Nvidia 70% Fiscal 2028 Growth Forecast Tops Analyst Estimates

A quarter that beat expectations, then a forecast that set new ones

Nvidia's fiscal second-quarter report on August 26 beat expectations by a wide margin. What changed the conversation, though, was a forecast for a year that hasn't started yet. CFO Colette Kress told analysts the company expects revenue to grow about 70% in fiscal 2028. The average estimate tracked by LSEG was 44%, and Nvidia had never before given full-year guidance this far ahead1112. Bloomberg's consensus was a little higher, around 45%, so the gap depends on which survey you use. Either way, it is more than 25 percentage points18. Kress also said customer forecasts suggest growth could double next year1011.

Fiscal 2028 starts around January 31, 2027, so this is guidance reaching well over a year out13. The company is also asking investors to take it on faith, because it has no track record with forecasts this long. Bernstein's Stacy Rasgon raised exactly that point on the call14.

The numbers behind the confidence

The quarter alone would have been a strong report. Revenue for the three months ended July 26 was $96.2 billion, up 106% from a year earlier and 18% from the previous quarter36. Estimates for the beat vary a little. Reuters-style coverage put consensus at $92.17 billion16, Bloomberg at $92.5 billion18, and Investing.com at $91.9 billion, which makes the surprise about 4.7%1. Nvidia's own guidance had pointed to about $91 billion, so it beat its own number and the Street's4.

Profits rose faster than sales. Adjusted earnings were $2.22 a share, against estimates of about $2.08 to $2.1011618. GAAP diluted EPS was $2.46, up 128% from a year earlier. GAAP net income rose 126% to $59.7 billion, and operating income rose 124% to $63.7 billion3. Gross margin was 75.0%6.

The data center business did most of the work again. It brought in $89 billion, up 117% from a year ago and above the roughly $85.7 billion to $85.8 billion analysts expected418. Hyperscalers accounted for about $48.7 billion. The rest, about $40.3 billion, came from a group Nvidia calls ACIE: AI clouds, industrial and enterprise customers. That group grew 138%8. Edge computing, which covers gaming and physical AI, brought in $7.2 billion against expectations of $6.6 billion20.

Guidance for the current quarter was also above forecasts. Nvidia projected $108 billion, plus or minus 2%. Estimates had ranged from about $104.6 billion to $105.2 billion818. That outlook assumes no data center compute revenue from China, which makes it more conservative than it first appears3.

Supply, not demand, sets the limit

The most important thing management said may be how it framed the 70%. Coverage consistently describes it as a supply-limited figure, not a demand estimate. Huang said demand is "much greater than 70%." He said the figure reflects what Nvidia's suppliers can reliably deliver, and that a global memory shortage is the main bottleneck1112. Yahoo Finance reported Huang suggesting growth could top 100% without the memory shortage20. One wealth manager described underlying demand as closer to 100% annual growth14.

That framing does two things. It makes 70% look like a floor rather than a stretch target, and it moves the main risk from customers to suppliers. Whether Nvidia can secure enough high-bandwidth memory and manufacturing capacity is now the key variable14. The company reported about $279 billion in supply chain commitments, which bullish analysts read as visibility into fiscal 20288. The Vera Rubin platform is in full production18 and is expected to make up about 20% of data center revenue this quarter6.

Customer commitments help the case. Amazon's agreement to buy 2 million Nvidia GPUs for delivery across 2027 and 2028 makes the long-range forecast more credible13. Huang later told reporters in Scotland that Nvidia would ship twice as many chips next year, which reinforced the guidance14.

Why Huang broke with precedent

Huang presented the unusual disclosure as coordination with partners, not salesmanship. His argument is that Nvidia's chips drive so much data center construction that the companies supplying land, power and components need the same information he has11. That is plausible. It also had the effect of resetting Wall Street's models in one call. Kress's projection added roughly $200 billion to the earlier consensus14. Based on fiscal 2027 consensus of about $396 billion, 70% growth implies about $673 billion in fiscal 2028 sales. That would put Nvidia ahead of Apple and Alphabet, with only Amazon larger among US tech companies1114.

The broader message concerns who is buying. Investors have long worried that a few hyperscalers serving a few frontier labs were behind almost all of Nvidia's growth. Huang argued that the customer base has widened to include regional AI firms, neoclouds, startups, enterprises and sovereign buyers1116. He said non-hyperscaler customers now make up about half of data center revenue and are growing about 100% a year14. The segment figures above back up that claim.

Where the coverage diverges

Reports of the stock's reaction vary widely. Several outlets say shares dipped after the release and then rose once Kress gave the fiscal 2028 figure1116. Estimates of the after-hours move range from about 4%1 to nearly 5%16 to more than 5%17. Investing.com puts the full reaction at 7%1.

The follow-through tells more. About three weeks later, the stock was up only 1.3% from its pre-report level, at $212.3879. By late September it was about 6% below its May all-time high of $235.7414. Analysts have not fully adopted the new target either. Yahoo Finance data showed the Street averaging 63% growth for fiscal 2028, still below the company's 70%20.

My reading is that investors accept the demand but are discounting the forecast, mainly because of margins and financing. Gross margin is guided to about 74% this quarter, a low of 71% to 72% in the fourth quarter as memory costs rise, and 72% to 73% in fiscal 2028814. Selling much more at a lower margin is a different kind of growth than the market has paid for over the past three years.

Cash flow is another concern. Second-quarter free cash flow was $21.34 billion, about a third of GAAP net income. Bearish analysts also point to rising receivables and growing reliance on customers Nvidia helps fund8. Kress mentioned plans for selective credit enhancement covering nearly two gigawatts of compute for another frontier lab. Huang said his only regret about investing in AI labs was not doing more of it sooner11. Coverage describes this as Nvidia's defense, not an independent assessment. The structure makes Nvidia both a supplier and a funder of its own demand12. Competition from AMD, Google and custom inference chips also hangs over any multi-year forecast811.

What to watch next

The next test is November 25. Consensus calls for adjusted EPS of about $2.47 on revenue near $108.69 billion, basically in line with Nvidia's own guidance1. Another beat-and-raise there would make the 70% figure look more like a floor, as management describes it.

Valuation is less of an obstacle than the headlines suggest. One analysis noted a forward P/E of about 23.5, below the Nasdaq average8. Consensus fiscal 2028 EPS of $15.52 implies 67% earnings growth19. The bigger risk is the one any company runs after making a promise this large. One assessment put it plainly: a forecast this big leaves almost no room to disappoint11.

On balance, the August report showed that demand for AI computing is still growing and spreading to more types of buyers. Whether Nvidia hits 70% will depend more on memory suppliers and on the health of the customers it is helping to finance than on demand itself. The stock's muted performance since the report suggests investors see it the same way.