Data & Sources
  As of:      2026-01-25 (FY2026 year-end figures) · shares as of 2026-08-21 · price as of 2026-10-09 close
  Source:     SEC EDGAR 10-K FY2026, filed 2026-02-25, accession 0001045810-26-000021 (`data/filings/NVDA/NVDA_2026_10-K.txt`) · XBRL companyfacts pack (`data/fixtures/NVDA.md`) · NASDAQ close $229.28 (from the pack)
  Retrieval:  pasted by user (local files; no browsing)
  Confidence: HIGH for reported FY2026 figures · MEDIUM for valuation (the fundamentals are 8.5 months old, and no FY2027 10-Qs were in the files)

> ⚠️ **Some data isn't in these files.** I couldn't find the 52-week range, any FY2027 quarterly results (Q1/Q2 10-Qs), consensus estimates, Form 4 trades, the proxy (Item 13 related parties), or the prior-year 10-K text. Anything that depends on them is marked *unavailable* below. The valuation multiples compare an October 2026 price with fiscal-year earnings that ended in January 2026.

---

## Executive Summary

NVIDIA's FY2026 10-K is a very clean filing for a company this size. Revenue was **$215.9B, up 65.5%** (Data Center $193.7B, +68%). Operating income was $130.4B and free cash flow (FCF) was **$96.7B, a 44.8% FCF margin**. PwC gave an unqualified opinion, the ICFR conclusion is effective, there are no restatements, and the filing landed 31 days after year-end. The balance sheet is effectively debt-free: **$54.1B net cash** once marketable securities are included. The data pack's −$2.1B "net debt" counts only cash and equivalents, so it understates the cash position by about $52B.

The filing's quality is high, but its risk profile changed. FY2026 brought:
- **Heavy dependence on two customers:** two direct customers were 36% of revenue (22% + 14%).
- **Inventory up 112%** while revenue rose 65%.
- **$95.2B of supply and capacity commitments**, plus $27B of cloud-service commitments.
- **A $17B "license" with Groq** that put $14.4B of goodwill on the books.
- **$17.5B put into private, often customer-linked AI companies.** Those holdings, together with Intel and other public stakes, produced $8.9B of the year's investment gains.

The main point: core economics are excellent, and GAAP earnings are slightly flattered by mark-to-market gains. The new risks are concentration and an ecosystem that NVIDIA increasingly finances itself. The risk is not accounting quality.

## Financial Health Dashboard

```
┌─────────────────────────────────────────────────────────────────┐
│  FINANCIAL REPORT ANALYSIS — NVDA 10-K FY2026 (FYE 2026-01-25)  │
├────────────────────────┬────────────────────────────────────────┤
│ Revenue Growth         │ +65.5% YoY  (organic; Groq immaterial) │
│ Gross Margin           │ 71.1%  (vs. prior: 75.0%  Δ −390 bps)  │
│ Operating Margin       │ 60.4%  (vs. prior: 62.4%  Δ −200 bps)  │
│ FCF Margin             │ 44.8%  (FCF Conversion: 80.5%)         │
│ Net Debt / (Cash)      │ $(54,088)M  (Debt/EBITDA: 0.06x)       │
│ DSO                    │ 65 days  (vs. prior: 65 days)          │
│ Management Tone        │ Positive, with expanded hedging        │
├────────────────────────┼────────────────────────────────────────┤
│ Filing Quality Score   │ 8 / 10                                 │
│ Earnings Quality Score │ 7 / 10                                 │
│ Balance Sheet Score    │ 9 / 10                                 │
└────────────────────────┴────────────────────────────────────────┘
```

## Phase 1 — Document Orientation

| Item | Finding |
|---|---|
| Filing | 10-K, fiscal year ended 2026-01-25 (52-week year) |
| Auditor / opinion | PricewaterhouseCoopers LLP. Unqualified ("present fairly, in all material respects"); ICFR audited and effective |
| Critical audit matter | Inventory provisions and excess purchase commitments, which depend on demand assumptions |
| Auditor change | None disclosed |
| Timeliness | Filed 2026-02-25, 31 days after year-end. Timely |
| Restatements | The cover-page error-correction / clawback box is unchecked; none disclosed |

## Phase 2 — MD&A

**2.1 Revenue by market and geography ($M)**

| Line | FY2026 | FY2025 | YoY | MD&A commentary |
|---|---|---|---|---|
| Data Center | 193,737 | 115,186 | +68.2% | Blackwell was the majority; compute +59%, **networking +142%** (NVLink, Ethernet, InfiniBand) |
| Gaming | 16,042 | 11,350 | +41.3% | Blackwell; **"supply constraints to be a headwind… Q1 FY27 and beyond"** |
| Pro Visualization | 3,191 | 1,878 | +69.9% | Blackwell, DGX Spark |
| Automotive | 2,349 | 1,694 | +38.7% | Self-driving platforms |
| OEM & Other | 619 | 389 | +59.1% | — |
| US (customer HQ) | 149,617 | 77,482 | +93.1% | 69% of revenue |
| Taiwan | 42,345 | 23,600 | +79.4% | — |
| **China incl. HK** | **19,677** | **25,048** | **−21.4%** | Down to 9.1% of revenue from 19.2%; H20 export license |
| Other | 4,299 | 4,367 | −1.5% | — |
| **Total** | **215,938** | **130,497** | **+65.5%** | Segments: C&N +67%, Graphics +57% |

- **Organic or acquired?** Organic. Groq was a technology license plus hired employees, with "no customer contracts, existing products" (Note 2). Acquisitions cost $1.5B net cash, and pro forma results were "not material."
- **Mix:** a move to full-rack Blackwell systems carries lower gross margin than Hopper HGX boards (MD&A).
- **Backlog / RPO:** RPO on contracts longer than one year is only **$2.3B** (42% recognized within 12 months). Deferred revenue additions were $11.1B, up from $5.1B. Hardware backlog isn't disclosed, so forward visibility from the filing is thin.

**2.2 Margins**

| Metric | FY2026 | FY2025 | Δ | MD&A explanation |
|---|---|---|---|---|
| Gross margin | 71.1% | 75.0% | −390 bps | Hopper → Blackwell systems, plus a **$4.5B H20 charge** |
| GM excluding H20 charge* | 73.2% | 75.0% | −180 bps | The underlying mix effect |
| Operating margin | 60.4% | 62.4% | −200 bps | Opex +41% against revenue +65%, so opex leverage held |
| Op margin excl. H20* | 62.5% | 62.4% | +10 bps | Flat on an underlying basis |
| Net margin | 55.6% | 55.8% | −20 bps | Held up by $9.0B of other income |
| FCF margin | 44.8% | 46.6% | −180 bps | Working-capital build |

*My adjustment, adding back the disclosed $4.5B charge. The 10-K itself presents no non-GAAP figures.

Inventory provisions were $7.2B gross against $3.7B, with net GM drag of 2.6 pts against 2.3 pts. Even excluding H20, provisions rose, which fits the CAM focus.

