Our Blog

Blog Index 

Nvidia's $96 Billion Quarter and the 70% Forecast That Reset the AI Boom Debate

Posted on 28th Aug 2026 06:05:58 in Artificial Intelligence, Machine Learning

Tagged as: Nvidia, Jensen Huang, AI chips, data center, earnings, Amazon, hyperscalers, memory

Nvidia has spent the past year answering the same uncomfortable question: is the AI boom fading? On August 26, 2026, the company answered it with the most direct evidence yet that the build-out is nowhere near done. Nvidia reported fiscal second-quarter revenue of $96.22 billion — up 106% from a year earlier and well above the $92.17 billion analysts expected — and then, on the earnings call, CFO Colette Kress dropped something no Nvidia executive has ever offered: a full-year revenue forecast. Revenue in fiscal 2028 will grow about 70%, the company said. Analysts had been modeling roughly 45%. The stock, which initially slipped after the report, reversed sharply and rallied more than 4% in after-hours trading, then surged again the next day, dragging the entire chip sector higher with it. This was not just another record quarter. It was Nvidia resetting the terms of the AI-boom debate in a single evening.

The Quarter in Numbers: Revenue Doubles Again

The headline figures speak for themselves. Revenue of $96.22 billion marked a 106% increase over the $46.7 billion Nvidia recorded a year earlier, extending a streak of triple-digit growth that would be extraordinary for a startup, let alone the world's most valuable company. Net income more than doubled to $53.95 billion from $24.76 billion, and adjusted earnings per share came in at $2.22 against the $2.10 consensus, according to LSEG data.

Beneath the top line, the data center business — the engine of Nvidia's AI franchise — generated $89.02 billion in revenue, up 117% year over year and 18% sequentially, comfortably ahead of the $86.3 billion estimate. Within that segment, hyperscale sales from public clouds and the largest consumer internet companies doubled to $48.7 billion. The newer reporting category covering AI clouds, industrials, and enterprises grew 138% to $40.3 billion, while the edge computing segment, which includes workstations, robotics, and automotive, rose 27% to $7.2 billion.

Perhaps more important than any single number was the shape of demand. For much of the past year, investors worried that Nvidia's growth depended on a handful of mega-customers. On the earnings call, CEO Jensen Huang argued the opposite: "This time last year, one lab alone was driving the buildout; today, we have a golden age of new AI labs and startups, multiple frontier labs scaling in parallel, a thriving open-model ecosystem and physical AI coming online."

The 70% Forecast: A First in Nvidia History

The moment that moved markets came mid-call, when Kress said Nvidia expects fiscal 2028 revenue to grow about 70% over fiscal 2027. Analysts, on average, had expected around 44% to 45%. Nvidia typically guides only one quarter ahead, so a full-year projection was unprecedented — and the optics were unmistakable. Based on the consensus fiscal 2027 revenue estimate of roughly $396 billion, a 70% increase implies about $673 billion in sales next year, which would vault Nvidia past Apple and Alphabet and leave only Amazon ahead of it among U.S. technology companies by revenue.

Huang was characteristically direct about why the number is not bigger: "Even though our demand is much greater than 70%, our supply allows us to confidently deliver 70%. And we're going to continue to work with our supply chain to increase on that." He added that the company issued the outlook because it has visibility into next year's computing requirements and wanted its partners — the firms providing land, power, and data center shells — to plan against the same information. "Everybody's putting a lot of resources in play," Huang said.

The forecast is backed by hard numbers from the customer side. Kress said the cloud industry's backlog now exceeds $2 trillion, and that capital expenditure by the top five hyperscalers is expected to reach nearly $800 billion in 2026 and climb to $1.3 trillion in 2027. Nvidia's own financial commitments tell a similar story: total commitments across supplier agreements, partnerships, and investments ballooned from $119 billion last quarter to $279 billion, driven primarily by procurement of memory — a direct signal of how aggressively the company is locking in supply for the ramp ahead.

Memory Scarcity and the Margin Squeeze

The one blemish in an otherwise triumphant report was gross margin. Nvidia said margins would bottom out in the fourth quarter of fiscal 2027 in the 71% to 72% range, pressured by relentlessly rising memory prices — a squeeze Kress attributed to the AI build-out itself. "Memory scarcity today is being driven in large part by the AI buildout itself," she said, adding that the company chose to be direct about the headwind "rather than let it linger as an open question." Memory prices are expected to be even higher next year, a drag for most hardware companies but a windfall for memory makers such as Micron, whose stock jumped more than 3.5% in after-hours trading.

