Nvidia jolted investors on Wednesday with a projection that revenue would rise 70% next fiscal year and said demand for its AI chips is growing at 100%, sending the stock higher in after-hours trading.

The preliminary growth forecast far exceeded analyst expectations. For fiscal 2028, Wall Street had been modeling revenue of about $570 billion, or growth of roughly 44% from fiscal 2027, the current year. Applied to fiscal 2027 revenue and expected revenue, Nvidia’s forecast implies fiscal 2028 revenue in the range of $690 billion to $700 billion, more than $100 billion above analyst estimates.

“We wanted to make sure that everybody has the same set of information,” CEO Jensen Huang said on a conference call Wednesday. “We’ve got a huge year coming up next year, and it’s going to be pretty extraordinary.”

“It is the case that we’ve never forecasted, never guided to a year in advance,” Huang said later on the call.

Melissa Otto, global head of Visible Alpha research at S&P Global, said the scale of the top-line outlook “blew away expectations,” especially since Nvidia does not usually provide guidance that far ahead.

“I think what wowed the market was that 70% fiscal year 2028 number that they gave that was way ahead of Visible Alpha consensus,” Otto said. “I think the whole market was like, ‘Whoa, 70%.’”

Chief financial officer Colette Kress said customer forecasts pointed to Nvidia’s growth doubling next year. Her comments, delivered after the market closed, helped push Nvidia shares up more than 4% in after-hours trading.

Nvidia is also dealing with significant supply constraints, as are other mega-cap technology companies. “Our entire supply chain is challenged, and it’s everybody; everybody is really running flat out,” Huang said.

He added that, without those constraints, revenue growth next year would likely be even higher.

“Even though our demand is much greater than 70%, our supply allows us to confidently deliver 70%,” Huang said in response to an analyst question.

The forecast followed quarterly results that beat expectations across the board. Revenue for the fiscal second quarter was $96.2 billion, up 106% from a year earlier and ahead of analysts’ estimates of $92.2 billion. The company earned $2.22 per share on a non-GAAP basis, above the $2.06 to $2.09 expected. Kress guided the next quarter to $108 billion in revenue, matching the buyside whisper range of $105 billion to $108 billion.

Huang said demand is coming from both hyperscale cloud providers and non-hyperscalers, including sovereign AI, neoclouds, AI startups and enterprises. The non-hyperscale group “represents about half our business, and that’s growing 100% a year,” he said.

He said much of the demand reflects changes in how AI systems consume compute, particularly as agentic AI becomes more common. Depending on the task, an AI agent can use 15 to 100 times more compute than a human user, he said. The number of agents inside companies is also expected to increase.

“We have 40,000 employees, roughly,” Huang said. “In the future, we’ll have 400,000 agents, 4 million agents, and those agents are running continuously.”

‘We see it differently’

As booming demand for AI chips has made Nvidia the world’s most valuable company, with a market capitalization of more than $5 trillion, critics have raised concerns about the company’s investments across the AI industry, from data center operators to frontier model makers such as Anthropic. Many have argued that the deals, financed from Nvidia’s large cash pile, create dangerous interdependencies in the AI ecosystem.

Kress defended the company’s approach during the call.

“We recognize the scale of this support, and we know some will call this circular financing,” the Nvidia finance chief said. “We see it differently.”

She said frontier AI labs have already demonstrated technology leadership, customer traction and “skyrocketing” usage.

“We expect them to become the largest technology companies in history,” Kress said. Their growth, she added, “isn’t limited by their technology or customer demand. It’s limited by compute.”

Nvidia disclosed maximum gross guarantee exposure of $108.5 billion, most of which stems from credit support for SB Energy’s Ohio tech campus, which will host Nvidia compute leased to OpenAI. The rest comes from $3.5 billion backing lease obligations for AI cloud partners. Nvidia has also disclosed investments of nearly $50 billion in frontier AI labs and has partnered with Apollo, BlackRock, Blackstone, Goldman Sachs and KKR to raise more than $500 billion in third-party capital for AI infrastructure.

Kress said those investments are low-risk and high-reward for Nvidia.

“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 our invested capital will be excellent, and our risk is limited,” Kress said.

She added that demand from AI labs will account for about a quarter of Nvidia’s business next year. Kress also said Nvidia’s platform is “fungible and durable” and can be used elsewhere if a partner changes its forward-looking projections.

For smaller AI-native cloud providers known as neoclouds, Kress said Nvidia offers a structure in which it guarantees payment for a minimum portion of a data center’s capacity, satisfying lenders. In return, Nvidia takes a share of the provider’s rental revenue above that threshold.

“Independent capital still underwrites every deal on its own merits. We’re not making loans,” Kress said. “In this model, we get paid twice—once on the hardware sale, and again through the share of rental revenue.”

Markets have not embraced the circular nature of these arrangements. Bill Birmingham, managing director at Rex Financial, said that when reports surfaced in July that Nvidia was in talks to guarantee as much as $250 billion in capacity for OpenAI in Ohio, the credit-default swap market repriced Nvidia’s five-year risk from 40 basis points to 82 basis points.

“The equity shed $250 (billion) in turn,” Birmingham wrote in a pre-earnings note seen by Fortune. “Even though the final number came in at $105B, the market read this as less demand and not less risk.”

Margin squeeze

One area that was less impressive was Nvidia’s gross margin guidance. The company projected third-quarter gross margin of 74%, down from 75% in the second quarter, Otto noted. Still, she said the market was already expecting some pressure.

“The market was expecting 72.6% for Q3, and the fact that they guided to 74% suggests that their gross margin is actually more resilient than the market was expecting,” Otto said.

In her CFO commentary, Kress said supply and capacity commitments jumped from $119 billion to $279 billion, driven by rising memory costs, which have become a persistent source of pressure across the technology sector. During the call, she further clarified that higher memory prices led Nvidia to reset expectations for next year.

John Belton, a portfolio manager at Gabelli Funds, said Nvidia likely got ahead of the memory-price issue by working early with suppliers on long-term agreements with locked-in prices. Huang said on the call that Nvidia had worked with suppliers well in advance to gain visibility into pricing.

“A long time ago, people asked me why it is that we’re working with memory suppliers when we’re a chip company,” he said. “Today, people understand it’s really quite genius that we were working on our supply chain so far upstream.”

Birmingham wrote that Nvidia has been raising prices to customers by about 15% to pass through memory-related inflation, which he said was a risk.

“It’s dangerous to raise prices when ROI for AI at the customer level is still unknown,” he wrote.

Kress said margins will bottom at 71% to 72% in the fourth quarter and settle around 72% to 73% next fiscal year as price increases take effect.

This story was originally featured on Fortune.com