NVDA / H14 / 2026-10-13
H14研究：税率与三期限现金流滚动H14 study: guided tax and three-horizon cash-flow roll-forwardH14研究：税率與三期限現金流滾動
研究日期2026-10-13；输入与实际市场收盘日2026-10-12。最新财报FY2027Q2，期末2026-07-26，发布2026-08-26、模型可得2026-08-27。利率采用2026-10-09严格先前观察；本轮历史评估截止2026-10-12。Research date October 13, 2026; inputs and actual market close October 12, 2026. Latest financials: FY2027Q2, ended 2026-07-26, released 2026-08-26, model-available 2026-08-27. Treasury uses the strictly prior 2026-10-09 observation; this historical evaluation ends 2026-10-12.研究日期2026-10-13；輸入與實際市場收盤日2026-10-12。最新財報FY2027Q2，期末2026-07-26，發佈2026-08-26、模型可得2026-08-27。利率採用2026-10-09嚴格先前觀察；本輪歷史評估截止2026-10-12。
H14未通过28项预先冻结的采用门槛。现行研究展示保留H09 + FAIR0整套版本：0M独立现金价值，6M/12M原H09市场PE价格预测。没有把H14税率或某个更好看的期限拼入主模型。H14 failed 28 predeclared adoption gates. The current research display retains the whole H09 + FAIR0 version: independent cash value at 0M and original H09 market-PE price forecasts at 6M/12M. H14 tax assumptions or selected horizons are not spliced into the headline.H14未通過28項預先凍結的採用門檻。現行研究展示保留H09 + FAIR0整套版本：0M獨立現金價值，6M/12M原H09市場PE價格預測。沒有把H14税率或某個更好看的期限拼入主模型。
最终保留主表：原H09 + FAIR0Final retained headline: original H09 + FAIR0最終保留主表：原H09 + FAIR0
| 期限Horizon期限 | 目标日 → 盈利窗口末日Target → earnings-window end目標日 → 盈利窗口末日 | Core EPS | Core PE | USD/股USD/shareUSD/股 | 身份Role身份 |
|---|---|---|---|---|---|
| 0M | 2026-10-12 → 2027-10-12 | 13.95945 | 18.76282× | $261.92 | 保留H09-FAIR0当前价值Retained H09-FAIR0 current value保留H09-FAIR0當前價值 |
| 6M | 2027-04-12 → 2028-04-12 | 16.71457 | 16.21268× | $270.99 | 保留H09市场PE价格预测Retained H09 market-PE price forecast保留H09市場PE價格預測 |
| 12M | 2027-10-12 → 2028-10-12 | 18.97455 | 16.21268× | $307.63 | 保留H09市场PE价格预测Retained H09 market-PE price forecast保留H09市場PE價格預測 |
10月12日供应商收盘记录修订October 12 vendor-close revision10月12日供應商收盤記錄修訂
10月13日再次核验FMP与Stock Analysis历史日线，10月12日供应商EOD记录从$226.43修订为$226.32。本轮基准与候选使用同一新核验输入，早期捕获的0M/6M/12M $261.92/$271.12/$307.78保持原样。0M独立估值不受P0修订影响，原H09未来目标按相同EPS与新市场PE重算。FMP and Stock Analysis historical daily records were rechecked on October 13. The October 12 vendor EOD record changed from $226.43 to $226.32. Baseline and candidate use the same newly verified inputs; the early captured 0M/6M/12M values of $261.92/$271.12/$307.78 remain unchanged. Independent 0M value is unaffected by the P0 revision; original H09 future targets use the same EPS and revised market PE.10月13日再次核驗FMP與Stock Analysis歷史日線，10月12日供應商EOD記錄從$226.43修訂為$226.32。本輪基準與候選使用同一新核驗輸入，早期捕獲的0M/6M/12M $261.92/$271.12/$307.78保持原樣。0M獨立估值不受P0修訂影響，原H09未來目標按相同EPS與新市場PE重算。
早期10月12日捕获版本原样归档Preserved early October 12 capture早期10月12日捕獲版本原樣歸檔
单一冻结候选如何改变三个期限How the single frozen candidate changes all horizons單一凍結候選如何改變三個期限