**2.3 Liquidity and capital resources**

- **Cash plus marketable securities: $62.6B.** Marketable securities now include public equities, and $10.5B of them are under lock-up (Note 7).
- **Debt:** $8.468B of senior notes. $1.0B is due within one year, $2.75B in one to five years, and $4.75B after that. The covenants are non-financial. The commercial paper program was raised from $575M to **$25.0B**, with nothing outstanding.
- **Debt-like item:** $4B still owed to Groq within one year, sitting in accrued liabilities.
- **OCF:** $102.7B. Capex $6.0B; FCF $96.7B.
- **Capital return:** $40.1B cash buybacks (282M shares) and $974M dividends, which together are 42.5% of FCF. A further **$7.9B** went to employee stock-plan tax withholding, which is effectively a cash cost of SBC. $58.5B of buyback authorization remains.
- **Off-balance-sheet commitments (Note 12):**
  - $95.2B manufacturing, supply and capacity, "substantially all" payable through FY2027
  - $27B multi-year cloud services
  - $11.4B investment commitments
  - $3.4B other
  - $3.5B lease guarantees to early-stage partners, with $712M in escrow

**2.4 Forward-looking language**

| Item | Language | Strength |
|---|---|---|
| Revenue | No numeric guidance in the 10-K (quarterly guidance comes in earnings releases, which aren't in these files) | Unavailable |
| Margin | None | Unavailable |
| Capex | "expect to increase capital expenditures in fiscal year 2027" | Directional |
| Gaming | "supply constraints to be a headwind… Q1 FY27 and beyond" | Specific (negative) |
| Product timing | "one-year product cadence, including our Rubin platform"; GB300 shipping since Q2 FY26 | Specific |
| China | H200 license granted Feb 2026; "have not generated any revenue… do not yet know whether any imports will be allowed into China"; 25% tariff applies | Hedged |
| OpenAI | "finalizing an investment and partnership agreement… no assurance" | Hedged |

## Phase 3 — Financial Statements

**3.1 Income statement quality ($M)**

| Line | FY2026 | FY2025 | Δ | Note |
|---|---|---|---|---|
| Revenue | 215,938 | 130,497 | +65.5% | |
| COGS | 62,475 | 32,639 | +91.4% | Includes the $4.5B H20 charge |
| Gross profit | 153,463 (71.1%) | 97,858 (75.0%) | +56.8% | |
| R&D | 18,497 (8.6%) | 12,914 (9.9%) | +43% | Compute and infrastructure +79% |
| SG&A | 4,579 (2.1%) | 3,491 (2.7%) | +31% | |
| Operating income | 130,387 (60.4%) | 81,453 (62.4%) | +60% | |
| Interest expense | 259 | 247 | | Coverage **503x** |
| **Other income, net** | **9,022** | **1,034** | **+$8.0B** | **Unrealized gains on private and public equity, including Intel** |
| Tax rate | 15.1% | 13.3% | | Includes a $711M valuation-allowance release (benefit); OBBBA effects recognized |
| Net income | 120,067 | 72,880 | +65% | |
| Diluted EPS | $4.90 | $2.94 | +67% | Diluted shares −1.2% |

- **Quality:** about 6.4% of pre-tax income ($9.0B of $141.5B) comes from mark-to-market gains, which don't recur and can reverse. Stripping out the $8.9B of equity gains (pre-tax), EPS growth would be close to the 60% operating-income growth, not 67%.
- **Revenue recognition:** no policy change. AR grew 66.8% against revenue at 65.5%, so DSO was flat (65.0 vs 64.5 days). No channel-stuffing signal.

**3.2 Balance sheet**

```
Asset Quality:
  Cash + marketable securities:  $62,556M  (30.2% of assets)
  Accounts Receivable:           $38,466M  DSO: 65 days (vs. prior: 65)
  Inventory:                     $21,403M  DIO: 125 days (vs. prior: 113)  ← +112% YoY
     WIP 8,822 (from 3,399) · Finished goods 8,774 (from 3,273) · Raw 3,807
  Goodwill & Intangibles:        $24,138M  (11.7% of assets; goodwill 20,832, up from 5,188)
  Non-marketable equity:         $22,251M  (from 3,387; 10.8% of assets, level-3-style marks)
  Deferred tax assets:           $13,258M

Liability Structure:
  Short-Term Debt:               $999M
  Long-Term Debt:                $7,469M
  Total Debt:                    $8,468M  (+$4B Groq payable, debt-like)
  Net Debt/(Cash):               $(54,088)M  (excluding the Groq payable)
  Interest Coverage:             503x
  Debt/EBITDA:                   0.06x  (EBITDA ≈ $133.2B = EBIT + D&A)

Equity Quality:
  Retained earnings:             $146,973M, growing (from 68,038)
  Buybacks charged to RE:        $40,158M in FY26
  AOCI:                          $178M, immaterial; no pension exposure
```

**3.3 Cash flow, the check on earnings**

```
Operating Cash Flow:        $102,718M
Less: Capex:                 $(6,042)M
Free Cash Flow:              $96,676M   FCF Margin: 44.8%
FCF Conversion:              80.5%      (prior: 83.5%; ~87% vs NI ex-equity gains, pre-tax)
FCF after Groq cash:         $83,676M   (69.7% of NI, if the $13.0B license is treated as capex)

Working Capital:
  AR change:                 $(15,399)M  use
  Inventory change:          $(11,324)M  use  ← biggest drag
  AP change:                 +$3,096M    source
  Accrued/other current:     +$5,257M    source
  Net WC impact:             ≈ $(15,949)M (incl. prepaid & LT liabilities)

SBC Add-Back:                $6,386M  (3.0% of revenue, down from 3.6%; below the 5% threshold)
D&A Add-Back:                $2,843M  (1.3% of revenue; capex/D&A = 2.1x, investing above maintenance)
Non-cash gains deducted:     $(8,918)M
```

FCF above 80% of net income passes the quality bar. The $13.0B Groq payment and the $17.5B of private investments sit in investing, below the FCF line, so reported FCF overstates the cash actually left after "strategic" spending.

## Phase 4 — Risk Factors and Legal

The prior-year 10-K text isn't in the files, so a strict new/removed comparison is **unavailable**. The items below are ones the FY2026 filing itself describes as FY2026 events:

```
NEW in FY2026 (by content):
  1. Groq IP license: "significant, nonrefundable payments"; integration may not deliver (Business/Tech)
  2. Multi-year cloud-service agreements ($27B): timing "changed and may continue to shift" (Business)
  3. OpenAI investment/partnership "no assurance" of completion (Business/Counterparty)
  4. Lease guarantees to early-stage partners ($3.5B) and a "commercial arrangements… counterparty risks" factor (Credit)
  5. H200 China license with a 25% US import tariff; China import approval unknown (Regulatory)
  6. Customers' data-center, energy and capital constraints named as a limit on deployments (Macro)
  7. Open-source models running on competitors' platforms (Competition)

ELEVATED:
  1. Customer concentration: top direct customer 22% (from 12%); top two 36%. One AI lab is
     "a meaningful amount" of revenue through cloud intermediaries
  2. Supply/demand mismatch: the H20 charge shows the risk already happened

REMOVED: unavailable (needs the prior 10-K text)
```

- **Going concern / covenant breach:** none. Covenants are non-financial and in compliance.
- **Legal:** the 2018 securities class action (*In re NVIDIA Corp. Securities Litigation*, 4:18-cv-07669) is still pending, and the derivative suits (N.D. Cal., D. Del., Del. Chancery *Horanic v. Huang*) are stayed. The filing gives no quantified exposure. Separately, unrecognized tax benefits total $4.0B, and the IRS is examining FY2023–24.

