Last updated: August 28, 2026 · By Vishal Swami, Founder & Lead AI Reviewer, AISagely
Nvidia projects $673B in sales as AI demand widens beyond the handful of hyperscalers that funded the first two years of the AI buildout. CFO Colette Kress told investors on August 26, 2026 that Nvidia expects 70% revenue growth in fiscal 2028, and if Wall Street's roughly $396 billion consensus for fiscal 2027 holds, that growth rate puts next year's sales near $673 billion.
Short answer: On August 26, 2026, Nvidia guided to 70% revenue growth in fiscal 2028, well above the 44% analysts expected. Applied to Wall Street's ~$396B fiscal 2027 consensus, that implies roughly $673B in sales — ahead of Apple and Alphabet, behind only Amazon among US tech firms. Nvidia says demand now comes from AI labs, neo-clouds, and enterprises, not just hyperscalers, though memory shortages are squeezing near-term margins.
I track chip and AI-infrastructure earnings the same way I track model releases for my best AI models guide: I read the primary filing before I read anyone's headline about it. In my testing of that habit here, the $673 billion figure isn't a number Nvidia printed anywhere in its press release. It's a derived projection — 70% applied to a consensus estimate that isn't even Nvidia's own guidance. That distinction matters more than the headline, and it's the first thing I checked when this story broke.
What you'll need to follow this story
You don't need a trading account or a finance degree, just three things. First, know Nvidia's fiscal calendar: its fiscal year ends in late January, so "fiscal 2027" is the year we're currently in, and "fiscal 2028" starts around February 2027. Second, keep a browser tab open on Nvidia’s own investor relations newsroom, where the actual quarterly press releases live — not the aggregator write-ups. Third, separate "guidance Nvidia gave" from "a number analysts and reporters derived from that guidance." The $673 billion figure is the second kind, and conflating the two is the single most common mistake in how this story is being repeated online.
Step-by-step: how to make sense of the $673B figure
1. Start with the quarter Nvidia actually reported
On August 26, 2026, Nvidia reported second-quarter fiscal 2027 revenue of $96.2 billion, up 106% year over year and 18% quarter over quarter, per its own earnings release. Data center revenue, the segment that includes AI chips like Blackwell and Vera Rubin, came in at $89.0 billion, up 117% year over year. Gross margin held at 75.0%. These are actuals, not projections.
2. Read the guidance Nvidia actually gave for next quarter
For Q3 fiscal 2027, Nvidia guided to $108.0 billion in revenue, plus or minus 2%, with gross margin around 74%. That's a real, near-term forecast from the company itself, one quarter out. It's a useful sanity check before jumping to a full fiscal-year-ahead number.
3. Understand where the 70% figure comes from
Kress told analysts Nvidia expects roughly 70% revenue growth in fiscal 2028 — the year that starts after the current one ends. That's far above the 44% average analyst estimate tracked by LSEG going into the earnings call, and Nvidia doesn't normally guide a full year ahead, which is why outlets flagged it as a change in disclosure.
4. See how $673B gets calculated
Nobody at Nvidia said "$673 billion." Reporters took Wall Street's existing consensus for fiscal 2027 full-year revenue, roughly $396 billion, and applied Kress's 70% growth rate to it. $396B × 1.70 ≈ $673B. If the fiscal 2027 consensus moves between now and next year's guidance, that $673B estimate moves with it — it isn't locked in.
5. Check the demand-versus-supply distinction Huang made
CEO Jensen Huang was specific on the call: "Our demand is much greater than 70%. Our supply allows us to confidently deliver 70%." In plain terms, Nvidia is saying it could sell more than the guidance implies, but memory-component shortages and rising input costs are the ceiling, not customer orders. That's also why Q4 gross margin is guided down to roughly 71-72%, the low point of the year.
6. Look at who's actually buying
Nvidia says AI labs will make up about 25% of next year's business, and neo-cloud GPU capacity is on pace to hit 8+ gigawatts by the end of 2026, up from about 3 gigawatts at the end of 2025. Amazon Web Services alone agreed to deploy 2 million additional Nvidia GPUs through 2027-2028. That's the "AI demand widens" half of the headline: fewer of Nvidia's dollars now come from a small handful of hyperscalers.
Example prompts you can copy
Use these with any AI chat tool to pull the current state of Nvidia's numbers instead of relying on a headline that may already be stale by the time you read it:
“ Pull Nvidia's most recent quarterly earnings release from its investor relations site. Report total revenue, data center revenue, gross margin, and next-quarter guidance, with the exact filing date. “
“ Nvidia's CFO guided to 70% revenue growth for fiscal 2028 on August 26, 2026, against a fiscal 2027 Wall Street consensus of about $396 billion. Has that consensus estimate changed since then? Cite the source and its date. “
“ Compare Nvidia's projected fiscal 2028 revenue (roughly $673 billion, derived from 70% growth guidance) against the most recent revenue figures for Apple, Alphabet, and Amazon. Cite each figure's source and reporting date. “
These work because they point the model at a specific filing and a specific date instead of asking it to summarize "Nvidia's AI growth" in general, which is where answers drift out of date fastest.