保持H09收入、GAAP毛利、费用、同比日收入动量、四季度半衰期与季节性路径；每个历史起点改用当时已公布的GAAP正常化税率指导中点。当前16–18%指导取17%，同时更新NOPAT、Core EPS、现金利息、债务税盾、WACC和终值。没有用当前税率回填历史，没有调参搜索。Retain H09 revenue, GAAP margins, costs, same-quarter daily-revenue momentum, four-quarter half-life and seasonality. Each origin uses the midpoint of its then-published normalized GAAP tax guidance. Current 16%–18% guidance gives 17%, applied consistently to NOPAT, Core EPS, cash interest, debt tax shields, WACC and terminal value. Current tax guidance is not backfilled into history; there is no parameter search.保持H09收入、GAAP毛利、費用、同比日收入動量、四季度半衰期與季節性路徑；每個歷史起點改用當時已公佈的GAAP正常化税率指導中點。當前16–18%指導取17%，同時更新NOPAT、Core EPS、現金利息、債務税盾、WACC和終值。沒有用當前税率回填歷史，沒有調參搜索。
指导的原始期限与推演假设分别注明：当前公司指导针对FY2027，模型将起点已知税率固定延续到stub、未来60个月及终值。这是模型外推，不是公司已经指导所有未来财年；没有为FY2028之后编造新指引。Core EPS仍为税后营业盈利，不等于公司GAAP净利润EPS。The original guidance period and projection assumption are separate: current company guidance is for FY2027, while the model holds the origin-known rate throughout the stub, sixty-month path and terminal. This is extrapolation, not company guidance for all future fiscal years. No new FY2028 or later guidance is fabricated. Core EPS remains after-tax operating earnings, separate from GAAP net-income EPS.指導的原始期限與推演假設分別註明：當前公司指導針對FY2027，模型將起點已知税率固定延續到stub、未來60個月及終值。這是模型外推，不是公司已經指導所有未來財年；沒有為FY2028之後編造新指引。Core EPS仍為税後營業盈利，不等於公司GAAP淨利潤EPS。
| 项目Item項目 | 冻结计算Frozen calculation凍結計算 |
|---|---|
| Core EPS_h | NOPAT in [target_h, target_h + 12M) / H09 reported diluted WA shares |
| FV_h | (remaining economic cash-flow PV_h + common terminal PV_h + modeled economic cash_h + investments − senior debt) / same H09 shares |
| Fair Core PE_h | FV_h / Core EPS_h |
| 共同终点与再投资Common terminal and reinvestment共同終點與再投資 | origin + 60M; g = 3%; ROIC = 25%; reinvestment = g / ROIC |
| 必要回报Required returns必要回報 | ke = prior 10Y + 6 points; kd = prior 10Y + 1 point; book capital weights |
| 当前税率／WACCCurrent tax / WACC當前税率／WACC | 17.00% / 10.4692% |
| 同口径股份数Same-basis share count同口徑股份數 | 24.285 billion |
0M、6M、12M都从同一经济现金路径向前滚动。目标日前现金保留一次，之后现金流折现一次，共同终点不在每个期限重新启动。每一期PE由独立价值除以同口径盈利得到，P0和分析师目标不决定倍数。单独修正固定税率会在原H09的EPS_h/EPS_0中相消，不能冒充未来目标改善。0M, 6M and 12M roll forward from one economic cash path. Cash generated before each target is retained once; subsequent cash flows are discounted once. The common terminal is not restarted at each horizon. Each PE is independent value divided by same-basis earnings; P0 and analyst targets do not set multiples. A uniform tax correction cancels in original H09 EPS_h/EPS_0 and cannot be relabeled as improved future targets.0M、6M、12M都從同一經濟現金路徑向前滾動。目標日前現金保留一次，之後現金流折現一次，共同終點不在每個期限重新啓動。每一期PE由獨立價值除以同口徑盈利得到，P0和分析師目標不決定倍數。單獨修正固定税率會在原H09的EPS_h/EPS_0中相消，不能冒充未來目標改善。
H14最终计算：未采用的条件公允价值Final H14 calculations: unadopted conditional fair valuesH14最終計算：未採用的條件公允價值
| 期限Horizon期限 | 目标日 → 盈利窗口末日Target → earnings-window end目標日 → 盈利窗口末日 | Core EPS | Core PE | USD/股USD/shareUSD/股 | 身份Role身份 |