## Phase 5 — Accounting Policies and Footnotes

| Estimate | Assessment |
|---|---|
| Revenue recognition | Standard ASC 606; distinct performance obligations; no change |
| Inventory / purchase commitments | **The highest-judgment area and the CAM.** $95.2B of commitments against $21.4B of inventory leaves a lot exposed to a demand shift |
| **Goodwill (Groq)** | **Unusual.** A *non-exclusive license plus hired staff* was booked as a business combination: $14.4B goodwill and $2.5B intangible (5-year life), "valued using a cost-to-recreate methodology." Goodwill is "primarily attributable to the workforce." The $14.4B won't amortize, so it never hits operating income unless impaired. The $2.5B intangible adds about $0.5B a year of amortization |
| Non-marketable equity | Measurement alternative (cost ± observable price changes). The $22.3B book value grew roughly 6.5x. Many investees are **customers** ("AI model makers that purchase our products"), which links the gains to revenue |
| Deferred tax assets | $711M **release** of valuation allowance, a positive sign |
| Pension / OPEB | None material |
| Related parties | Item 13 is incorporated from the proxy: **unavailable** |
| Segments | Unchanged (Compute & Networking, Graphics) |

## Phase 6 — Management Tone

```
Section              Tone (1–5)   Key language
──────────────────────────────────────────────────────────────────────────
MD&A Overview          5          "data center scale AI infrastructure company reshaping all industries"
Revenue Discussion     5          "major platform shifts – accelerated computing and AI"
Margin Discussion      3          Margin decline attributed to model transition + H20; candid on provisions
Guidance Language      3          Directional only; Gaming "headwind"; China "do not yet know"
Risk Factors Framing   3          Many new "no assurance" clauses (Groq, OpenAI, investments, guarantees)
CEO Letter             n/a        Not in a 10-K
──────────────────────────────────────────────────────────────────────────
Overall Tone:        4 / 5
Change vs. Prior:    Unavailable (no prior 10-K text). Hedging density around ecosystem deals visibly up.
```

Hedging flags: "there can be no assurance that we will realize a return" (private investments), "may not become profitable in the near term, or at all" (investees), "may not deliver anticipated benefits" (cloud agreements). The negatives are stated in active voice, so I don't see evasive phrasing.

**Guidance accuracy scorecard (Phase 6.3) and the YoY/sequential quarterly comparison (Phase 7):** unavailable, because there are no earnings releases or 10-Qs in the files.

## Phase 8 — Insider Activity

From Item 9B, the only insider data in the files:
- **CFO Colette Kress** adopted a 10b5-1 plan (2025-12-18) to sell up to **500,000 shares** through 2027-03-23.
- **Director John Dabiri** adopted a plan for about 3,984 shares.

Scheduled 10b5-1 sales carry little signal. Open-market buys and sells (Form 4) are **unavailable**.

## Valuation Snapshot (price $229.28, 2026-10-09)

| Metric | Value | Note |
|---|---|---|
| Market cap | ~$5.53T | 24,100M shares × $229.28 |
| EV | ~$5.47T | Less $54.1B net cash |
| P/E (FY2026 EPS $4.90) | 46.8x | Earnings are about 8.5 months old; trailing-twelve-month EPS is unavailable |
| EV / FY26 revenue | 25.3x | |
| EV / FY26 EBIT | 42.0x | |
| FCF yield | 1.75% | |

These multiples look backward. If FY2027 has grown at anything like FY2026's pace, the real multiple is lower, but nothing in these files lets me check that.

## Key Insights

**Positives (bullish signals):**
- 65.5% revenue growth at a 60.4% operating margin, all organic (MD&A, FY2026 Summary).
- Networking +142% shows the platform is broadening beyond GPUs (Segment discussion).
- Underlying operating margin was flat at about 62.5% excluding H20, despite the shift to full systems.
- FCF of $96.7B, 80.5% conversion; SBC fell to 3.0% of revenue (Cash Flow Statement).
- $54.1B net cash, 503x interest coverage, no going-concern, covenant or auditor issues (Liquidity; Item 9A).
- Share count down 0.7% even after SBC; $58.5B of buyback authorization left.

**Negatives (bearish signals):**
- Two customers are 36% of revenue, and the top one rose from 12% to 22% (Concentration of Revenue).
- Inventory up 112% against revenue +65%, DIO up 12 days. A $4.5B write-down already happened with H20 (Balance sheet; Note 9).
- **Circular-financing exposure:** $17.5B into private AI firms that buy NVIDIA products, $3.5B of lease guarantees to early-stage partners, a pending OpenAI deal, and $27B of cloud capacity bought back from customer-CSPs (MD&A; Note 12).
- $8.9B of GAAP profit is unrealized investment gains, which can reverse (Other income).
- The Groq "license" was booked as $14.4B of non-amortizing goodwill (Note 2).
- China revenue fell 21%, and H200 sales there are uncertain and carry a 25% tariff.

**Neutral / watch items:**
- $95.2B of supply commitments due through FY2027: watch inventory provisions and the CAM.
- The $25B commercial paper program is undrawn. Ask why the capacity is being set up.
- FY2027 capex is guided up. The IRS is examining FY2023–24, with $4.0B of uncertain tax positions.

## Red Flags and Anomalies

| # | Red Flag | Severity | Where Found | Implication |
|---|---|---|---|---|
| 1 | Two direct customers = 36% of revenue; top one up from 12% to 22% | HIGH | MD&A, Concentration of Revenue | One customer's capex pause would move consolidated results |
| 2 | Vendor-financing pattern: investments in customers, partner lease guarantees, CSP cloud buy-backs | HIGH | MD&A Recent Developments; Notes 7, 10, 12 | Part of the demand may be NVIDIA-funded; counterparties are often early-stage |
| 3 | Inventory +112% vs revenue +65%; $95.2B purchase commitments; inventory provisions are the CAM | MED | Balance sheet; Note 12; Auditor's report | Write-down risk on an architecture transition or demand shift (H20 precedent) |
| 4 | Groq license booked as $14.4B goodwill (workforce), "cost-to-recreate" valuation | MED | Note 2 | Never amortized through earnings; $13B cash sits outside FCF |
| 5 | $8.9B non-cash equity gains in net income; $10.5B of locked-up public equity | MED | Other income; Cash Flow; Note 7 | EPS growth slightly above core growth; reversal risk |
| 6 | China revenue −21%; H200 license comes with a 25% tariff and uncertain imports | LOW | Geographic note; MD&A | Mostly already in FY26 numbers; mainly upside optionality now |
| 7 | 2018 securities class action still pending | LOW | Note 12 | Long-running; no quantified exposure |

