$NVDANASDAQNVIDIA CorporationEarnings Deep Dive · Q2 FY2027 · as of 2026-08-26 · reported 2026-08-26

Nvidia guided to a $108 billion quarter. The market asked who is paying for it.

A $96.2 billion print, 106% above last year, at a 75% gross margin, with a guide $4 billion above the street: the fourteenth straight quarter above Nvidia's own outlook. The stock has gone nowhere for ten months anyway, because the debate has moved from demand to financing - purchase commitments jumped to $279 billion, receivables stretched to 60 days, a $25 billion bond was sold and $108.5 billion of lease guarantees now sit behind customers who are growing faster than their balance sheets.

Market Cap
$5.08T
largest company in the world
Off the High
-11%
52-week closing high $235.47
Q3 Guide
$108.0B
street carried $103.8B
Fwd P/E
~17x
CY2027 consensus EPS, J.P. Morgan
01 — The Print

Q2 FY2027 earnings, decoded

The August 26 print beat consensus on every line and the guide beat it again; the new numbers are in the cash flow statement, not the income statement. Click a row for the read.

Revenue
$96.2B
+106% YoY
vs $92.1B consensus; +18% QoQ
Data Center
$89.0B
+117% YoY
92% of revenue; Hyperscale $48.7B, AI clouds & enterprise $40.3B
EPS (non-GAAP)
$2.22
+6% vs street
consensus $2.09; GAAP $2.46 on $7.8B equity gains
Gross Margin
75.0%
+2.5 pts YoY
flat QoQ; Q3 guided to 74.0% ± 50 bp
Free Cash Flow
$21.3B
-56% QoQ
$48.6B in Q1; working capital and cash taxes
Follow every dollar — the income statement, as a flow
Ribbon width scales with dollars. Slate = segments, green = profit, red = costs.
Hyperscale $48.7BAI Clouds & Enterprise $40.3BEdge Computing $7.2BRevenue $96.2BGross Profit $72.1BCost of Revenue $24.1BOperating Income $63.7BOperating Expenses $8.4BNet Income (GAAP) $59.7BTaxes & Other $4B
Line by line — with the why
Bars scaled to revenue. Click a row for the read.
Fourteenth straight quarter above the company's own guide; Blackwell Ultra is the vast majority of it
Memory is the swing factor: supply commitments rose to $279B, mostly HBM procurement
A 75% margin on $96B is more gross profit than Micron's entire quarterly revenue
Up 55% YoY on compute infrastructure and compensation; guided to $9.2B for Q3
66-cent operating margin; the stock's question is duration, not profitability
$11.8B tax at 16.5% net of $7.8B other income, dominated by equity-securities gains
Non-GAAP net income $54.0B strips the investment gains; that is the cleaner earnings power
Guidance
Q3 FY27 Revenue$108.0B ± 2%$103.8B street, $91.0B ± 2% (Q2 guide)
Q3 FY27 Gross Margin74.0% ± 50 bp75.0% (Q2 guide and actual)
Q3 FY27 Opex (non-GAAP)~$9.0B$8.2B (Q2 actual)
China Data Center compute$0 assumed$0 assumed; Hopper to China <1% of DC in Q2
Balance sheet
Cash & Marketable Debt$56.6B
Senior Notes Issued in Q2$25.0B
Receivables (DSO)$63.1B (60 days, from 45)
Inventory$31.6B, up $5.8B for Rubin
Purchase Commitments$279B, from $119B
Lease Guarantees (max)$108.5B
02 — The Ecosystem

Who NVIDIA buys from & sells to

Ninety-two cents of every Nvidia dollar is data center, so the whole AI supply chain reads this print. Upstream: the foundry and the memory makers it just committed $279 billion to. Downstream: the hyperscalers and AI clouds it now helps finance. Click any node.

The value chain, mapped
Hover or tap a node to light up its links
TSMTSMC000660.KSSK hynixMUMicron2317.TWHon Hai (Foxconn)MSFTMicrosoftGOOGLAlphabetORCLOracleCRWVCoreWeave$NVDAthe pivot
Tap a ticker — and I'll tell you exactly why it rises (or doesn't) with $NVDA.

Flow weight is a subjective 1-10 read of how tightly each name's results move with Nvidia's data center cycle, not a revenue rank. Private customers that matter as much (OpenAI, Anthropic, SpaceXAI) have no ticker to show.

NVDA BUYS FROM
NVDA SELLS TO ↓
Trades with the AI-capex tape
AVGO BroadcomAMD Advanced Micro DevicesMRVL Marvell
Picks and shovels for the buildout
VRT VertivETN EatonHUBB Hubbell
Neoclouds
NBIS NebiusSPCX SpaceX
03 — The Debate

Bull, base & bear case

Nobody argues about the quarter. The fight is over whether a company that is now guaranteeing its customers' data center leases still deserves a growth multiple, or whether it has quietly become the lender of last resort to its own demand.

Compute is revenue now. The buyers just needed someone to underwrite the buildings.
The capex curve still compounds
  • Guide above the bar, again $108B for October versus $103.8B carried by the street, with no China compute assumed; the fourteenth consecutive quarter above outlook.
  • Rubin arrives on time Vera Rubin is in full production at CoreWeave, Google Cloud, Azure, Oracle and Nebius, inventory is up $5.8B ahead of the Q3 ramp, and J.P. Morgan expects a 5-10% blended ASP step-up.
  • Demand is broadening AI clouds, enterprise and sovereign revenue grew 138% year over year to $40.3B, catching up to hyperscale; one lab drove last year's buildout, dozens drive this one.
  • The multiple is not the problem Around 17x CY2027 consensus EPS after ten months of sideways tape; the stock has de-rated while earnings doubled.
BULL CASE · 12-MO TARGET
$340
≈ +62% vs $210
Assumes FY28 (calendar 2027) EPS near $14 as Rubin ramps and margins hold in the mid-70s, and the market pays roughly 24x for it. Consistent with the KeyBanc, Cantor and HSBC targets of $330-360 set in the week before the print.
SUBJECTIVE ODDS
~30%
The three targets, side by side
12-month targets vs the $209.66 close. The spread tells you how two-sided this is.
BULL
$340
BASE
$290
BEAR
$180
Today ≈ $209.66
04 — Build Your Own

Price-target calculator

Pick your calendar-2027 EPS and the multiple you would pay for it. The Nvidia debate is a dispute about the second number.