|---|---|---|---|---|---|
| 0M | 2026-10-12 → 2027-10-12 | 13.63099 | 18.79810× | $256.24 | 保留H09-FAIR0当前价值Retained H09-FAIR0 current value保留H09-FAIR0當前價值 |
| 6M | 2027-04-12 → 2028-04-12 | 16.32128 | 16.46797× | $268.78 | 保留H09市场PE价格预测Retained H09 market-PE price forecast保留H09市場PE價格預測 |
| 12M | 2027-10-12 → 2028-10-12 | 18.52809 | 15.21299× | $281.87 | 保留H09市场PE价格预测Retained H09 market-PE price forecast保留H09市場PE價格預測 |
表中末日不含，盈利窗口从各目标日开始持续12个月。三个期限均为起点假设成立时的条件公允价值，不能假设市场必定在目标日兑现；也没有把0M独立PE固定传播到未来两期。Window end dates are excluded; each earnings window begins on its target date and lasts twelve months. All horizons are conditional fair values under origin assumptions, without assuming markets realize value on the target date. The independent 0M PE is not held fixed at future horizons.表中末日不含，盈利窗口從各目標日開始持續12個月。三個期限均為起點假設成立時的條件公允價值，不能假設市場必定在目標日兑現；也沒有把0M獨立PE固定傳播到未來兩期。
三年完整回测：条件值与实际价格Complete three-year comparison: conditional values and realized prices三年完整回測：條件值與實際價格
本轮评估截止2026-10-12，实际价格标签沿用截至2026-10-06的冻结档案；最新成熟目标为2026-08-28，早于档案末日。没有新增成熟样本，不把原历史价格档案冒充本轮全量行情刷新。Evaluation ends 2026-10-12, using frozen price labels through 2026-10-06. The latest mature target is August 28, 2026, before the archive ends. No additional outcome has matured; the original archive is not relabeled as a full market-data refresh.本輪評估截止2026-10-12，實際價格標籤沿用截至2026-10-06的凍結檔案；最新成熟目標為2026-08-28，早於檔案末日。沒有新增成熟樣本，不把原歷史價格檔案冒充本輪全量行情刷新。
| 期限Horizon期限 | 模型Model模型 | 成熟／待定NMature / pending N成熟／待定N | MAPE | MALE | 中位APEMedian APE中位APE | P90 APE | 有符号偏差Signed bias有符號偏差 |
|---|---|---|---|---|---|---|---|
| 6M | H14 | 10 / 2 | 29.60% | 0.38807 | 33.28% | 51.32% | -27.93% |
| 6M | H09 | 10 / 2 | 14.33% | 0.14744 | 12.21% | 30.89% | -1.72% |
| 6M | PriceUnchanged | 10 / 2 | 21.14% | 0.25371 | 19.73% | 37.75% | -18.88% |
| 6M | 旧15%现金法控制，未选用Old 15% cash-method control; unselected舊15%現金法控制，未選用 | 10 / 2 | 29.35% | 0.37980 | 31.94% | 50.24% | -26.51% |
| 12M | H14 | 8 / 4 | 40.74% | 0.55726 | 47.19% | 53.35% | -39.55% |
| 12M | H09 | 8 / 4 | 14.91% | 0.16237 | 10.46% | 26.31% | -3.65% |
| 12M | PriceUnchanged | 8 / 4 | 32.20% | 0.42024 | 32.34% | 47.51% | -32.20% |
| 12M | 旧15%现金法控制，未选用Old 15% cash-method control; unselected舊15%現金法控制，未選用 | 8 / 4 | 40.26% | 0.54537 | 46.00% | 52.55% | -38.51% |
保留原三年12个季度起点；6M成熟10、待2，12M成熟8、待4。实际价格取目标日及之后首个交易日常规收盘，不取期间最高价。MAPE分母为实际价格；不是准确率。H14的误差明显高于原H09，旧15%现金法控制同样较差，不能把差异全部归因于税率。Preserve all twelve original quarterly origins within three years: 6M has ten mature/two pending, and 12M eight mature/four pending. Realized labels use the regular close on the first trading day on or after the target, not the window high. MAPE uses realized prices in the denominator and is not an accuracy rate. H14 errors materially exceed original H09; the old 15% cash-method control also performs worse, so the gap cannot be attributed entirely to tax.保留原三年12個季度起點；6M成熟10、待2，12M成熟8、待4。實際價格取目標日及之後首個交易日常規收盤，不取期間最高價。MAPE分母為實際價格；不是準確率。H14的誤差明顯高於原H09，舊15%現金法控制同樣較差，不能把差異全部歸因於税率。