## Accounting Quality Score

| Criterion | Score (0–3) | Notes |
|---|---|---|
| Revenue recognition clarity | 3 | Standard ASC 606; DSO flat; no policy change |
| Non-GAAP reconciliation quality | 2 | No non-GAAP in the 10-K; GAAP net income includes $8.9B mark-to-market gains |
| FCF conversion rate | 2 | 80.5%, at the threshold; ~70% if Groq cash counts as capex |
| Working capital trends | 1 | Inventory doubled; $27.6B of combined AR and inventory build |
| Footnote transparency | 3 | Commitments, guarantees, Groq and investments all disclosed in detail |
| Auditor opinion | 3 | PwC unqualified; ICFR effective; timely filing |
| Related party transactions | 2 | Item 13 deferred to the proxy, which is unavailable; customer-investee overlap is related-in-substance |
| **Total** | **16 / 21** | **Average, upper end.** Clean books; the questions are about business structure |

---

## Thesis Invalidation

**If signal is BULLISH — thesis breaks if:**
- auditor issues going concern warning OR material restatement announced
- Inventory provisions plus purchase-obligation charges exceed **4% of revenue** in any fiscal year (FY26 gross was 3.3%), or inventory keeps growing more than 2x faster than revenue for another year
- The top direct customer's share rises **above 30%**, or any top-two customer cuts purchases more than 25% year on year. Also if non-marketable equity marks are **written down more than $5B**, which would show the investee ecosystem is cracking

**If signal is BEARISH — thesis breaks if:**
- new auditor clears all concerns AND restatement resolved favorably
- Revenue growth stays above 40% with GM back at or above 73% and DIO back under 100 days. That would show the inventory build was demand-backed
- Customer concentration falls (top two under 25%) while Data Center growth continues

**Re-run this analysis when:**
- [ ] Next earnings release (the Q2/Q3 FY2027 10-Qs aren't in these files; load them first)
- [ ] Price moves ±15% from $229.28
- [ ] 60 days have elapsed (by 2026-12-10)
- [ ] Material news event (OpenAI deal close, China H200 import decision, Groq integration, a large customer capex change)

## Investment Signal

```
╔══════════════════════════════════════════════╗
║              INVESTMENT SIGNAL               ║
╠══════════════════════════════════════════════╣
║ Signal:      BULLISH                         ║
║ Confidence:  MEDIUM                          ║
║ Horizon:     LONG-TERM                       ║
║ Score:       6.8 / 10                        ║
╠══════════════════════════════════════════════╣
║ Action:      HOLD                            ║
║ Conviction:  MODERATE                        ║
╚══════════════════════════════════════════════╝
```

Score Guide: 8.0–10.0 Strongly Bullish | 6.0–7.9 Moderately Bullish | 4.0–5.9 Neutral | 2.0–3.9 Moderately Bearish | 0.0–1.9 Strongly Bearish
Confidence: HIGH (strong data, clear signals) | MEDIUM (mixed signals) | LOW (limited data, conflicting signals)
Horizon: SHORT-TERM (1 week–3 months) | MEDIUM-TERM (3 months–1 year) | LONG-TERM (1+ years)

The signal is BULLISH because the filing is strong: 16/21 accounting quality, 44.8% FCF margin, $54B net cash. The action is HOLD rather than BUY for three reasons. Customer concentration and customer-financing exposure are rising. The price implies 46.8x earnings that are 8.5 months old. And no FY2027 quarterly data was in the files to confirm the trend. Confidence is MEDIUM because of that data gap, not because of doubts about the reported figures.

**Disclaimer:** Educational analysis only. Not financial advice.

# Fact Check: `output/nvda/01-financial-report-analyst.md` (NVDA 10-K FY2026)

```
Data & Sources
  As of:      2026-01-25 (FY2026 fiscal year end) · cover-page shares as of 2026-02-20 · price as of 2026-10-09 close (from the pack)
  Source:     NVIDIA Form 10-K for the FY ended 2026-01-25, filed 2026-02-25, accession 0001045810-26-000021
              (data/filings/NVDA/NVDA_2026_10-K.txt, plus the .json filing metadata) · XBRL data pack data/fixtures/NVDA.md
              (scripts/fetch-fundamentals.js, retrieved 2026-10-11)
  Retrieval:  pasted by user (local files; no browsing, as instructed)
  Confidence: HIGH for reported figures (Tier 1, checked line by line) · LOW–MEDIUM for price and share count (user-supplied pack only)
```

> ⚠️ **No live market data was used.** Every claim was checked against the two user-supplied files. The $229.28 price and the 24,100M share count come only from the data pack, so verification of those stops at the pack. They were not independently confirmed.

**Verification summary:** 114 claims · 85 verified · 22 recomputed (22 agree) · 2 mismatches · 1 unverifiable · 4 stale (flagged on recomputed multiples) · 4 qualitative (2 supported, 1 qualified, 1 unsupported) · **Verification Score 8.5 / 10 · Caps fired: none**

---

## 1. Signal-driving claims

| ★ | Claim | Verdict | Source · location | Delta |
|---|---|---|---|---|
| ★1 | Revenue $215.9B, +65.5% | ✅ | [1] Income Statement p.51: 215,938 / 130,497 → +65.47% | 0 |
| ★2 | FCF $96.7B, 44.8% margin | ✅ + 🔁 agrees | [4] Cash Flows p.55: OCF 102,718 − capex 6,042 = 96,676; ÷ 215,938 = 44.77% | 0 |
| ★3 | $54.1B net cash | 🔁 agrees, with a definition caveat | [2] p.53 / [7] p.42: 62,556 − 8,468 = 54,088 | 0. It includes $12.9B of public equities, $10.5B of them locked up. Excluding equities it is **$41.2B** |
| ★4 | P/E 46.8x (FY26 EPS $4.90) | 🔁 agrees · 🕒 | [1] diluted EPS $4.90; 229.28 ÷ 4.90 = 46.79x | 0. EPS is 8.5 months old, which the report discloses |
| ★5 | Two direct customers = 36% (22% + 14%); top one up from 12% | ✅ | [6] Concentration of Revenue p.41; [13] Note 16 p.78 | 0 |

The other number the signal rests on, the 16/21 accounting-quality total, sums correctly (3+2+2+1+3+3+2).