Implied value = EPS × P/E multiple

IMPLIED SHARE PRICE
$211
+1% vs $210

Price target = forward EPS × P/E. The defaults ($12.40 of CY2027 EPS at 17x) roughly reproduce today's $209.66 close and match J.P. Morgan's read that the stock trades near 17x next-year consensus. First-half FY27 non-GAAP EPS was $4.09; the Q3 guide implies roughly $2.45 more.

05 — The Street

What Wall Street says

Sixty-one analysts, a Strong Buy consensus, and an average target 46% above the stock: the street has been right about the numbers and wrong about the tape for a year.

Strong Buy
61 analysts · as of 2026-08-26
95%BUY OR BETTER
Strong Buy
10
Buy
49
Hold
2
Sell
1
Strong Sell
0
HIGH TARGET
Street high, S&P Global survey · 2026-08-26
$500
AVERAGE TARGET
S&P Global consensus, 61 analysts · 2026-08-26
$306
LOW TARGET
Street low, S&P Global survey · 2026-08-26
$180
The week before the print, KeyBanc ($330), Rosenblatt ($325), Benchmark ($335), Cantor ($350) and HSBC ($360) all reiterated; J.P. Morgan kept $280 and wrote that beats alone no longer move the shares.
06 — The Tape

Technical picture

One year, one doubling of earnings, one stock that finished roughly where it started the spring: the multiple did all the moving.

Price, moving averages & momentum
Daily close with the 50- and 200-day moving averages above, RSI(14) below.
52W HIGH $235.47$164.98$182.60$200.23$217.85$235.4712mo9mo6mo3monowRSI (14)705030
LAST CLOSE
$209.66
2026-08-26
52-WK HIGH
$235.47
2026-05-14
52-WK LOW
$164.98
2026-03-30
50-DMA
$207.75
+1% vs last close
200-DMA
$195.37
+7% vs last close
RSI (14)
46
neutral
THE READ

From $181.36 a year ago to a $235.47 closing high on 2026-05-14, then a grind back to $209.66 into the print, 11% off the high while trailing earnings more than doubled. The 52-week low of $164.98 came on 2026-03-30. Four of the last five earnings days closed red despite beats, and the stock came into this one down eight of the last nine sessions. First reaction to the print: about half a percent lower in early after-hours trading. The tape is pricing the financing debate, not the quarter.

Past performance does not guarantee future results. Prices are daily closes from Yahoo Finance.

07 — The Call

Final assessment

The quarter answered the demand question. It did not answer the financing one, and the stock will not re-rate until it does.

HOLD · ADD BELOW $200

The business is compounding at 17x next year's earnings; the balance sheet is compounding faster. Own it, but let the financing debate set your entry.

CONVICTIONMedium
12-mo Base Target
$290
+38% from $209.66
Fwd P/E
~17x
CY2027 consensus EPS
Q3 Guide Step
+$11.8B
vs +$14.6B in Q2
Guarantees + Commitments
$387B
$108.5B leases, $279B supply

Judged as an operating company, this was as clean as a $96 billion quarter gets: every line above consensus, margin held at 75%, Rubin shipping on schedule, and a $108 billion guide that assumes nothing from China. Judged as a balance sheet, the quarter opened a new chapter. Purchase commitments jumped from $119 billion to $279 billion, receivables stretched to 60 days on extended terms, free cash flow halved, and Nvidia now guarantees up to $105 billion of a single customer's data center leases while raising $25 billion of debt to fund buybacks and investments.

None of that is a thesis break. It is the reason the multiple sits at 17x instead of 30x, and it is why five straight beats have produced four red earnings days. At $210 the stock discounts a lot of the financing risk already; below $200 it discounts more than the operating evidence supports. Above $240 the market would be paying for a re-rating that management's own guarantees make harder to earn.

▲ WHAT PUSHES US MORE BULLISH
  • Rubin's first full quarter A Q3 print at or above $108B with gross margin holding 74%+ despite HBM costs would show the ASP step-up is real and the margin dip is mix, not pricing power lost.
  • DSO normalizes Receivables back toward 45 days in Q3 would mean the extended terms were a one-quarter accommodation, not the new business model.
  • China reopens Any H200 compute revenue from China is pure upside to a guide that assumes zero; J.P. Morgan pegs it at roughly $3B per 100,000 units.
▼ WHAT FLIPS US CAUTIOUS
  • Guarantees phase in The Ohio guarantees become effective as sites reach service from fiscal 2029; every milestone that triggers exposure without matching customer cash flow is a rating event.
  • Custom silicon at scale Jalapeño's 2027 rollout, Google's TPUs and Broadcom's XPU pipeline all land in the same year Rubin needs its second wave of orders.
  • The second derivative If the October quarter adds fewer than $11.8B and January guides for less again, the market will call the top of the cycle whether or not it is.
The financing question, in full: Circular Reference (Macro Deep Dive #4)
This brief is one analyst's opinion, produced for research and education. It is not financial advice and no figure here is a recommendation to buy or sell. All figures are as of 2026-08-26 and come from the sources listed on this page; NVDA is held in The Desk's fictional model portfolio.