The crowd, in its wisdom or madness, usually rushes to reward a spectacular quarter. This time it has not. Nvidia reported fiscal Q2 2027 results after the close on August 26 108. Adjusted EPS was $2.22 against a $2.09 consensus 95,99,107,123,137,138, and revenue was $96.22 billion against expectations of $92.27 billion 119. The shares did jump 8.7%, well beyond the ±5.9% the options market had priced 110. They then set a record close of $239.24 on October 6 133 and drifted back. The latest close of $217.94 49 is about 10% below the $243.37 12-month high 102,116,143. From April 29 to October 23 the stock gained only about 1% 75,79.
This is the central tension in the material. Reported results and forward estimates keep rising, and the price has gone sideways. What follows asks what the quarter established, where its quality is open to question, and why the November report matters more than the August beat. Almost all of the evidence dates from mid-to-late October 2026, so it is current.
What the quarter established
The operating facts are widely corroborated. Revenue grew 106% year over year, a figure carried by 30 sources 28,29,30,31,32,34,43,44,52,59,63,64,65,84,87,96,118,122,127,136,144. The more precise 105.9% appears elsewhere 67,86,95,99,101,107,112,123,124,126,137,139. Lower figures of +83.4% 39 and 85% 20,35,41,42,47,71,72,73,75,77,78,79,80,89,90,91,92,94,100,102 also circulate. The first is cited in peer-comparison material, and the source material does not reconcile them. Revenue also cleared the $91 billion, ±2%, that management had guided 8,66,108. Sequential growth was about 18% 122.
The growth is concentrated. Data Center revenue was $89.0 billion, up 117% 40,43,44,52,66,113,122,125,127,136,144, or roughly 92.5% of the total 129. Networking grew 138%, faster than compute 61. Profitability remains extreme. Non-GAAP gross margin was 75.0% 18,19,83, the quarterly net margin was 62.97% 25,86,95,99,107,111,112,118,123,124,126,137,139, and net income was $59.7 billion 29,46,59.
The estimates have followed. Fiscal 2028 EPS consensus is about $15.91, up from $12.76 90 days earlier 61. In the preceding 30 days seven analysts raised estimates and none cut 61. That implies about 71% growth from roughly $9.31 in fiscal 2027 144. Sentiment is lopsided. About 98% of ratings are buys 99, and consensus targets imply roughly 47–60% upside 57,106,138.
One caution applies to the figures that circulate. The beat streak is variously described as four of four 71,77,80, five 61,73, six 138,144 or seven of the last eight 49. The direction is clear, but the count is not.
Beneath the headline: three questions of quality
The first question concerns the bottom line. GAAP EPS of $2.46 exceeded adjusted EPS of $2.22 132. Net gains from equity securities of $7.8 billion contributed to Q2 net income 46. First-half net gains on equity stakes were $23.7 billion, mostly unrealized 45. Adjusted EPS is therefore the cleaner read of operations, and the "128% GAAP EPS growth" figure in the posts should be read with that in mind 105.
The second question is cash. Q2 operating cash flow was $24.1 billion, about 40% of profit 63. Free cash flow was about $21.3 billion 29,121. Receivables stretched to about 60 days of sales from 45 61,84,87,121,135,136. These are largely single-source figures. They are directionally consistent with working capital absorbing cash during rapid scale-up. Cash generation is not weak in absolute terms: first-half operating cash flow was $74.4 billion 34,45, and the first-half free cash flow cited was $69.9 billion 132.
The third question is who is buying. One direct customer was 16% of Q2 revenue 59,62,140. Five companies each account for 10%–22% of accounts receivable 98, and another source puts five customers at 70% of receivables 98. Management has said the five largest cloud providers are a little over half of Data Center sales 50.
The ecosystem being financed
Concentration matters more because Nvidia is underwriting part of its own demand. Supply and capacity commitments reached $279 billion in the July 10-Q 85,120, up from $119.0 billion a quarter earlier 12,16,59,85,87. Other obligations include $20 billion of data-center leases intended for later assignment to third parties 76,85,117 and guarantees of up to $108.5 billion 145. Equity stakes stand at $99 billion 45. The figures are measured differently and should not be summed. They agree in direction. Management expects roughly a quarter of next year's business to come from balance-sheet-supported customers 138.