---

## 2. Claim ledger

Type codes: R = reported · D = derived · M = market · F = fact · Q = qualitative. ★ marks signal-driving claims. References [n] are listed in §8.

| # | Claim (report §) | Type | Verdict | Source · location |
|---|---|---|---|---|
| | **Orientation (Ph 1)** | | | |
| 1 | FYE 2026-01-25, 52-week year | F | ✅ | [8] Note 1 "Fiscal Year" p.56 |
| 2 | Filed 2026-02-25, 31 days after FYE | F/D | ✅ | Filing metadata (.json); audit report dated 2026-02-25 [14] |
| 3 | PwC unqualified; ICFR effective | F | ✅ | [14] pp.49–50; Item 9A p.45 |
| 4 | CAM = inventory provisions and excess purchase commitments | F | ✅ | [14] p.50 |
| 5 | No auditor change | F | ✅ | [14] "auditor since 2004" |
| 6 | Restatement / clawback boxes unchecked | F | ✅ | [17] cover |
| | **Revenue (Ph 2.1)** | | | |
| 7★ | Revenue 215,938 / 130,497, +65.5% | R | ✅ | [1] |
| 8 | Data Center 193,737 / 115,186, +68.2% | R | ✅ | [13] p.79 |
| 9 | DC compute +59%, networking +142% | R | ✅ | [6] p.40; [13] 162,361/102,196 and 31,376/12,990 |
| 10 | Gaming 16,042 / 11,350, +41.3%, plus supply-headwind quote | R | ✅ (quote abridged, see §5) | [13]; [5] p.37 |
| 11 | ProViz 3,191 / 1,878, +69.9% | R | ✅ | [13] |
| 12 | Automotive 2,349 / 1,694, +38.7% | R | ✅ | [13] |
| 13 | OEM & Other 619 / 389, +59.1% | R | ✅ | [13] |
| 14 | US 149,617 / 77,482, +93.1%, 69% of revenue | R | ✅ | [13] p.78 |
| 15 | Taiwan 42,345 / 23,600, +79.4% | R | ✅ | [13] |
| 16 | China incl. HK 19,677 / 25,048, −21.4%; 9.1% of revenue vs 19.2% | R/D | ✅ | [13] |
| 17 | Other 4,299 / 4,367, −1.5% | R | ✅ (−1.56%, within 0.1 pp) | [13] |
| 18 | C&N +67%, Graphics +57% | R | ✅ | [6] p.40 |
| 19 | Groq: no customer contracts; acquisitions $1.5B net; pro forma not material | F | ✅ | [8] Note 2 p.61; [4] 1,535 |
| 20 | RPO $2.3B, 42% within 12 months | R | ✅ | [10] Note 9 p.69 |
| 21 | Deferred revenue additions $11.1B vs $5.1B | R | ✅ | [10] 11,137 / 5,083 |
| | **Margins (Ph 2.2)** | | | |
| 22 | GM 71.1% vs 75.0%, −390 bps | R | ✅ | [5] p.37 |
| 23 | H20 charge $4.5B | R | ✅ | [5] p.36; [6] p.41 |
| 24 | GM excl. H20 73.2%, −180 bps | D | 🔁 agrees | 157,963 ÷ 215,938 = 73.15% |
| 25 | Op margin 60.4% vs 62.4%, −200 bps | R | ✅ | [6] % table p.40 |
| 26 | Opex +41% | R | ✅ | [5] |
| 27 | Op margin excl. H20 62.5%, +10 bps | D | 🔁 agrees | 62.47% vs 62.42% (+5 bps unrounded) |
| 28 | Net margin 55.6% vs 55.8% | R | ✅ | [6] |
| 29★ | FCF margin 44.8% vs 46.6%, −180 bps | D | 🔁 agrees | 60,853 ÷ 130,497 = 46.63% |
| 30 | The 10-K presents no non-GAAP figures | F | ✅ | "non-GAAP" does not appear anywhere in the filing |
| 31 | Provisions $7.2B vs $3.7B; net GM drag 2.6 vs 2.3 pts | R | ✅ | [6] Gross Margin p.41 |
| 32 | "Even excluding H20, provisions rose" | D | ⚠️ **MISMATCH** | [6]: 7.2 − 4.5 = **2.7 vs 3.7, a fall** |
| | **Liquidity (Ph 2.3)** | | | |
| 33 | Cash + MS $62.6B; 30.2% of assets | R/D | ✅ | [7] p.42; [2] 62,556 ÷ 206,803 |
| 34 | $10.5B under lock-up | R | ✅ | [9] Note 7 fn.1 p.65 |
| 35 | Notes $8,468M; maturities 1.0 / 2.75 / 4.75B; non-financial covenants | R | ✅ | [7] p.44; [11] Note 11 p.70 |
| 36 | CP program $575M → $25.0B, none outstanding | R | ✅ | [11] |
| 37 | $4B Groq payable in accrued liabilities | R | ✅ | [8] Note 2; [10] accrued purchase consideration 3,921 |
| 38★ | OCF 102,718; capex 6,042; FCF 96,676 | R/D | ✅ | [4] |
| 39 | Buybacks $40.1B, 282M shares | R | ✅ cash-flow basis | [4] 40,086; [3] 282M. MD&A states $40.4B (equity statement 40,388) |
| 40 | Dividends $974M | R | ✅ | [4] |
| 41 | Buybacks + dividends = 42.5% of FCF | D | 🔁 agrees | 41,060 ÷ 96,676 = 42.47% |
| 42 | $7.9B employee stock-plan tax withholding | R | ✅ | [4] 7,948 |
| 43 | $58.5B buyback authorization remaining | R | ✅ | [7] p.43 |
| 44 | $95.2B supply commitments, "substantially all" through FY27 | R | ✅ | [12] Note 12 p.70 |
| 45 | $27B cloud commitments | R | ✅ | [12] |
| 46 | $11.4B investment commitments | R | ✅ | [12] |
| 47 | $3.4B other commitments | R | ✅ | [12] |
| 48 | $3.5B lease guarantees, $712M escrow | R | ✅ (located in Note 10, not Note 12) | [11] Note 10 p.69 |
| | **Forward-looking (Ph 2.4)** | | | |
| 49 | "expect to increase capital expenditures in fiscal year 2027" | F | ✅ | [7] p.44 |
| 50 | One-year cadence / Rubin; GB300 since Q2 FY26 | F | ✅ | [5] p.36 |
| 51 | H200 license Feb 2026; no revenue; 25% tariff | F | ✅ | [5] p.36 |
| 52 | OpenAI "no assurance" | F | ✅ | [7] p.44 |
| | **Income statement (Ph 3.1)** | | | |
| 53 | COGS 62,475 / 32,639, +91.4% | R | ✅ | [1] |
| 54 | GP 153,463 / 97,858, +56.8% | R | ✅ | [1] |
| 55 | R&D 18,497 / 12,914, +43%, 8.6% / 9.9%; compute and infrastructure +79% | R | ✅ | [1]; [6] p.41 |
| 56 | SG&A 4,579 / 3,491, +31%, 2.1% / 2.7% | R | ✅ | [1]; [6] |
| 57 | Op income 130,387 / 81,453, +60% | R | ✅ | [1] |
| 58 | Interest expense 259 / 247; coverage 503x | R/D | ✅ 🔁 | 130,387 ÷ 259 = 503.4 |
| 59 | Other income 9,022 / 1,034, +$8.0B, incl. Intel gains | R | ✅ | [1]; [6] p.42 |
