Title: NVIDIA Gone Parabolic Show: App Economy Insights (How They Make Money) Author: Bertrand Hartman Date: 2026-05-22 URL: https://www.appeconomyinsights.com/p/nvidia-gone-parabolic Note: Written Substack post (free edition), saved verbatim. Prose article — no (mm:ss) timestamps. Body reproduced as published. --- > SpaceX just published its S-1 filing, so we'll spend the next few days digging into what could become the biggest IPO ever. Stay tuned for our full breakdown next week. "Demand has gone parabolic." That was Jensen Huang's closing message after NVIDIA's Q1 FY27 earnings call. His explanation was simple: agentic AI has moved from promise to production, turning token generation into a revenue stream. That's the clearest expression yet of NVIDIA's token economy thesis. Compute is no longer just infrastructure spending. It is becoming the raw material for AI revenue. Today at a glance: 1. NVIDIA's Q1 FY27 2. Business highlights 3. Key quotes from the call 4. What to watch moving forward 1. NVIDIA Q1 FY27 — NVIDIA's fiscal year ends in January, so the April quarter was Q1 FY27. Data Center revenue remains off the charts. Income statement: - Revenue accelerated +85% Y/Y to $81.6 billion ($2.6 billion beat). - Data Center +92% Y/Y to $75.2 billion. - Edge Computing +29% Y/Y to $6.4 billion. - Gross margin was 75% (+14pp Y/Y). - Operating margin was 66% (+16pp Y/Y). - Non-GAAP EPS $1.87 ($0.10 beat). Cash flow: - Operating cash flow +84% Y/Y to $50.3 billion. - Free cash flow +86% Y/Y to $48.6 billion. Balance sheet: - Cash and cash equivalents: $80.5 billion. - Debt: $8.5 billion. Q1 FY27 Guidance (next quarter): - Revenue +11% Q/Q and +95% Y/Y to $91.0 billion ($3.0 billion beat). - Gross margin 75% (flat Q/Q). So, what to make of all this? - Data Center is still the whole story: Data Center revenue surged 92% Y/Y to $75.2 billion and now represents 92% of the business. NVIDIA also changed its breakdown (previously Compute vs. Networking), shifting the focus from what it sells to who is building AI factories: - Hyperscale surged 115% Y/Y to $37.9 billion, driven by customers like AWS, Azure, and Google Cloud. - AI Clouds, Industrial, and Enterprise (ACIE) was nearly as large, rising 74% Y/Y to $37.4 billion. This segmentation helps counter the notion that NVIDIA relies solely on a handful of Big Tech buyers. Demand is broadening across neoclouds, sovereign AI, industrial deployments, and enterprise customers. - Inference is becoming the engine: NVIDIA's growth is no longer just about training bigger models. As AI apps move from chatbots to agents, every query, image, video, and coding task requires real-time compute. That is why Jensen keeps framing AI infrastructure around tokens, not chips. - Margins remain the lie detector: Gross margin stayed at 75% despite the complexity of the Blackwell ramp, higher memory content, and the shift toward full rack-scale systems. If competition were biting or demand were softening, margins would likely show it first. - Cash flow is becoming absurd: Free cash flow nearly doubled to $48.6 billion in a single quarter. NVIDIA is converting growth into cash at a scale that provides strategic flexibility for supply-chain investments, partnerships, and ecosystem support. - Investment gains flattered reported profits: Net profit of $58.3 billion included nearly $16 billion in equity investment gains. NVIDIA did not break them out by company, but the large publicly disclosed holdings include Intel, CoreWeave, and Coherent, with Intel likely the biggest contributor in Q1. - The $91 billion guide raises the bar again: Guidance implies an 11% sequential increase and 95% Y/Y growth. Big picture: NVIDIA's Q1 FY27 results support the same message Jensen delivered at GTC. The AI boom is moving from training clusters to full AI factories built for inference, agents, and token generation. The numbers suggest the cycle is still expanding, not digesting. 2. Business highlights China becomes a call option: China is no longer a clean zero, but it is not back either. The US has reportedly cleared H200 sales to roughly 10 Chinese companies, including Alibaba, Tencent, ByteDance, and JD.com. But the chips have not shipped yet because Beijing has not given the green light, partly because it wants local companies to rely more on domestic alternatives. NVIDIA is still assuming no China Data Center compute revenue in its outlook. The business is accelerating without China, while any actual restart would become incremental upside. If China reopens even partially, it becomes a call option on top of an already booming business. Vera opens a new growth layer: The most surprising development from the call may have been NVIDIA's CPU ambitions. Management said Vera opens a $200 billion TAM and that NVIDIA has visibility to nearly $20 billion in standalone CPU revenue this year. Vera is more than a companion chip for Rubin. It could become a new growth pillar across AI factories, storage, security, and confidential computing. AI factories go physical: NVIDIA announced a strategic partnership with IREN to support the deployment of up to 5 gigawatts of NVIDIA DSX-aligned AI infrastructure across IREN's data center pipeline. NVIDIA also received a five-year right to buy up to 30 million IREN shares at $70. Since the stock trades below that level, this is less a near-term investment than an upside kicker if IREN becomes a major partner. The bottleneck is no longer just access to GPUs. It is power, land, cooling, networking, deployment speed, and operating know-how. 3. Key quotes from the earnings call CEO Jensen Huang on the token economy: "AI can now do productive and valuable work. Tokens are now profitable. Model makers are in a race to produce more. In the AI era, compute capacity is revenue and profits." Jensen says the AI boom has shifted from experimentation to monetization. Bears see hyperscaler spending as a cost. Jensen sees compute as the input required to generate AI revenue. On the CPU opportunity: "The world has billions of human users. My sense is that the world is going to have billions of agents. [...] Every one of those agents are going to spin off sub-agents, and every time they spin these off, you're going to need to do inference. That's where the thinking happens. All of the thinking happens on GPUs. All of the orchestration essentially runs on CPUs." The mental model for Vera: GPUs do the thinking, while CPUs coordinate the work. 4. What to watch next NVDA is up nearly 20% YTD, still outperforming the S&P 500 by a wide margin (the index is less than 8%). The latest 13F filings for Q1 2026 showed that some funds were still buying, such as Altimeter and Tiger Global. The stock remains one of the most widely held names, although many funds are still underexposed relative to its 8% weight in the S&P 500. At ~27x forward earnings, NVIDIA continues to trade mostly in line with the rest of Big Tech. With adjusted EPS surging 140% Y/Y, you could argue it looks cheap. NVIDIA's growth is supply-constrained. Here's what I'm watching: - ACIE growth: NVIDIA's new disclosure shows that AI Clouds, Industrial, and Enterprise are nearly as large as Hyperscale, and growing faster sequentially. If this continues, it would support the idea that the AI buildout is broadening beyond hyperscaler CapEx. - Rubin cadence: Vera Rubin production shipments are expected to begin in Q3 and ramp into Q4. A clean handoff from Blackwell to Rubin would support NVIDIA's $1 trillion Blackwell and Rubin revenue outlook through 2027. A delay could create the demand air pocket that bears have been waiting for. - Token economics: Jensen says tokens are now profitable, but the real test is whether AI labs and AI-native companies can turn token revenue into durable profits after compute, R&D, and customer acquisition. The bear case is that AI infrastructure demand eventually normalizes. The bull case is that agentic AI turns compute into a new industrial base. NVIDIA is building for the latter. Disclosure: the author owns AAPL, AMD, AMZN, GOOG, META, MSFT, and NVDA in the App Economy Portfolio.