as powered by hundreds of thousands of Blackwell GPUs – is now live. AWS has committed to adding more than one million Blackwell and Rubin GPUs this year. Google will offer Blackwell-based compute in its cloud, including confidential computing capability.
The Blackwell Ultra (GB300) delivered a 2.7x throughput improvement and 60% reduction in cost-per-token compared to its predecessor on GV300 over the prior six months, a performance trajectory that makes the economics of inference increasingly compelling for model builders. The H100 rental price rising 20% year to date and the A100 pricing up 15% indicate that infrastructure owners are generating returns beyond the depreciable life of their hardware, which, in turn, supports continued capital deployment.
The Anthropic partnership disclosed on the call is worth noting. Kress stated that NVIDIA has deepened its collaboration with Anthropic as a strategic partner for expanding its compute capacity, with support flowing through AWS, Azure, CoreWeave, and others. NVIDIA now lists Anthropic alongside OpenAI, xAI, Meta, Google, and others, as major frontier labs building on its infrastructure. For customers assessing the compute dependencies of frontier model providers, this disclosure confirms that NVIDIA’s position at that layer of the stack is, if anything, strengthening.
The New ACIE Segment Is More Than a Reporting Change
NVIDIA reorganized its data center reporting into two sub-segments: Hyperscale and ACIE (AI Cloud, Industrial, Enterprise). Hyperscale – the major public cloud and consumer internet platforms – contributed $38 billion (approximately 50% of data center revenue), growing 12% quarter over quarter. ACIE contributed $37 billion, growing 31% quarter over quarter, with AI cloud revenue within it more than tripling year over year.
During Q&A, Huang described two distinct demand categories: first, the major hyperscalers where CapEx is already running toward a trillion dollars and heading toward $3-4 trillion; and second, a broader category comprising AI-native regional clouds, enterprise on-premises deployments, industrial infrastructure, and sovereign AI clouds. His argument was that NVIDIA should grow faster than hyperscale CapEx precisely because this second category is large, fragmented, and structurally less addressable by custom silicon. He said: “If they don’t have the compute, they won’t have the revenues. Compute is revenues. Compute is profit.”
The ACIE segment’s 31% sequential growth is a concrete indicator that this second-category thesis is already playing out in revenue. The number of partner data centers exceeding 10MW nearly doubled in a year to more than 80 sites – a supply-side data point that suggests the infrastructure enabling ACIE demand is scaling materially.
Vera CPU: A New Revenue Layer Above the $1 Trillion Pipeline
The most strategically significant new disclosure on the earnings call was NVIDIA’s characterization of the Vera CPU as a standalone revenue opportunity. During Q&A, Huang clarified that the $20 billion CPU revenue expectation for the year refers specifically to standalone Vera CPUs, not CPUs bundled as part of the Vera Rubin system. He positioned the Vera CPU as an agentic workload processor: GPUs handle “the thinking” in inference, while CPUs handle orchestration, I/O, memory management, and tool use, all of which scale with agentic deployments.
Crucially, when asked about upside above NVIDIA’s previously stated $1 trillion platform revenue visibility (which was discussed at GTC in March 2026 and covered Blackwell and Rubin), Huang identified three incremental sources: first, growing share of frontier AI model compute; second, standalone Vera CPU revenue (which he said was not included in the $1 trillion figure and expects to be the second-largest source of upside); and third, LPX. Production shipments of Vera Rubin are scheduled to begin in Q3, with ramp continuing into subsequent quarters.
Whether $20 billion in standalone CPU revenue materializes at the pace suggested is a legitimate open question. NVIDIA is entering the CPU market with established incumbents (AMD EPYC, Intel Xeon, Arm-based custom silicon from hyperscalers), and the agentic AI workload characterization, while directionally plausible, is not yet validated at scale. But the TAM framing – a $200 billion CPU addressable market – and the explicit exclusion of Vera CPU from the prior $1 trillion visibility figure are meaningful signals of where NVIDIA expects its next growth layer to come from.
LPX: Confirmed as Niche, With a Defined Use Case
When asked about traction with custom merchant silicon (CPX, LPX), Huang was direct. LPX – NVIDIA’s SRAM-based accelerator, designed for low latency and high token rate – has limited throughput, limited model size capacity, and limited context processing capability. He said its use case is not broad: it is suited to customers operating large portfolios of token services where a subset requires premium, very-high-token-rate inference. He estimated that today it represents significantly less than 20% of the market, with 20% as a possible ceiling over time.