| 60 | Tax rate 15.1% / 13.3%; $711M VA release; OBBBA | R | ✅ | [6] p.42 |
| 61 | NI 120,067 / 72,880, +65% | R | ✅ | [1] |
| 62 | Diluted EPS 4.90 / 2.94, +67%; diluted shares −1.2% | R/D | ✅ 🔁 | [1] 24,514 / 24,804 = −1.17% |
| 63 | Gains are 6.4% of pre-tax income | D | 🔁 agrees | 9,022 ÷ 141,450 = 6.38% |
| 64 | EPS growth ex-gains ≈60% (not 67%) | D | 🔁 agrees (approx.) | ex-gains pre-tax +59.7%; at each year's tax rate, EPS growth ≈58% |
| 65 | AR +66.8%; DSO 65.0 vs 64.5 | D | 🔁 agrees | [2] 38,466 / 23,065; ×365 on revenue |
| | **Balance sheet (Ph 3.2)** | | | |
| 66 | AR 38,466 | R | ✅ | [2] |
| 67 | Inventory 21,403, +112%, DIO 125 vs 113; WIP / FG / raw materials | R/D | ✅ 🔁 | [2]; [10] Note 9; DIO on COGS |
| 68 | Goodwill + intangibles 24,138 (11.7% of assets); goodwill 20,832 from 5,188 | R/D | ✅ 🔁 | [2] |
| 69 | Non-marketable equity 22,251 from 3,387; 10.8% of assets; ~6.5x | R/D | ✅ 🔁 | [2]; [9] Note 8 (6.57x) |
| 70 | DTA 13,258 | R | ✅ | [2] |
| 71 | Debt 999 / 7,469 / 8,468 | R | ✅ | [2] |
| 72★ | Net cash (54,088) | D | 🔁 agrees (caveat) | See ★3 |
| 73 | EBITDA ≈ $133.2B; Debt/EBITDA 0.06x | D | 🔁 agrees | 130,387 + 2,843; 8,468 ÷ 133,230 = 0.064 |
| 74 | RE 146,973 from 68,038; buybacks to RE 40,158; AOCI 178 | R | ✅ | [2]; [3] |
| 75 | Pack's net debt understates cash by ~$52B | D | 🔁 agrees | 54,088 − 2,137 = 51,951 |
| | **Cash flow (Ph 3.3)** | | | |
| 76 | FCF conversion 80.5% (prior 83.5%) | D | 🔁 agrees | 96,676 ÷ 120,067; 60,853 ÷ 72,880 |
| 77 | ~87% vs NI ex-gains | D | 🔁 agrees | 96,676 ÷ 111,149 = 87.0% |
| 78 | FCF after Groq 83,676 = 69.7% of NI | D | 🔁 agrees | [4] Groq (13,000) |
| 79 | Working-capital lines; net ≈ (15,949) | R/D | ✅ 🔁 | [4] sum of six lines |
| 80 | SBC 6,386; 3.0% vs 3.6% of revenue | R/D | ✅ 🔁 | [4] |
| 81 | D&A 2,843; 1.3% of revenue; capex/D&A 2.1x | R/D | ✅ 🔁 | [4] |
| 82 | Equity gains deducted (8,918) | R | ✅ | [4] |
| 83 | $17.5B into private companies; includes AI model makers that buy NVIDIA products | R/F | ✅ | [5] p.37; [4] 17,502 |
| | **Risk and legal (Ph 4)** | | | |
| 84★ | Concentration 22% + 14% = 36%; top one up from 12% | R | ✅ | [6]; [13] |
| 85 | One AI lab "meaningful amount" of revenue via cloud | F | ✅ | [6] p.41 |
| 86 | Groq "significant, nonrefundable payments" | F | ✅ | [16] Item 1A |
| 87 | Cloud timing "changed and may continue to shift"; "may not deliver anticipated benefits" | F | ✅ | [16] |
| 88 | Counterparty-risk factor | F | ✅ | [16] |
| 89 | Energy / capital constraints; open-source risk | F | ✅ | [5] p.36 |
| 90 | 4:18-cv-07669 pending; derivative suits stayed | F | ✅ | [12] pp.71–72 |
| 91 | UTB $4.0B; IRS examining FY2023–24 | R | ✅ | [7] p.44 |
| | **Footnotes (Ph 5)** | | | |
| 92 | Groq: $14.4B goodwill, $2.5B intangible, 5-year life, cost-to-recreate, workforce; ~$0.5B/yr amortization | R/D | ✅ 🔁 | [8] Note 2 (quote truncated, see §5) |
| 93 | Measurement alternative | F | ✅ | [5] p.38; [9] Note 8 |
| 94 | Segments C&N and Graphics | F | ✅ | [5] |
| 95 | Item 13 incorporated from the proxy | F | ✅ | [15] p.47 |
| | **Insider (Ph 8)** | | | |
| 96 | Kress plan adopted 2025-12-18, 500,000 shares, through 2027-03-23 | F | ✅ | [15] Item 9B p.46 |
| 97 | Dabiri plan ~3,984 shares | F | ✅ | [15] |
| | **Valuation** | | | |
| 98 | Price $229.28 (2026-10-09) | M | ✅ (user-supplied) | [18] |
| 99 | Shares 24,100M as of 2026-08-21 | M | ❓ | Not in the 10-K. The 10-K cover gives **24.3B as of 2026-02-20** [17] |
| 100 | Market cap ~$5.53T | D | 🔁 agrees on report inputs | $5.57T on the 10-K count (+0.8%) |
| 101 | EV ~$5.47T | D | 🔁 agrees | $5.52T on 24.3B |
| 102★ | P/E 46.8x | D | 🔁 agrees · 🕒 | |
| 103 | EV / revenue 25.3x | D | 🔁 agrees · 🕒 | 25.5x on 24.3B |
| 104 | EV / EBIT 42.0x | D | 🔁 agrees · 🕒 | 42.3x on 24.3B |
| 105 | FCF yield 1.75% | D | 🔁 agrees · 🕒 | 1.74% on 24.3B |
| | **Insights / composites** | | | |
| 106 | Share count −0.7% | D | 🔁 agrees | [3] 24,304 vs 24,477 |
| 107 | "$27.6B of combined AR and inventory build" | D | ⚠️ **MISMATCH** | [4] 15,399 + 11,324 = **$26.7B** |
| 108 | FY26 gross provisions = 3.3% of revenue | D | 🔁 agrees | 7.2 ÷ 215.9 |
| 109 | Accounting Quality 16/21 | D | 🔁 agrees | sum of rows |
| 110 | Score 6.8 sits in the BULLISH band | F | ✅ | Band rule ≥ 6.0 |
| | **Qualitative** | | | |
| 111 | "Ecosystem NVIDIA increasingly finances itself" / "circular financing" | Q | **qualified** | The evidence (investments, guarantees) holds, but Item 1A says: "We have not entered into any financing arrangements." [16] |
| 112 | "$27B of cloud capacity bought back from customer-CSPs" | Q | **[unsupported]** | Note 12 describes multi-year cloud agreements "to support our research and development". It says nothing about buying back capacity |
| 113 | Groq booked as a business combination | Q | supported | $14.4B goodwill recognized [8] |
| 114 | "No channel-stuffing signal" | Q | supported | DSO flat (row 65) |