There is a counterweight. The debt/equity ratio is 0.12 3,5,7,9,13,14,17,99,102,143, and about $26 billion went to shareholders in Q2 38,40,52,127. Even so, the headline margins rest in part on customers' continued access to funding. The wider evidence points the same way. Alphabet recorded its first negative free-cash-flow quarter since its IPO in Q2 2026, a point reported by nine sources 21,22,23,115,120. Meta reported negative free cash flow in Q3 115. One author sets roughly $200 billion of expected AI-native revenue 103 against $281 billion of required cloud revenue 103. The material also cites a JPMorgan estimate of about $650 billion per year of new revenue needed 120. We have seen this pattern before, in the railway mania, where a genuine technology met capital committed faster than the end-revenue that could justify it. The comparison is mine and is not evidence. It is, however, the shape of the question the sources raise.
Guidance: strong, conditional, and not yet a margin story
Management guided fiscal Q3 revenue to $108 billion, ±2% 43,44,52,127. That is about 12% above Q2 63 and implies about 89% year-over-year growth 109. Consensus of about $109 billion sits in line with the guide 48. Beyond that, management expects roughly 70% growth in fiscal 2028 29,33,36,43,44,63,67,84,104,128,135,138. Supply is tight 61, and Jensen Huang said demand is "much greater than 70%" 87,136. The pre-call average expectation was about 44% 84. Post-call consensus is about $692 billion of fiscal 2028 revenue, up 68% 50. Interpretively, guidance pulled the Street toward management rather than the reverse.
The conditions are explicit. The outlook assumes no Data Center compute revenue from China 1,2,4,6,10,11,15,24,27,43,44,46. It also depends on customers completing data centers 44. Growth is expected to slow to about 30% in the year ending January 2029 50.
Margin is the swing factor. Q3 gross margin is guided to 74%, ±0.5 points 63, and is expected to bottom at 71%–72% in fiscal Q4 on rising memory costs 50,63,121. Price increases are scheduled for early 2027 50. One bull argument expects them to lift margins to 72%–73% 135. One unnamed-analyst claim sees compression of 10 to 15 points over 18 months 141. That is a single-source claim, far beyond anything management has guided, and I treat it as a scenario only. The guided erosion is modest and cost-driven. The remedy has not yet been tested.
The market's verdict: a multiple that has reset
Beneath the numbers lies human nature, and here it is doubt about durability. Trailing P/E readings cluster near 27, at 26.83 95,123,124,126,130,139, 27.55 60,86,107 and 27.31 58. One source says that is the lowest level since June 2019 144, against a ten-year average of 61.3 96. It also sits below a peer median of 31.2 73,80,88. Outliers of 43 67,83 and 45 135 appear to be stale or on a different basis. Forward multiples run from 13x 97 to 15x 142 on fiscal 2028 earnings, and the spread follows which year's earnings are used. The PEG ratio is contested at 1.5x in twelve sources 26,56,86,95,107,112,124,130,137,139, 1.81 37,123,126 and 0.478 101,138. That disagreement comes from the growth input, not from the fundamentals.
Not all of the weakness was Nvidia's own. Over the five trading days to October 16 the stock fell about 6.6% 54,55, while its closest peers fell about 6.1% 51. Cautionary signals exist but are thinner. Insider sales were $967.6 million over 90 days in one source 114 and $399 million over three months in another 102. Put flow stayed elevated 53. Against that, a $150 billion buyback was authorised 102.
What it implies: November is the test
The evidence suggests the market has not lost faith in the reported numbers. It is pricing their durability: the $279 billion of commitments, the memory-cost dip, customer financing and the China exclusion. Since earnings have outpaced the price, the multiple has compressed 134. That is usually the setup in which a further beat-and-raise can re-rate a stock. The opposing view is that the August beat is understood and the bar for November is higher. The sources cannot settle the matter.
The fiscal Q3 report is dated November 17 after the close in most sources 41,42,68,69,70,71,72,74,75,77,79,80,81,82,89,90,93,94,110,131. Others say November 18 78,80,131, and the date is not confirmed 122. The options market priced a ±9.1% move, about $19.70 per share 110, against a 5.1% average absolute realized move 110. It has overestimated the post-earnings move 77% of the time over 13 quarters 110. Event risk therefore looks richly priced. Guidance and margin comments often move the stock more than the reported quarter 127. That puts the 71%–72% margin floor and the early-2027 price increases at the centre of the test.
The quality questions are about cash, concentration and financing. Operating cash flow was about 40% of profit 63, and commitments reached $279 billion 120. Five customers account for 70% of receivables 98.
The price has not followed. Trailing P/E near 27 is the lowest since 2019 by one account 144, and the shares sit about 10% below their high 102,116,143.
November will show which reading is right. The report's guidance on margin and on the conditional fiscal 2028 outlook 44,63 will matter more than the headline quarter 127.