The margin commentary arrived alongside a new risk-factor disclosure: Nvidia's quarterly filing identified indebtedness as a standalone risk for the first time, noting $33.5 billion in senior notes outstanding and a $25 billion commercial paper program. The company said its rising obligations could "adversely affect" its financial condition and cash flows — a candid footnote to a strategy of financing customers and partners that has drawn criticism in some corners as "circular financing." Kress pushed back directly on the characterization: "We know some will call this circular financing. We see it differently. We believe these investments, measured against the strength of their demand, the business they create for us, the ecosystem they build on Nvidia's platform, and the equity returns on invested capital will be excellent. And our risk is limited."

The Amazon Deal and the Physical AI Push

Alongside the results, Nvidia announced a major expansion of its partnership with Amazon Web Services. AWS will deploy an additional 2 million Nvidia GPUs across fiscal years 2027 and 2028, building on a 1-million-GPU agreement announced in March, and will adopt Nvidia's new Vera CPU both standalone and integrated with the upcoming Rubin systems. Huang told CNBC that Amazon can also purchase "millions of CPUs." Beyond silicon, Kress said Amazon will adopt Nvidia's full physical AI stack — Omniverse, Cosmos, Isaac, and Jetson — to power its fleet of warehouse robots, extending the partnership from cloud infrastructure into the robotics layer of Amazon's operations.

The deal illustrates the broadening of Nvidia's customer base that Huang keeps emphasizing. He described a new class of buyers he calls ACIE — regional AI companies, neoclouds, startups, and enterprises — that were previously "invisible" and now represent what he believes could become larger than the cloud market itself. Earlier in August, Nvidia launched a program with six major financial firms to help such customers and frontier labs finance their infrastructure, addressing one of the biggest practical obstacles to broader AI adoption: the capital intensity of building data centers.

Market Reaction: A Chip Rally and a Re-rating

Wall Street needed only minutes to reprice the story. Nvidia shares, up just 12% for the year heading into the print — a frustrating lag behind rivals after a more-than-1,000% surge over three years — rose more than 4% in after-hours trading as the 70% forecast sank in. On Thursday the rally accelerated: the stock gained as much as 6.8% intraday, its best session since June 1, putting it on track to add roughly $295.7 billion in market value in a single day. The move dragged the sector along in a nearly $150 billion chip-stock rally, with Intel, Micron, Broadcom, and SK Hynix rising between 1.3% and 3.5%, and AI cloud firms backed by Nvidia, including CoreWeave and Nebius, climbing 2% to 4.5%. At least 16 brokerages raised their price targets on Nvidia following the results, according to LSEG data.

The valuation case Nvidia is now making is striking: despite being the market's most valuable company, it trades at a forward price-to-earnings multiple of roughly 17.9 — far below AMD's 37.2 and Intel's 46.2. Shareholders were also rewarded directly. Nvidia returned a record $26 billion during the quarter through buybacks and dividends, working through an $80 billion repurchase authorization announced in May, and Kress signaled that excess free cash flow will increasingly flow back to investors. Huang, meanwhile, made his own conviction clear with a line that captured the evening: "Investing in these companies are a once-in-a-generation opportunity. The only regret that I have is that I didn't invest more and sooner." He also predicted that OpenAI and Anthropic "will likely go public soon, and others will follow."

What It Means for the AI Trade

Nvidia's quarter does not settle every question about the AI economy. The memory crunch will keep pressuring margins into next year, the circular-financing debate will not disappear, and a company carrying hundreds of billions in supply commitments assumes real execution risk. But the signal the market chose to hear is hard to miss: the demand Nvidia can see is limited by what it can build, not by what the world wants to buy. A 70% growth forecast — issued voluntarily, a year ahead of time, and described as supply-constrained rather than demand-constrained — is the strongest data point yet that the AI infrastructure cycle has years left to run. For the chip sector, the hyperscalers, and the broader technology market, that single number changed the conversation from "how long can this last?" to "how fast can supply catch up?"

Sources

Recent Posts

Catgories

Archives

whatsapp me