Is NVIDIA worth buying at $156.14, with enough cushion if AI spending slows?

Not enough cushion yet — HOLD, BUY window closed: on our lowered forecast the $166.62 base DCF fair value sits only 6.7% above the share price, and SemiAnalysis's revised outlook implies $129.69, pending a Q3 guidance reset.

NVIDIA · Q2 FY2027 · Published · Data cut-off: 2026-11-17 · By DeepCell AI · Sources

1 · Key Question

Our valuation cushion has narrowed on a lowered forecast, so Q3 results and the new outlook must clear before we buy


Our base DCF fair value sits only slightly above the November 16 close, so we recommend holding.

Purchase orders, startup stakes, and customer balances in Q2 filings Q2 FY2027

Supply, cloud & lease purchase orders$422.0B
Strategic startup & ecosystem equity stakes$93.9B
Unrealized Q2 equity investment gains$7.8B
Uncollected customer invoices (receivables)$63.1B

Customer lease guarantees, cash collection rates, and collection days are detailed ahead. Public filings do not disclose which buyers received credit backing.

Why we wait: Q3 earnings (Nov 17) test a forecast we have already cut once

  • The pullback did not make the stock cheaper against our model.Shares have fallen since November 6, and our re-run base DCF fair value has fallen further.
  • Sustained AI customer demand matters far more to value than worst-case lease guarantee payouts.All three stress cases value the stock below that close, and carry no probability weights.
  • Cash-flow valuation does not price the CEO change or the new export-licensing rule.We wait for Q3 results and the new outlook before revisiting the rating.

2 · Executive Summary

HOLD (BUY window closed): $166.62 base DCF fair value sits only 6.7% above the $156.14 share price


The share price fell, but our forecast fell further. The Base Case keeps a small margin; the AI Capex Peaks Early stress case, which follows SemiAnalysis’s revised outlook, puts fair value 16.9% below the November 16 close.

Fair value per share across scenarios vs. current price USD per share

Per sharevs. Current price
Wall Street average target (after November cuts)$267.65+71.4%
Optimistic ceiling (15x exit multiple)$220.50+41.2%
Base Case (DCF fair value)$166.62+6.7%
Current share price (Nov 16 close)$156.14—
AI Cloud & Enterprise Stalls (stress case)$143.46-8.1%
AI Capex Peaks Early (SemiAnalysis path, stress case)$129.69-16.9%
Big Tech Hyperscalers Only (stress case)$106.38-31.9%

Stress scenarios represent severe boundary tests rather than probability-weighted forecasts.

Why not buy now: every headline bargain rests on a forecast we just cut

  • A 10.9x FY2028E P/E ratio is only a bargain if our forecast holds.Those earnings have not been reported yet.
  • Wall Street still sees upside our model does not.Analysts are cutting targets, led by HSBC’s downgrade to Hold, yet the average remains above our upper ceiling.
  • Outside the Base Case there is no margin at all.All three stress cases value the stock below that close.

Timing. Q3 earnings (Nov 17) bring the first outlook under the interim CEO. We revisit the rating only after re-running the model on it.

3 · Recommendation: Hold, BUY Window Closed

Shares have closed below our old $182.32 buy threshold for six sessions; on our lowered forecast the threshold falls to $129.69, below today's share price


The November 16 close sits below our old buy threshold but above the re-run one: the breach prompted a model update, and the update does not support a BUY.

Current cushion above the re-run threshold & upcoming catalysts

Nov 16 close vs. re-run AI-capex-peak threshold$26.45
Q3 FY2027 earnings (tests $108.0B revenue outlook)17 November
Huang’s closed-door House testimony (as reported)18 November

The price breach prompted a model update; our lowered forecast is why it does not open a BUY. Sources: SemiAnalysis Accelerator & HBM Model update of 13 November 2026, as reported by OfficeChai (secondary; page not opened) · NVIDIA Form 8-K, Exhibit 99.1: NVIDIA Announces CEO Transition (12 November 2026) · The Hill: BIS limits remote computing exports (16 November 2026; secondary; page not opened)

4 · November Selloff, Forecast Cut & CEO Change

Record Q2 revenue of $96.2B, paired with a guided Q3 gross margin dip to 74.0%


Exhibit 1: Quarterly Revenue: Reported, Guided & Forecast USD billions, FY2027

Strong Q2 results confirm near-term AI chip demand, while leaving multi-year buyer durability open.

