Nvidia Earnings: $68 Billion in 90 Days. Wall Street Still Doesn't Fully Believe It

NVIDIA just did something nobody's ever done before.

Let's put $68 billion in perspective.

That's more than Netflix makes in a year. It's more than Goldman Sachs brings in.
NVIDIA, a company that most people 10 years ago thought of as "the graphics card company," just generated $68.1 billion in a single quarter.

And it grew 73% from a year ago.

But here's what really caught my attention: they guided next quarter to $78 billion. Wall Street expected $72 billion. NVIDIA blew right past that before the quarter even started.

Seems 73% revenue jump fails to impress Nvidia investors. 

Let's walk through it.

Let me give you the highlights that actually move the needle.

Revenue: $68.1B vs Wall Street expected $66.2B. Beat by $1.9 billion.

EPS: $1.62 vs expected $1.53. Beat by 9 cents. That might sound small, but at this scale it adds up to billions in extra profit.

Gross margin: 75%. Think of this as NVIDIA keeping 75 cents for every dollar it earns. That's exceptional. Apple-level exceptional.

Net income: $43 billion. Doubled year-over-year. The entire company was worth less than this five years ago.

Q1 FY27 Guidance: $78 billion. The most important number in the whole report. It tells you the next quarter is expected to be even bigger.

And one more number that deserves its own line: $216 billion, that's NVIDIA's full-year revenue for fiscal 2026. Up 65% from last year. The year before that? $130.5 billion. The year before that? $60.9 billion.

See the pattern? This company is compounding at a rate that's genuinely hard to wrap your head around.

Here's the simple version: AI runs on chips. Specifically, NVIDIA's chips.

When companies like Amazon, Google, Meta, and Microsoft build out AI systems, the kind that generate text, images, video, analyze data, and increasingly, make decisions, they need enormous amounts of computing power. They buy that power from NVIDIA.

The Data Center business, which is NVIDIA's AI-chip segment, brought in $62.3 billion last quarter alone. That's 75% growth YoY, and it now makes up more than 91% of total revenue.

But Jensen Huang, NVIDIA's CEO, made something very clear on the earnings call. This isn't just "companies buying servers." Something structural has shifted.

What does "agentic AI" mean? Think of it this way: the first wave of AI was chatbots answering questions. The new wave is AI doing things, autonomously completing multi-step tasks, writing code, managing workflows, running analyses. Each of those tasks requires generating "tokens," and tokens require compute. And compute? That means NVIDIA chips.

Huang put it bluntly: "Compute equals revenues."

That's not marketing spin. It's a real economic loop. 

The more AI a company deploys, the more revenue it generates, which funds more AI investment, which requires more compute. NVIDIA sits at the center of that loop.

One of the biggest investor concerns going into this earnings report was: "What if Amazon, Google, and Microsoft slow down their AI spending?"

That concern got answered loud and clear.

The biggest hyperscalers are collectively expected to spend around $650 billion on AI infrastructure in 2026 alone, up 70% from last year. Amazon's AI capex is expected to grow $200 billion, up 50% YoY. Google's, up 100%. Meta's, up 75%.

NVIDIA's CFO Colette Kress pointed this out on the call. She wasn't bragging. She was giving investors a blueprint for demand visibility.

And NVIDIA isn't just waiting around. The company has already strategically secured inventory and supply commitments to meet demand beyond the next several quarters, further out than usual. 

That's a signal that the company sees this growth lasting well into 2027.

NVIDIA just shipped its first Vera Rubin samples to customers. Production shipments are on track for the second half of this year.

Why does this matter? Vera Rubin is expected to deliver 10 times lower inference costs per token compared to Blackwell. In plain English: the same AI work, done for one-tenth the price. That makes it dramatically more attractive for enterprise customers, and it drives even more adoption.

Huang says it will "extend that leadership even further." Based on what Blackwell has already done, that's not a claim investors should brush off.

This is where I have to be straight with you: I'm not a financial advisor, and this isn't investment advice. But here's the context that's worth thinking about.

Going into today's premarket session, $NVDA is trading around $192–195. Motley Fool noted it's trading at roughly 25 times forward earnings which, for a company growing revenues at 73% YoY, many analysts consider reasonable.

$NVDA is up about 5% YTD. It's not cheap, but it's not the speculative bubble it was being called six months ago.

Gene Munster at Deepwater Asset Management put it well: the real debate isn't whether AI is real. The debate is what growth looks like in 2027 and 2028. NVIDIA's Q1 guidance: $78 billion, 77% growth YoY, is a pretty strong opening argument.

There are real risks to watch. Gaming revenue came in at $3.7 billion, below the $4 billion expectation. Supply constraints are expected to pressure gaming in Q1 and beyond. And NVIDIA explicitly excluded China data center revenue from its guidance due to export restrictions.

AMD is going after Nvidia's customers with cheaper GPUs and its new open-standard Helios platform. CEO Lisa Su confirmed Helios systems are on track to ship in the second half of 2026, targeting a 60% growth rate in data center revenue over the next few years. 

Broadcom plays it differently. UBS analyst Timothy Arcuri estimates Broadcom's AI revenue could climb to roughly $60 billion in fiscal 2026, up nearly 3x YoY, driven by custom TPU chips that are significantly more affordable than Nvidia's Blackwell, priced between $10,500–$15,000 versus $40,000–$50,000.

These aren't deal-breakers. But they're worth knowing.

Three months ago, plenty of smart people were asking whether the AI trade was over.

NVIDIA just answered that question with a $68 billion quarter, a $78 billion outlook, a next-generation chip sample already in customer hands, and a CEO who sounds more confident than ever.

The agentic AI era, where AI actually does things rather than just answers questions, is just getting started. And NVIDIA's customers are spending as if they believe it.

Whether you own NVIDIA, are thinking about it, or are just watching from the sidelines, this report matters. It's the clearest signal yet about where AI infrastructure spending is headed.

Keep watching GTC 2026: NVIDIA's annual tech conference is March 16 in San Jose. It's expected to bring major product announcements and more clarity on the Vera Rubin roadmap.

That's the next date on your calendar.