---

## 3. Mismatches in detail

**M1, row 32: "Even excluding H20, provisions rose, which fits the CAM focus."**
- **Source:** gross provisions were "$7.2 billion and $3.7 billion … *including* $4.5 billion associated with H20" [6, p.41].
- **Recomputed:** excluding H20, gross provisions were **$2.7B vs $3.7B, down 27%**. Net of the $1.5B / $0.689B releases they were about $1.2B vs $3.0B.
- **Likely cause:** the H20 component was never subtracted before the comparison.
- **Effect:** removes one supporting point for Red Flag #3. The flag still stands on inventory +112% and the $95.2B commitment base. The signal does not change.
- **Related:** the 3.3% "FY26 gross" baseline in Thesis Invalidation is arithmetically correct (row 108), but it includes the one-off H20 charge. Excluding it, the figure is about 1.25%.

**M2, row 107: "$27.6B of combined AR and inventory build."**
- **Source:** 15,399 + 11,324 = **$26,723M** [4]. The balance-sheet deltas give the same answer (15,401 + 11,323).
- **Size:** overstated by $0.9B (+3.3%).
- **Likely cause:** transposed digits.
- **Effect:** none. The working-capital criterion score of 1 is unaffected.

**Unverifiable, row 99: 24,100M shares "as of 2026-08-21."**
- The pack labels this as the "cover-page count". But this accession's cover page reads "24.3 billion" as of **2026-02-20** [17], and a filing made on 2026-02-25 cannot carry an August date.
- The figure almost certainly comes from a later filing, likely the Q2 FY2027 10-Q cover, which is not in the files. That 10-Q's cover page would settle it.
- On the 10-K's own 24.3B count, every valuation multiple moves by under 1% (rows 100–105). P/E is per-share and does not change. The valuation conclusions survive either way.

---

## 4. Recomputation table

```
Revenue growth     215,938 ÷ 130,497 − 1                 = 65.47%   ✓ 65.5%
FCF                102,718 − 6,042                        = 96,676   ✓
FCF margin         96,676 ÷ 215,938                       = 44.77%   ✓ 44.8%
FCF conversion     96,676 ÷ 120,067                       = 80.52%   ✓ 80.5%   (prior 60,853 ÷ 72,880 = 83.50% ✓)
GM ex-H20          (153,463 + 4,500) ÷ 215,938            = 73.15%   ✓ 73.2%
Op margin ex-H20   (130,387 + 4,500) ÷ 215,938            = 62.47%   ✓ 62.5%
Net cash           62,556 − 8,468                         = 54,088   ✓  (ex-public equities: 62,556 − 12,886 − 8,468 = 41,202)
EBITDA; D/EBITDA   130,387 + 2,843 = 133,230; 8,468 ÷ …   = 0.064x   ✓ 0.06x
Interest coverage  130,387 ÷ 259                          = 503.4x   ✓
DSO / DIO          38,466 ÷ 215,938 × 365 = 65.0;  21,403 ÷ 62,475 × 365 = 125.0 (prior 64.5 / 112.7)  ✓
Capital return     (40,086 + 974) ÷ 96,676                = 42.47%   ✓ 42.5%
Provisions ex-H20  7.2 − 4.5 = 2.7 vs 3.7                 → FELL     ✗ report "rose"
AR + Inv build     15,399 + 11,324                        = 26,723   ✗ report 27,600
Market cap         24,100 × 229.28 = $5.526T ✓ · (24,300 × 229.28 = $5.572T)
EV                 5,525.6 − 54.1 = $5.472T ✓ · (5,517.4)
P/E                229.28 ÷ 4.90                          = 46.79x   ✓
EV/Rev · EV/EBIT   5,471.6 ÷ 215.9 = 25.3x ✓ · ÷ 130.4 = 42.0x ✓
FCF yield          96,676 ÷ 5,525,648                     = 1.75%    ✓
Groq amortization  2.5B ÷ 5 yrs                           = 0.5B/yr  ✓  (consideration 13.0 + 4.0 ≈ GW 14.4 + IA 2.5 ✓)
Accounting score   3+2+2+1+3+3+2                          = 16/21    ✓
```

---

## 5. Report consistency findings

- **Numbers vs. narrative:** one contradiction, M1. Everything else in the narrative matches its tables.
- **Internal consistency:**
  - Buybacks are quoted on the cash-flow basis ($40.1B) throughout and on the retained-earnings basis ($40,158M) where labelled. That is consistent.
  - The MD&A's $40.4B is a third basis the report never cites, which is fine.
- **Net-cash definition (★3):**
  - The $54.1B figure counts $12.9B of publicly held equities, $10.5B of which are locked up.
  - §2.3 mentions the lock-up, but the dashboard and the signal rationale don't. Excluding equities, net cash is $41.2B.
  - This is a disclosure gap, not a wrong number.
- **Quotes edited inside quotation marks:**
  - The Gaming quote reads "…Q1 FY27 and beyond". The filing says "to Gaming in the first quarter of fiscal 2027 and beyond".
  - The Groq goodwill quote stops at "primarily attributable to the workforce" and drops "and future development of the licensed technology". That omission slightly strengthens the report's "unusual" framing.
- **Characterizations:**
  - Row 111 should acknowledge the filing's explicit "We have not entered into any financing arrangements".
  - Row 112's "bought back" wording has no support in the filing.
- **Citation locations:**
  - The lease guarantees are listed under "Off-balance-sheet commitments (Note 12)". They are in Note 10, and they are credit derivatives carried at fair value. The report's own Red Flag #2 cites Note 10 correctly.
  - "$4.5B write-down … (Note 9)": the H20 figure is in the MD&A, while Note 9 holds the $4.0B + $3.2B provisions that include it.
  - Both are imprecise but not fabricated, because the figures exist in the filing. No citation is unopenable, so the zero cap does not fire.
- **Signal vs. score:** 6.8 sits correctly in the BULLISH band. BULLISH with HOLD is explained in the report's note.
- **The report's own `Data & Sources` header:**
  - It reads `Retrieval: pasted by user (local files; no browsing) · Confidence: HIGH for reported FY2026 figures · MEDIUM for valuation`.
  - The verification supports it: 85 of 85 reported figures checked against the 10-K matched.
  - The one gap: "shares as of 2026-08-21" is attributed to sources that cannot contain an August date. The header is not overstated, so no cap fires.
- **Side note on the data pack (not the report):** `fetch-fundamentals.js` appears to label the latest `dei:EntityCommonStockSharesOutstanding` value as this 10-K's cover-page count. It may be pulling the newest filing's value instead of the cited accession's. Worth checking before the next demo.

---

## 6. Corrected report: changed passages and citation map

Only the passages that change are reprinted below, with the original struck through. Every other figure stands as written and carries the citations in the map that follows.