Quarterly revenue & margin trajectory USD billions except gross margin

FY2027 Quarterly Headline MetricsQ1 FY2027AQ2 FY2027AQ3 FY2027 guideQ4 FY2027E
Total Revenue$B81.696.2108.0121.0
Data Center$B75.289.0100.1112.3
Gross Margin %%74.9%75.0%74.0%71.5%

Tight chip supply supports near-term growth, though management's FY2028 target is not a guaranteed floor. Our Base Case puts FY2028E revenue at $650.7B, up 60.0% from $406.8B in FY2027E and below management's ~70% outlook.

Q3 revenue/margin and Q4 margin trough (71.5%, driven by memory costs) reflect guidance; Q4 revenue is projected.

5 · Q2 Results & Q3 Outlook

Emerging AI clouds and enterprises (ACIE) drove most Q2 growth — though filings do not isolate financed buyers


Exhibit 2: Data Center Revenue by Customer Group USD billions, Q1 and Q2 FY2027

Emerging AI clouds & enterprises (ACIE) delivered 58.9% of the $13.8B quarter-over-quarter Data Center jump; Big Tech hyperscalers drove the rest.

Quarter-over-quarter Data Center revenue by buyer group Q1 to Q2 FY2027

Q1 FY2027Q2 FY2027
AI Clouds & Enterprise (ACIE)$32.2B$40.3B
Big Tech Hyperscalers$43.1B$48.7B
Data Center$75.2B$89.0B

Customer growth is not proof of vendor financing: SEC filings do not link credit support to specific buyers.

ACIE groups AI cloud startups (neoclouds), sovereigns, and enterprises together. We use ACIE as a stress-test proxy rather than assuming all ACIE sales are financed.

6 · Who Drove Data Center Growth

Sustained AI customer demand matters far more to value than worst-case lease guarantee payouts


Worst-case lease guarantee payouts cost at most $4.47 per share, whereas slowing AI chip demand cuts fair value by more than ten times as much.

Per-share valuation impact by risk factor USD per share

Per-share hit
Customer lease guarantees (0% to 100% worst-case payout)$4.47
AI Capex Peaks Early (drop from base DCF value)$36.93
AI Cloud & Enterprise Stalls (drop from base DCF value)$23.16
Big Tech Hyperscalers Only (drop from base DCF value)$60.24

Stress cases represent severe downside boundaries rather than customer-level credit forecasts.

Purchase orders and guarantees are disclosed; financed customer demand is not

  • Supply commitments reflect chip production scale.Purchase orders are mainly memory and fab capacity through FY2029, not customer loans.
  • Customer lease guarantees are bounded credit backstops.We deduct an explicit dollar reserve in our valuation bridge.
  • Multi-year AI demand is the real swing factor.Emerging AI clouds (ACIE) drove most Q2 growth and serve as our demand stress-test proxy.

7 · Demand Risk vs. Guarantee Risk

Supply-chain purchase orders and customer lease guarantees carry completely different risks


Exhibit 3: Supply Purchase Orders vs. Customer Lease Guarantees USD billions, Q2 FY2027

Customer lease guarantees make up 20% of the $530.5B combined headline figure; adding them together confuses chip supply with credit risk.

Why purchase orders and credit backstops belong in separate buckets

  • Purchase commitments lock in NVIDIA's own supply chain.Includes memory and wafer supply, cloud compute contracts, and $25B in upcoming data-center leases.
  • Guarantees pay out only if backed customers default on leases.Deducted directly from equity value using an explicit payout reserve rather than lumped into operating costs.

Maximum guarantees comprise $3.5B for AI cloud partners and the remainder for OpenAI's SB Energy campus.

8 · Supply Orders vs. Customer Guarantees

Purchase obligations cluster in FY2027–FY2029 to lock in memory and manufacturing for AI systems


Component & fab supply, chiefly high-bandwidth memory, dominates obligations: normal operating scale, not customer financing.

Total commitments by category all years

Component & fab supply$279B
Cloud services$29B
AI cloud agreements$36B
Leases, equity and capex$78B

$106B comes due in FY2028E alone. Obligations stay elevated through FY2029, then taper sharply to longer-term leases and capex.

9 · Supply & Capacity Commitments

Slower Q2 cash collection is a trend to watch, not proof of customer credit trouble


Operating cash flow dipped to $24.1B vs. $59.7B in GAAP profit: profit collected in cash fell from 86.2% to 40.4%.

Exhibit 5: Operating Cash Flow vs. GAAP Net Income USD billions, Q1 FY2027A and Q2 FY2027A

Customer collection time (DSO) slowed in Q2, while remaining healthier than our warning threshold.