现金滚动回答价值随经营与留存现金如何变化，固定市场PE更直接承接起点的市场定价。更一致的财务体系没有在本轮获得更好的市场价格误差。旧15%现金法仅用于归因，是冻结的控制，不因某个分数较好而替代H14成为第二候选。Cash roll-forward asks how conditional value evolves with operations and retained cash; held market PE carries forward origin market pricing more directly. A more consistent financial framework did not produce lower market-price errors here. The old 15% cash method is a frozen attribution control, not a second candidate selected for a better score.現金滾動回答價值隨經營與留存現金如何變化，固定市場PE更直接承接起點的市場定價。更一致的財務體系沒有在本輪獲得更好的市場價格誤差。舊15%現金法僅用於歸因，是凍結的控制，不因某個分數較好而替代H14成為第二候選。
分析师12M目标：同日差与成熟价格分开Analyst 12M targets: separate same-date gaps from mature-price errors分析師12M目標：同日差與成熟價格分開
| 比较Comparison比較 | 模型Model模型 | N | MAPE | MALE | P90 APEP90 APEP90 APE |
|---|---|---|---|---|---|
| 同日12M平均目标差Same-date average 12M target disagreement同日12M平均目標差 | H14 | 12 | 37.13% | 0.43109 | 54.72% |
| 同日12M平均目标差Same-date average 12M target disagreement同日12M平均目標差 | H09 | 12 | 16.86% | 0.14611 | 35.93% |
| 共同成熟实际12M价格误差Common mature realized 12M price error共同成熟實際12M價格誤差 | H14 | 8 | 40.74% | 0.55726 | 53.35% |
| 共同成熟实际12M价格误差Common mature realized 12M price error共同成熟實際12M價格誤差 | H09 | 8 | 14.91% | 0.16237 | 26.31% |
| 共同成熟实际12M价格误差Common mature realized 12M price error共同成熟實際12M價格誤差 | 分析师平均目标Average analyst target分析師平均目標 | 8 | 18.39% | 0.22916 | 40.70% |
同日Stock Analysis平均12M目标N=12，共同成熟实际12M价格N=8、待4。目标接近度与实际价格误差分别报告，分析师数据不参与拟合。没有可匹配的6M分析师历史目标，不能把12M平均目标缩短期限、折算或冒充6M基准。6M仅与相应6M实际价格比较。Same-date Stock Analysis average 12M targets have N=12; common mature realized 12M prices have N=8, with 4 pending. Target agreement and realized-price errors are reported separately; analysts never enter fitting. Matching historical 6M analyst targets are unavailable. Twelve-month averages are not shortened, converted or relabeled as 6M benchmarks; 6M is compared with actual 6M prices only.同日Stock Analysis平均12M目標N=12，共同成熟實際12M價格N=8、待4。目標接近度與實際價格誤差分別報告，分析師數據不參與擬合。沒有可匹配的6M分析師歷史目標，不能把12M平均目標縮短期限、折算或冒充6M基準。6M僅與相應6M實際價格比較。
28项门槛失败，完整版本不替换Twenty-eight failed gates; no whole-version replacement28項門檻失敗，完整版本不替換
| 门槛组Gate family門檻組 | 失败数Failures失敗數 |
|---|---|
| 6M误差与固定阶段门槛6M error and fixed-stage gates6M誤差與固定階段門檻 | 12 |
| 12M误差与固定阶段门槛12M error and fixed-stage gates12M誤差與固定階段門檻 | 10 |
| 同日分析师目标差门槛Same-date analyst-target-gap gates同日分析師目標差門檻 | 3 |
| 成熟12M分析师配对门槛Mature 12M analyst paired gates成熟12M分析師配對門檻 | 3 |
