Nvidia CEO Jensen Huang told the Goldman Sachs Communacopia + Technology conference on Thursday that his company can grow revenue 70% year over year, and he says he’s confident about it. According to TechCrunch AI, Huang repeated guidance Nvidia first shared last month alongside another record quarter. Analysts peg Nvidia’s current fiscal year at roughly $400 billion in revenue, so 70% growth would push next year to around $680 billion.
That’s the number driving the conversation. Here’s why it matters and what’s actually behind it.
What Huang said
Huang pushed back on the idea that Nvidia just sells chips. “Most people think Nvidia builds a chip. I mean, you need airplanes to ship what we build,” he said. His point: the product has changed shape entirely.
- A modern “GPU,” as he defines it, is a full connected system, not a $399 gaming card. “One GPU now is not $399. It’s $8.5 million dollars,” Huang said, describing 2 million parts linked by NVLink and drawing 250,000 kilowatts.
- Orders for one flagship system, which pairs 36 Grace CPUs with 72 Blackwell GPUs, are growing 27% month over month.
- Huang framed Nvidia as “a foundational platform of the AI ecosystem,” noting it runs models from Anthropic, OpenAI, Google, and open-weight projects.
Why he thinks he can see the future
Huang’s confidence rests on visibility. Nvidia sits at the center of the supply chain, from memory-chip suppliers to data center buildouts to startups. “We’re tracking every single gigawatt of land, power, shell around the world. Literally everything on the planet,” he said. (“Shell” means the empty data center building before the computers go in.)
He leaned on that web of partners as proof: neoclouds, OEMs, cloud providers, and AI-native companies all “reporting back” to Nvidia. “We’re working with everybody, and so we kind of know where everything is.”
What stands out here is that this is a demand-signal argument, not a hype line. Huang is claiming Nvidia has a clearer real-time read on AI infrastructure spending than almost anyone else.
The circular-deal question
The elephant in the room is Nvidia’s habit of investing in companies that then buy its hardware. TechCrunch AI notes the obvious historical parallel: vendor-financing schemes helped sink internet buildout suppliers like Lucent Technologies.
Huang’s answer was cheeky. “It’s not circular because we put a little bit of money in, and a lot of money comes back,” he said, joking, “I look at the spreadsheet, we put in $1 and $100 comes back in.” More seriously, he said Nvidia won’t invest until a company already has real customer contracts generating revenue. He put that figure at $100 billion in such contracts. “I’m not taking any risks. I need a sure thing.”
Why this matters
The backdrop is a steady drumbeat of “will the party end” questions. Competition is coming from every direction:
- Hyperscalers Amazon, Microsoft, and Google are each building their own chips.
- AI labs Anthropic and OpenAI are designing custom silicon too.
- Public rival Cerebras and startups like Etched want a slice of the market.
This is significant because Huang isn’t just defending market share, he’s projecting acceleration into that competitive pressure. If Nvidia hits $680 billion, the custom-silicon threat looks a lot less urgent for at least another year.
What to watch
Huang admitted the soft spot himself: much of AI’s current growth comes from AI-native startups raising huge rounds and spending most of it on their own compute. That’s real revenue, but it’s concentrated and fast-moving.
As the industry matures, expect companies to squeeze more out of every GPU and every token. Efficiency gains eventually eat into raw hardware demand. The tech industry’s one reliable rule is that all big things get disrupted, and Nvidia is now the biggest thing in AI.
For now, the guidance stands and the demand is here. Whether Huang’s crystal ball holds through next year is the story to track. Full details are available at the original source.