**§2.2, after the margins table:**
> Inventory provisions were $7.2B gross against $3.7B, with net GM drag of 2.6 pts against 2.3 pts [6]. ~~Even excluding H20, provisions rose, which fits the CAM focus.~~ **Excluding the $4.5B H20 charge, gross provisions fell to about $2.7B from $3.7B [6] (recomputed); the year-on-year increase is the H20 event.** *(Correction: the H20 amount is included in the $7.2B and was not subtracted.)*

**§2.3, last bullet:**
> $3.5B lease guarantees to early-stage partners, with $712M in escrow ~~(Note 12)~~ **(Note 10; credit derivatives at fair value, not material)** [11]

**Executive Summary and Dashboard, net cash:**
> $54.1B net cash once marketable securities are included [2][7] (recomputed) **(includes $12.9B of public equities, $10.5B locked up [9]; $41.2B excluding equities)**

**Valuation Snapshot:**
> | Market cap | ~$5.53T | 24,100M shares [?] × $229.28 [18] **(10-K cover: 24.3B shares as of 2026-02-20 [17] → $5.57T)** |

**Negatives, circular financing:**
> **Circular-financing exposure:** $17.5B into private AI firms that buy NVIDIA products [5][4], $3.5B of lease guarantees to early-stage partners [11], a pending OpenAI deal [7], and ~~$27B of cloud capacity bought back from customer-CSPs~~ **$27B of multi-year cloud-service commitments [12] [unsupported as "bought back"]**. **The 10-K states "We have not entered into any financing arrangements" [16]; "vendor financing" is this report's characterization.**

**Accounting Quality Score, Working capital row:**
> Inventory doubled; ~~$27.6B~~ **$26.7B** [4] (recomputed) of combined AR and inventory build

**Citation map for the unchanged figures:**
- Executive Summary: [1][4][5][6][8][12][14]
- Dashboard: [1][2][4] (recomputed)
- Phase 1: [8][14][17] and filing metadata
- Phase 2.1: [6][8][10][13]
- Phase 2.2: [5][6] (rows 24, 27, 29 recomputed)
- Phase 2.3: [3][4][7][8][9][10][11][12]
- Phase 2.4: [5][7]
- Phase 3.1: [1][6]
- Phase 3.2: [2][3][9][10]
- Phase 3.3: [4][8]
- Phase 4: [6][12][16]
- Phase 5: [5][8][9][15]
- Phase 8: [15]
- Valuation: [1][2][18] (recomputed; 🕒 FY26 EPS)
- Key Insights and Red Flags: as above
- Thesis Invalidation: the "3.3%" baseline is [6] (recomputed) and includes H20

## References

- [1] NVIDIA Corp., Form 10-K for the FY ended 2026-01-25, filed 2026-02-25, accession 0001045810-26-000021. Consolidated Statements of Income, p.51. `data/filings/NVDA/NVDA_2026_10-K.txt`, pasted by user.
- [2] Same filing. Consolidated Balance Sheets, p.53.
- [3] Same filing. Consolidated Statements of Shareholders' Equity, p.54.
- [4] Same filing. Consolidated Statements of Cash Flows, p.55.
- [5] Same filing. Item 7 MD&A: Recent Developments, FY2026 Summary, Critical Accounting Estimates, pp.36–38.
- [6] Same filing. Item 7 MD&A: Results of Operations (% table, Reportable Segments, Concentration of Revenue, Gross Margin, Opex, Other Income, Income Taxes), pp.40–42.
- [7] Same filing. Item 7 MD&A: Liquidity, Capital Return, Indebtedness, Material Cash Requirements, pp.42–44.
- [8] Same filing. Note 1 Fiscal Year (p.56); Note 2 Groq (p.61); Note 5 Goodwill (p.63).
- [9] Same filing. Note 7 Cash Equivalents and Marketable Securities (pp.65–66); Note 8 Non-marketable Equity Securities (pp.66–67).
- [10] Same filing. Note 9 Balance Sheet Components, including deferred revenue and RPO, pp.67–69.
- [11] Same filing. Note 10 Facility Lease Guarantees (p.69); Note 11 Debt (p.70).
- [12] Same filing. Note 12 Commitments (p.70) and Litigation (pp.71–72).
- [13] Same filing. Note 16 Segment Information: geographic revenue (p.78), revenue by end market (p.79).
- [14] Same filing. Report of Independent Registered Public Accounting Firm, PwC, dated 2026-02-25, pp.49–50; Item 9A, pp.45–46.
- [15] Same filing. Item 9B Other Information (p.46); Part III Items 10–14 (pp.46–47).
- [16] Same filing. Item 1A Risk Factors: Groq license, cloud service agreements, and "Commercial arrangements expose us to counterparty risks". Page numbers were not captured in the text extract.
- [17] Same filing. Cover page: error-correction checkboxes; "24.3 billion" shares outstanding as of 2026-02-20.
- [18] User-supplied: `data/fixtures/NVDA.md`, generated by `scripts/fetch-fundamentals.js` on 2026-10-11. Price $229.28 (NASDAQ close 2026-10-09) and shares 24,100M. Pasted by user; not independently verified.

---

## 7. Thesis Invalidation for these verdicts

- **A new filing restates a relied-on figure.** An 8-K or 10-K/A turns the affected ✅ verdicts into 🕒 until they are re-checked.
- **Valuation:**
  - **Shares:** loading the Q2 FY2027 10-Q settles row 99, either confirming 24,100M or replacing it.
  - **Trailing EPS:** the same 10-Q gives trailing-twelve-month EPS, which retires the 🕒 on rows 102–105.
- **The price** is checked against an independent quote. Today it rests only on the pack.
- **Re-run when:**
  - [ ] The FY2027 10-Qs or the FY2027 10-K land in `data/filings/NVDA/`
  - [ ] The report is regenerated with the corrections above
  - [ ] The 2026-10-09 price passes 90 days (2027-01-07)
  - [ ] Before this report is published as a site demo

## 8. Investment Signal

```
╔══════════════════════════════════════════════╗
║              INVESTMENT SIGNAL               ║
╠══════════════════════════════════════════════╣
║ Signal:      BULLISH                         ║
║ Confidence:  MEDIUM                          ║
║ Horizon:     LONG-TERM                       ║
║ Score:       8.5 / 10                        ║
╠══════════════════════════════════════════════╣
║ Action:      HOLD                            ║
║ Conviction:  MODERATE                        ║
╚══════════════════════════════════════════════╝
```

Score Guide: 8.0–10.0 Strongly Bullish | 6.0–7.9 Moderately Bullish | 4.0–5.9 Neutral | 2.0–3.9 Moderately Bearish | 0.0–1.9 Strongly Bearish
Confidence: HIGH (strong data, clear signals) | MEDIUM (mixed signals) | LOW (limited data, conflicting signals)
Horizon: SHORT-TERM (1 week–3 months) | MEDIUM-TERM (3 months–1 year) | LONG-TERM (1+ years)

Mirrors the report's signal, quoted rather than re-derived: BULLISH / HOLD / MODERATE / LONG-TERM, at the report's own 6.8.

The Score above is the Verification Score. It breaks down as follows:

| Component | Weight | Score | Basis |
|---|---|---|---|
| Coverage | 30% | 9 | 107/114 claims (93.9%) verified or recomputed-agree |
| Signal-driving accuracy | 30% | 9 | All five within tolerance, with the net-cash definition and P/E vintage caveats |
| Recomputation | 15% | 8 | 22 of 24 derived figures agree |
| Source quality | 15% | 8.5 | Tier 1 with page-level locations; price and shares come from the user-supplied pack only |
| Freshness | 10% | 6 | Fundamentals are 8.5 months old (disclosed); the share-count date is untraceable |

Weighted total: 8.5. Caps fired: none.

Verification would support HIGH confidence, but the box keeps the report's own MEDIUM. That rating reflects the missing FY2027 data, which verification doesn't change.

**Disclaimer:** Educational analysis only. Not financial advice. Verification is limited to the sources available at run time; an unverifiable claim is neither confirmed nor refuted.