Nvidia again posted results above expectations, but Bernstein argued the bigger story sits beyond the second fiscal quarter and in the company’s read on the next product cycle and demand through calendar 2027.
In a report dated Aug. 27, Bernstein raised its price target on Nvidia to $400 from $315 and maintained an Outperform rating. The firm also increased its fiscal 2028 revenue forecast to $690.3 billion from $539.6 billion and lifted its non-GAAP earnings per share estimate for the same year to $15.80 from $12.52.
Bernstein’s case for the upgrade centers on two points: Blackwell demand is still growing quickly, and Rubin has moved into a phase of meaningful volume contribution. In the firm’s view, Rubin may set up the largest product cycle in Nvidia’s history.
Second-quarter revenue and earnings both topped expectations
Nvidia reported second-quarter revenue of $96.221 billion, above the market expectation of $92.293 billion and ahead of the company’s own $91 billion guide. Non-GAAP EPS came in at $2.22, topping the consensus estimate of $2.09.
The beat was driven mainly by the data center business. Revenue from that segment reached $89.023 billion, up 18% from the prior quarter and 117% from a year earlier. That was also above the market expectation of $85.949 billion.
Within data center, revenue from hyperscale cloud providers was $48.710 billion, up 13% quarter over quarter and 102% year over year. AI cloud, industrial and enterprise customers contributed $40.313 billion, up 25% sequentially and 138% from a year earlier. Bernstein used that mix to argue Nvidia’s expansion is not dependent only on a small group of large technology companies, as AI cloud providers and enterprise customers are also scaling their spending.
The company’s so-called edge computing businesses, including gaming, professional visualization, automotive and other operations, generated $7.198 billion in revenue, up 13% from the previous quarter and about 28% from a year earlier. That result also came in above expectations.
Profitability held largely steady. Second-quarter non-GAAP gross margin was 75%, in line with both company guidance and market expectations, while non-GAAP operating margin reached 66.5%, about 0.6 percentage points higher than the prior quarter.
Bernstein sees data center revenue topping $100 billion next quarter
Nvidia guided to third-quarter revenue of $108 billion at the midpoint, above the market expectation of $104.6 billion. Based on Bernstein’s calculations, the implied non-GAAP EPS is about $2.46, also above the market expectation of $2.36.
Bernstein estimates that most of the roughly $12 billion in incremental quarter-over-quarter revenue will again come from the data center unit. The firm said data center revenue in the third quarter could exceed $100 billion, with Rubin contributing around 20%.
That marks a shift. Rubin is no longer just a distant product expectation and is beginning to have a real effect on current-period revenue. At the same time, Blackwell remains in a growth phase, and Rubin is ramping, leaving demand for the two platforms overlapping in the near term.
Bernstein also focused on Nvidia’s much higher growth outlook for fiscal 2028, which covers most of calendar 2027. Using management’s indication of roughly 70% growth, the firm said full-year revenue could reach $680 billion to $700 billion.
Bernstein had previously projected fiscal 2028 revenue at $539.6 billion. It now puts that figure at $690.3 billion, an increase of nearly 28%. Its fiscal 2029 revenue forecast was also raised, from $623 billion to $954.1 billion.
The report noted that this outlook is still constrained by supply capacity. Management said that if supply of key components such as memory and wafers improves further, existing demand could support year-over-year revenue growth of as much as 100%. Bernstein treated that as a statement about demand potential rather than formal company guidance.
Margin pressure is real, but Bernstein said it does not break the earnings case
Compared with the revenue outlook, gross margin was the softer point in the earnings release.
Nvidia expects third-quarter non-GAAP gross margin of about 74%, below the market expectation of 74.8%. In the fourth quarter, that figure may fall to 71% to 72%, mainly because of rising memory prices. As pricing adjustments take effect, the company expects gross margin to recover to 72% to 73% in the next fiscal year, though that would still sit below its earlier view of about 75%.
Bernstein said the margin pressure is now confirmed, but still manageable. Even with memory costs rising sharply, Nvidia’s gross margin is only expected to decline by a few percentage points, which the firm said shows the company still has strong pricing power.
Under Bernstein’s updated model, Nvidia’s fiscal 2028 non-GAAP EPS will reach $15.80, up 26% from its prior forecast. Its fiscal 2029 EPS estimate was lifted to $22.28 from $14.81.
In practical terms, weaker margin may reduce some of the earnings leverage that comes with faster revenue growth. Even so, the scale of the revenue expansion is large enough that overall profit expectations have still moved sharply higher.
$279 billion in supply commitments has become a central variable
As spending on AI infrastructure grows, Nvidia’s main constraint is shifting from orders to supply.
By the second fiscal quarter, the company’s supply-related commitments had jumped to $279 billion from $119 billion in the previous quarter. Including cloud service agreements, data center leases and equity investments, total commitments reached $366 billion.
During the same period, inventory rose to $31.575 billion, and inventory days increased to 118 from about 114. Bernstein said that likely reflects advance stocking for the launch and ramp of the Vera Rubin platform.
The report said Nvidia’s balance sheet is becoming a competitive barrier alongside its technology. Commitments at this scale allow the company to secure memory, wafers, packaging and other key resources early, while also investing in cloud infrastructure and ecosystem partners.
For smaller rivals, Bernstein said, that kind of purchasing power and supply-chain control would be difficult to replicate even if they have viable chip designs. In that framing, AI chip competition is no longer only about the performance of a single chip. It now also involves capital, capacity, system delivery and ecosystem expansion.
Why Bernstein thinks $400 is achievable
Bernstein’s new $400 target is based on average non-GAAP EPS of $19.04 across fiscal 2028 and fiscal 2029, applying a valuation multiple of about 21 times earnings. The target moved higher mainly because Bernstein sharply increased its earnings forecasts, not because it assigned a richer valuation multiple. In fact, the previous framework used a 25 times multiple.
Using Nvidia’s Aug. 26 closing price of $209.66, the $400 target implies about 91% upside.
That valuation still rests on several assumptions: Rubin needs to ramp on schedule, the supply chain needs to support rapid growth, higher memory prices cannot keep eroding gross margin, and major cloud providers must remain willing to sustain heavy AI capital spending.
Bernstein’s bottom line is that Nvidia’s recent share-price stagnation does not mean its growth cycle has peaked. As Rubin starts contributing revenue, the market’s earnings estimates for calendar 2027 may face a concentrated upward reset. From here, the bigger question is less whether demand exists and more whether Nvidia can secure enough supply to turn a large order book into recognized revenue.