预设要求两期限MALE至少相对改善5%、MAPE下降，且中位、P90、最大误差和有符号偏差不恶化；各固定阶段成熟N≥2时，MAPE和MALE不恶化。另要求同日分析师差不恶化，以及共同成熟12M的MAPE/中位/P90不差于分析师。H14多项失败，不能只凭0M价格差稍小就采用。The frozen screen requires at least 5% relative MALE improvement and lower MAPE at both horizons, without worse median, P90, maximum errors or absolute signed bias. Fixed stages with at least two mature origins must not worsen MAPE or MALE. Same-date analyst disagreement must not worsen, and common mature 12M MAPE/median/P90 must be no worse than analysts. H14 fails multiple requirements; a slightly smaller 0M price gap does not justify adoption.預設要求兩期限MALE至少相對改善5%、MAPE下降，且中位、P90、最大誤差和有符號偏差不惡化；各固定階段成熟N≥2時，MAPE和MALE不惡化。另要求同日分析師差不惡化，以及共同成熟12M的MAPE/中位/P90不差於分析師。H14多項失敗，不能只憑0M價格差稍小就採用。
全覆盖与算术审计通过表示模型可复算，不改变采用结果。失败后未更换税率、ERP、半衰期、终值或择优混合期限；本轮保留整个原H09 + FAIR0基准。后续版本应先提出新的财务证据与冻结规则，再进行独立新研究。Complete coverage and a passed arithmetic audit establish reproducibility, not adoption. Tax, ERP, half-life and terminal assumptions are not retuned after failure, and horizons are not mixed by score. The whole original H09 + FAIR0 reference remains. Any next version needs new financial evidence and frozen rules before a separate study.全覆蓋與算術審計通過表示模型可復算，不改變採用結果。失敗後未更換税率、ERP、半衰期、終值或擇優混合期限；本輪保留整個原H09 + FAIR0基準。後續版本應先提出新的財務證據與凍結規則，再進行獨立新研究。
0M没有股价提供的公允价值真值0M has no price-based fair-value ground truth0M沒有股價提供的公允價值真值
12个起点全部可估计。H14的0M同日市场价偏离MAPE 35.70%，原H09-FAIR0 35.96%；这些只描述定价差，不能证明哪套Fair Value更准确。税率修正改善来源一致性，不等于市场价格接近度改善或未来预测成功。All twelve origins are estimable. H14 0M contemporaneous market-price-gap MAPE is 35.70%, versus 35.96% for original H09-FAIR0. These describe pricing differences and cannot establish which Fair Value is more accurate. The tax correction improves source consistency, without establishing closer market pricing or successful future forecasts.12個起點全部可估計。H14的0M同日市場價偏離MAPE 35.70%，原H09-FAIR0 35.96%；這些只描述定價差，不能證明哪套Fair Value更準確。税率修正改善來源一致性，不等於市場價格接近度改善或未來預測成功。
当前H14终值现值占经营价值71.56%；现金余额是经济价值账户的模型滚动，并非已核验的当日银行余额。资产融资期末索偿、其他义务、投资专属税负和既有期权/RSU未完全映射。本金预留不等于完整债务扣清，结果不是严格FCFF或已证明保守下限。Current H14 terminal PV contributes 71.56% of operating value. Cash is a modeled economic-value account, not a verified same-day bank balance. Ending asset-financing claims, other obligations, investment-specific taxes and existing option/RSU claims are incompletely mapped. A principal reserve does not settle the full debt bridge; this is not strict FCFF or a proven conservative floor.當前H14終值現值佔經營價值71.56%；現金餘額是經濟價值賬户的模型滾動，並非已核驗的當日銀行餘額。資產融資期末索償、其他義務、投資專屬税負和既有期權/RSU未完全映射。本金預留不等於完整債務扣清，結果不是嚴格FCFF或已證明保守下限。
实现验证与研究边界Implementation verification and research limits實現驗證與研究邊界
独立复算PASS，共203,270项检查、0失败；候选采用状态仍为未通过。Independent reproduction passed 203,270 checks with zero failures; candidate adoption remains rejected.獨立復算PASS，共203,270項檢查、0失敗；候選採用狀態仍為未通過。