Exhibit 6: Customer Collection Time (Days Sales Outstanding) average days to collect, quarter-end

Track the multi-quarter trend, not one quarter's noise. Management cites chip inventory build, tax timing, and extended payment terms for creditworthy Big Tech buyers. Startup equity gains (10.9% of pretax profit) are excluded from core EPS.

10 · Cash Flow & Customer Collections

Even if every customer lease guarantee defaulted at once, the direct hit is $4.47 per share


Our guarantee reserve is the smallest adjustment in the valuation bridge, deducting $27.1B to reach $4,046.4B in equity value.

Sizing the customer guarantee reserve

Maximum potential customer lease guarantees$108.5B
Assumed payout rate (analyst estimate)25%
Dollar reserve deducted from equity value$27.1B

This reserve covers direct lease-guarantee payouts; broader demand slowdowns are tested in our stress scenarios.

Key note. No historical default data exists for AI data-center guarantees; knock-on revenue or invoice losses sit in our demand stress cases.

11 · How Guarantees Affect Fair Value

Our base DCF fair value sits only 6.7% above a share price trading between it and the $129.69 AI-capex-peak slowdown case


Exhibit 8: Fair Value per Share by Scenario vs. Current Price USD per share; Wall Street range spans lowest to highest analyst target

Our valuation range spans severe downside stress tests to an optimistic exit ceiling. We treat them as guideposts, not probability-weighted odds.

Core valuation assumptions

Discount rate (WACC)11.6%
Long-run terminal growth (Base DCF)3.5%
Optimistic ceiling (EV / EBITDA exit multiple)15x
Perpetual growth implied by exit multiple6.1%

The 15x exit multiple implies perpetual growth above GDP and serves strictly as an upper ceiling.

12 · Fair Value Across Scenarios

Three clear signals prompt a rating review; none triggers an automatic trade


Crossing any threshold triggers an updated valuation check; rating changes depend on updated fair value vs. market price.

Price and Treasury-yield triggers have both fired, and the model update they prompted lowered fair value

Share price enters slowdown rangeTriggered: shares have closed below our old AI-capex-peak value since November 9. After our forecast cut that value falls, and the November 16 close sits above it.
10-year Treasury yield risesTriggered: input updated from 4.79% to 5.10% (November 12). A 0.25 percentage-point rise in the 10-year Treasury yield trims fair value by more than all customer guarantees combined.
Soft Q3 results vs. Q3 FY2027 guideTested November 17: slower customer collections or weaker buyer mix beyond the decision-checklist thresholds, or an FY2028 outlook below our Base Case.

Where each signal stands today

Closes below old AI-capex-peak threshold6 sessions
10-year US Treasury yield input (Nov 12, updated)5.10%
Q2 jump in customer collection days+15 days

Current stance: HOLD. Both triggers prompted this update; Q3 results (Nov 17) decide whether the next one changes the rating.

13 · Three Signals That Reopen Our Rating

Hold: reopen the BUY window only if Q3 outlook, governance and valuation all clear


Warning signals that weaken the buy case

ThresholdNow
AI Clouds & Enterprise (ACIE) QoQ growthturns negative+25.2%
Customer collection time (DSO)above 70 days60 days
OpenAI / SB Energy lease guaranteeactivates ahead of schedule$105.0B maximum
Inquiry, testimony or export ruleimplicates NVIDIA compliancepending

Positive signals that strengthen the buy case

ThresholdNow
Big Tech Hyperscalers share of QoQ Data Center growthabove 50%41.1%
Customer collection time (DSO)below 55 days60 days

HOLD, BUY window closed. A lower share price alone does not reopen it; Q3 earnings and the new outlook do.

Three conditions, all required

  1. 1Nov 17: Q3 earnings meet or beat guidance.Revenue and margin match or top guidance; the FY2028 outlook holds at or above our Base Case.
  2. 2Post-Nov 18: Governance stays contained.Nothing implicates NVIDIA’s own export-control compliance.
  3. 3After the re-run: Valuation cushion returns.Updated base DCF fair value sits well above the share price.

Any signal prompts a forecast refresh before trading. Customer-level financing data and SemiAnalysis’s note at source remain blind spots.

Key test: Q3 results vs. Q3 FY2027 guide14

14 · Decision Checklist After Q3 Earnings

Sources

Not investment advice. Any rating here is this model's view of value against price at the data cut-off.

Built in DeepCell Every figure on these slides opens its formula and its source. See how a review like this is made.

Share this review

Send the link: it opens with the picture below, and every number on the page still opens its formula and source. The picture is free to download and share.

More valuation reports

All valuation reports