Common mistakes to avoid
In my testing of how this story spread across aggregators in the first 24 hours, the most common mistake was treating $673 billion as a number Nvidia guided to. It isn't — it's Wall Street's fiscal 2027 consensus multiplied by Nvidia's own 70% fiscal 2028 growth rate, and both inputs can change before that year even starts. Second, people conflate the $96.2 billion Q2 fiscal 2027 actual with the $108 billion Q3 guide, as if the bigger number already happened; it hasn't been reported yet as of this writing. Third, coverage that leads with "AI demand widens" often skips the supply-side caveat Huang gave on the same call: memory shortages are pressuring margins down to roughly 71-72% by Q4, so wider demand isn't converting to profit at the same rate it was a year ago. Fourth, don't assume this guidance changes what you pay for any AI product today — Nvidia sells chips to cloud providers and labs, not directly to most end users, so a GPU supply story and your ChatGPT or Claude subscription price are several steps removed from each other.
Q2 FY2027 actuals vs. Q3 FY2027 guidance vs. the FY2028 projection
| Period | Revenue | Data center revenue | Gross margin | Status as of Aug 28, 2026 |
|---|---|---|---|---|
| Q2 FY2027 (reported Aug 26, 2026) | $96.2B (+106% YoY) | $89.0B (+117% YoY) | 75.0% | Actual, reported |
| Q3 FY2027 (quarter in progress) | $108.0B ±2% (guided) | Not separately guided | ~74.0% ±50bps (guided) | Company guidance, not yet reported |
| FY2028 (fiscal year starts ~Feb 2027) | ~$673B (derived) | Not disclosed | Guided toward ~71-72% low point before recovering | Analyst-derived from 70% growth rate on consensus |
The pattern in that table is the whole story: the closer you get to today, the more solid the number is. The $96.2 billion is confirmed. The $108 billion is Nvidia's own near-term guide. The $673 billion is two layers removed — a growth rate applied to someone else's estimate for a year that hasn't started.
Tools that make this easier
If you want to track earnings and guidance like this without re-reading a 10-Q every quarter, an AI research assistant can summarize the primary filing faster than most news aggregators publish their take — my AI tool ratings page and best AI models guide cover which models I'd actually trust with a numbers-heavy summary versus which ones tend to round figures loosely. The financing side of this same AI buildout is covered in my piece on why the AI trade increasingly runs on borrowed money, and in Nvidia’s $750 billion in AI deals reigniting circular-financing fears, which is the concentration-risk flip side of "demand widens." The political angle behind Nvidia's growth is in my breakdown of Nvidia’s new PAC and its DC lobbying push. The power-grid consequence of all this GPU deployment is mapped state by state in AI data centers are driving up power bills, and the same "is this valuation real" question shows up in Morgan Stanley’s debate over SpaceX’s AI-driven valuation.
My take
The $673 billion number is a real projection built from real inputs, but it's less solid than the headline makes it sound, because it's a growth rate glued onto someone else's forecast rather than a figure Nvidia stands behind directly. What I'd actually watch is the 70% growth guidance itself — that's Nvidia's own claim, it's a break from how the company normally discloses, and Huang's "demand is much greater than 70%" comment tells you the real constraint is supply, not customer appetite. I'd treat the broadening customer base (AI labs, neo-clouds, enterprises) as the more durable story than the specific dollar figure, since customer concentration has been the biggest single risk to Nvidia's growth story for two years running. Watch the Q3 fiscal 2027 report for whether $108 billion actually lands, and watch fiscal 2027's full-year number before trusting any fiscal 2028 estimate built on top of it.
Frequently Asked Questions
Did Nvidia actually say sales would hit $673 billion?
No. Nvidia's CFO guided to 70% revenue growth for fiscal 2028 on the August 26, 2026 earnings call. The $673 billion figure is a derived estimate, calculated by applying that 70% growth rate to Wall Street's roughly $396 billion consensus for fiscal 2027 — it's not a number Nvidia disclosed in its own press release.
What did Nvidia actually report for its most recent quarter?
For Q2 fiscal 2027, reported August 26, 2026, Nvidia posted $96.2 billion in total revenue (up 106% year over year) and $89.0 billion in data center revenue (up 117% year over year), with a 75.0% gross margin, according to Nvidia's own newsroom release.
Why is Nvidia's growth guidance limited by supply instead of demand?
CEO Jensen Huang said demand is running well above the 70% growth figure, but memory-component shortages and rising input costs cap how much Nvidia can actually ship. That's also why guided gross margin dips to roughly 71-72% in Q4 fiscal 2027 before recovering, as the company pays more for constrained components.
Does this affect what I pay for ChatGPT, Claude, or other AI tools?
Not directly. Nvidia sells GPUs to cloud providers, AI labs, and enterprises, not to individual subscribers. A chip-supply and revenue story like this one sits several layers below the subscription price you see for any consumer AI product, and there's no evidence in this earnings report that it has moved those prices.
How is Nvidia's customer base changing?
Nvidia says AI labs will represent about 25% of its business next year, and neo-cloud GPU capacity is on track to reach 8+ gigawatts by the end of 2026, up from roughly 3 gigawatts at the end of 2025. Amazon Web Services also agreed to deploy 2 million additional Nvidia GPUs through 2027-2028, spreading demand beyond the handful of hyperscalers that dominated Nvidia's early AI revenue.