独立实现复算通过：H09经营美元保持一致，股份口径一致，税率作用于全现金与折现路径，P0扰动不改变H14任一期条件价值。来源日期、未来财务与分析师隔离、现金守恒和旧版本保护同时核验。Independent implementation reproduction passed: H09 operating dollars and share basis remain consistent, tax applies throughout the cash and discount path, and P0 perturbations leave every H14 conditional value unchanged. Source dates, future-financial and analyst isolation, cash conservation and prior-version preservation are also checked.獨立實現復算通過：H09經營美元保持一致，股份口徑一致，税率作用於全現金與折現路徑，P0擾動不改變H14任一期條件價值。來源日期、未來財務與分析師隔離、現金守恆和舊版本保護同時核驗。
实现检查不是准确率认证；没有正式Champion、独立OOS或完整不可变PIT。原季度样本反复使用且未来盈利窗重叠，只能标回顾性研究。当前6M/12M尚未兑现；历史价差不能将条件公允价值自动变成可靠市场价格预测。Implementation checks are not accuracy certification. Formal Champion status, independent OOS and complete immutable PIT are absent. Quarterly origins have been reused and future earnings windows overlap, so this is retrospective research. Current 6M/12M values remain pending; historical gaps do not turn conditional fair value into a reliable market-price forecast.實現檢查不是準確率認證；沒有正式Champion、獨立OOS或完整不可變PIT。原季度樣本反覆使用且未來盈利窗重疊，只能標回顧性研究。當前6M/12M尚未兑現；歷史價差不能將條件公允價值自動變成可靠市場價格預測。
网站仅发布自有模型输出、方法、聚合比较与审计摘要。逐日Stock Analysis Pro目标、逐起点财务、现金日账和来源原文留在内部研究记录。The website publishes original model outputs, methods, aggregate comparisons and audit summaries only. Daily Stock Analysis Pro targets, origin-level financials, daily cash ledgers and source documents remain internal.網站僅發佈自有模型輸出、方法、聚合比較與審計摘要。逐日Stock Analysis Pro目標、逐起點財務、現金日賬和來源原文留在內部研究記錄。
固定单因素敏感性：不选为目标价Fixed one-factor sensitivity: not selected as targets固定單因素敏感性：不選為目標價
| 变动Change變動 | 0M | 6M | 12M |
|---|---|---|---|
| WACC_MINUS1PP | $295.17 | $308.33 | $322.04 |
| WACC_PLUS1PP | $226.54 | $238.60 | $251.21 |
| TERMINAL_G2PCT | $240.72 | $252.47 | $264.73 |
| TERMINAL_G4PCT | $276.37 | $289.94 | $304.11 |
| SHARES_0.98 | $261.47 | $274.26 | $287.62 |
| SHARES_1.02 | $251.21 | $263.51 | $276.34 |
这些变动没有参与评分、不是置信区间，也不作为更高或更贴市价的替代头条。低WACC或较高终值增长可以提高估值，但不构成本轮优化证据。These variations are unscored, are not confidence intervals and are not higher or price-matching alternative headlines. Lower WACC or higher terminal growth can raise value but do not provide optimization evidence here.這些變動沒有參與評分、不是置信區間，也不作為更高或更貼市價的替代頭條。低WACC或較高終值增長可以提高估值，但不構成本輪優化證據。
来源与永久记录Sources and permanent record來源與永久記錄
NVIDIA FY2027Q2 GAAP tax guidanceNVIDIA FY2027Q2 GAAP tax guidanceNVIDIA FY2027Q2 GAAP tax guidance
NVIDIA original financial reportsNVIDIA original financial reportsNVIDIA original financial reports
Treasury daily 10Y ratesTreasury daily 10Y ratesTreasury daily 10Y rates
RA-NVDA-HORIZON-14-v0.1 · Protocol SHA256 28d06620ab69f34451a5ab0cddf955bcbc8558479b14dcd8de0ffbf0d0b0bd78
Public aggregate SHA256 1c230d7ca6b874a02e7e936008abc4691594fce2ea451084d4be34c4b49f7dc8
NVDA个股研究保留模型Retained NVDA equity-research modelNVDA個股研究保留模型 · 早期10月12日版本Early October 12 version早期10月12日版本 · H09原始研究Original H09 studyH09原始研究 · NVDA全部研究版本All NVDA research versionsNVDA全部研究版本