This framing has direct implications for the competitive positioning of SRAM-based inference accelerators more broadly, including products from companies such as Groq (recently acquired by NVIDIA) and others pursuing similar architectural bets. Huang did not dismiss the category outright – he said NVIDIA is ready to work with service providers to enable it – but the message is consistent with what he has said previously: these are specialized products for a constrained use case, not a structural alternative to Blackwell for general inference.
The Sovereign AI Revenue Line
Sovereign AI revenue grew more than 80% year over year, with NVIDIA infrastructure now deployed across nearly 40 countries. The $50 trillion in aggregate GDP cited by Kress is a framing device rather than an analytical construct, but the underlying footprint data is substantive.
What is notable is how NVIDIA defines sovereign AI. The term is used primarily as a revenue category descriptor rather than a strategic or policy concept. During Q&A, Huang folded sovereign AI clouds into the same “second category” as AI-native regional clouds, enterprise on-premises deployments, and industrial infrastructure. His definition was functional: deployments where latency, reliability, or regulatory requirements make dependence on a public hyperscaler cloud impractical or politically unacceptable. He cited chip fabrication plants as an example of industrial contexts where “connecting to a cloud service provider doesn’t make any sense.”
This is a narrower framing than how sovereign AI is increasingly discussed in policy contexts, where the emphasis is on national control of AI infrastructure, data residency, and strategic autonomy from any external technology provider, including NVIDIA itself. NVIDIA’s definition is, understandably, aligned with its commercial interest: sovereign AI is a revenue category that justifies why custom hyperscaler silicon does not threaten the ACIE segment. The Futurum Group’s own research on sovereign AI strategy tracks a more nuanced picture, in which countries’ motivations range from data sovereignty to industrial policy to explicit strategic autonomy from US technology supply chains.
Guidance and Final Thoughts
For Q2 FY2027, NVIDIA guided to revenue of approximately $91 billion, plus or minus 2%, with GAAP gross margin of 74.9% and non-GAAP gross margin of 75%. The guidance explicitly excludes any China data center compute revenue, reflecting continued export licensing uncertainty around H200 shipments.
NVIDIA’s updated segmentation and platform messaging suggest the company is trying to prove that AI infrastructure demand is becoming structurally diversified rather than concentrated in a handful of hyperscale customers. The segmentation, inference framing, and CPU expansion create clearer scorecards for durability, but they also raise expectations for execution across more product lines. Momentum remains tied to whether AI factory buildouts broaden and remain repeatable across hyperscalers, AI clouds, and enterprise buyers. The push into CPUs and inference economics also expands the debate from accelerator leadership toward broader control of AI system architecture and workload efficiency. However, the broader the platform scope becomes, the more execution risk shifts toward supply coordination, product cadence consistency, and sustaining pricing power as customers evaluate alternative inference and systems strategies.
See the full press release on NVIDIA’s Q1 FY 2027 financial results on the company website.
Declaration of generative AI and AI-assisted technologies in the writing process: This content has been generated with the support of artificial intelligence technologies. Due to the fast pace of content creation and the continuous evolution of data and information, The Futurum Group and its analysts strive to ensure the accuracy and factual integrity of the information presented. However, the opinions and interpretations expressed in this content reflect those of the individual author/analyst. The Futurum Group makes no guarantees regarding the completeness, accuracy, or reliability of any information contained herein. Readers are encouraged to verify facts independently and consult relevant sources for further clarification.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other Insights From Futurum:
NVIDIA Q4 FY 2026 Earnings Highlight Durable AI Infrastructure Demand
NVIDIA Q3 FY 2026: Record Data Center Revenue, Higher Q4 Guide
Sovereign AI: What Nations Want (And What They’ll Actually Get)
Author Information
Nick Patience is VP and Practice Lead for AI Platforms at The Futurum Group. Nick is a thought leader on AI development, deployment, and adoption - an area he has researched for 25 years. Before Futurum, Nick was a Managing Analyst with S&P Global Market Intelligence, responsible for 451 Research’s coverage of Data, AI, Analytics, Information Security, and Risk. Nick became part of S&P Global through its 2019 acquisition of 451 Research, a pioneering analyst firm that Nick co-founded in 1999. He is a sought-after speaker and advisor, known for his expertise in the drivers of AI adoption, industry use cases, and the infrastructure behind its development and deployment. Nick also spent three years as a product marketing lead at Recommind (now part of OpenText), a machine learning-driven eDiscovery software company. Nick is based in London.
