NVDA / V2 / METHOD SUMMARY
V2：可比增长与分阶段估值V2: comparable growth and staged valuationV2：可比增長與分階段估值
主公式与时间窗口Core formula and time windows主公式與時間窗口
T0 = valuation origin; T12 = T0 + 12 calendar months; T24 = T0 + 24 calendar months Current NTM EPS: [T0, T12) Target-date NTM EPS: [T12, T24) Pₛ(T12) = EPSₛ[T12,T24) × Exit P/Eₛ(T12) E[P(T12)] = Σ wₛ × (EPSₛ × P/Eₛ)
先逐季预测，再按目标窗口的重叠天数汇总净利润，并除以该窗口的加权平均稀释股数。不能把当前NTM、目标日NTM或更远财年EPS互换。情景先分别算价再加权，不能用平均EPS乘平均P/E。Forecast by quarter, allocate net income by overlap days, then divide by the window’s weighted-average diluted shares. Current NTM, target-date NTM and more distant fiscal-year EPS are not interchangeable. Price each scenario before weighting; mean EPS times mean P/E is not equivalent.先逐季預測，再按目標窗口的重疊天數彙總淨利潤，併除以該窗口的加權平均稀釋股數。不能把當前NTM、目標日NTM或更遠財年EPS互換。情景先分別算價再加權，不能用平均EPS乘平均P/E。
季度内按天均匀分摊是一项插值假设，无法代表实际交付节奏。价格标签满12个月，与目标日NTM盈利已经全部披露，是两个不同的成熟条件。Uniform daily allocation within a quarter is an interpolation assumption, not an actual delivery schedule. A matured 12-month price label and fully reported target-date NTM earnings are separate maturity conditions.季度內按天均勻分攤是一項插值假設，無法代表實際交付節奏。價格標籤滿12個月，與目標日NTM盈利已經全部披露，是兩個不同的成熟條件。
从总收入外推到业务驱动From aggregate extrapolation to business drivers從總收入外推到業務驅動
| 阶段Stage階段 | 方法Method方法 |
|---|---|
| 1 | 第一预测季度以公司指引为近端约束；情景偏离需要独立说明，不默认每季超指引。Anchor the first forecast quarter to company guidance; explain scenario deviations rather than assume recurring beats.第一預測季度以公司指引為近端約束；情景偏離需要獨立説明，不默認每季超指引。 |
| 2 | 随后2–3季分开受影响产品与其他业务，核对需求、交付、供应和客户部署；客户CapEx不等于公司收入。For the next 2–3 quarters, separate affected products from other business and test demand, delivery, supply and customer deployment. Customer capex is not company revenue.隨後2–3季分開受影響產品與其他業務，核對需求、交付、供應和客户部署；客户CapEx不等於公司收入。 |
| 3 | 远端增长分阶段收敛；增长持续期、成熟增速和利润率单独论证，不把短期冲击永久化。Transition longer-term growth by stage; justify growth duration, mature growth and margins separately without making temporary shocks permanent.遠端增長分階段收斂；增長持續期、成熟增速和利潤率單獨論證，不把短期衝擊永久化。 |
g_b,h = g_b,infinity + (g_b,4 - g_b,infinity) × 2^(-(h-4)/H_b)
这条半衰期公式是V2方法说明中的可执行收敛形式，不是已证明最优的参数模型，也不声称后续V2.2运行逐项采用该公式。V2.2前瞻运行使用统一年度收入与利润路径衔接；两个实现层次分别记录。This half-life equation is an executable convergence form in the V2 method description, not an optimally validated parameter model or a claim that the later V2.2 run uses this exact equation throughout. The V2.2 forward run links unified annual revenue and earnings paths; the implementation layers remain distinct.這條半衰期公式是V2方法説明中的可執行收斂形式，不是已證明最優的參數模型，也不聲稱後續V2.2運行逐項採用該公式。V2.2前瞻運行使用統一年度收入與利潤路徑銜接；兩個實現層次分別記錄。
可比收入不是加回损失收入Comparable revenue is not adding back lost sales可比收入不是加回損失收入
将“收入水平少了一块”与“剩余业务增速下降”分开。可比基数调整只改变对增长趋势的判断，不无条件把无法交付的收入加回；缺乏可靠销量或单价时，不伪造单位销量模型。Separate a loss of revenue level from slower growth in the remaining business. A comparable-base adjustment changes the interpretation of growth; it does not restore undeliverable sales. Do not invent unit/price models when reliable volume and pricing evidence is unavailable.將“收入水平少了一塊”與“剩餘業務增速下降”分開。可比基數調整隻改變對增長趨勢的判斷，不無條件把無法交付的收入加回；缺乏可靠銷量或單價時，不偽造單位銷量模型。
毛利与费用联动Linked margins and expenses毛利與費用聯動
毛利可随产品组合、换代、采购和存货成本恢复或恶化。研发与其他费用分别预测，再与已知全年费用指引核对。诊断版费用弹性公式只用于敏感性隔离，不把0.5当作实证估计值。Gross margin may recover or deteriorate with product mix, transitions, procurement and inventory costs. Forecast R&D and other expenses separately and reconcile known annual expense guidance. The diagnostic elasticity equation isolates sensitivity; 0.5 is not an empirically established estimate.毛利可隨產品組合、換代、採購和存貨成本恢復或惡化。研發與其他費用分別預測，再與已知全年費用指引核對。診斷版費用彈性公式只用于敏感性隔離，不把0.5當作實證估計值。
OPEXnew = OPEXold × (Revenuenew / Revenueold)^η
利润、SBC与现金流Earnings, SBC and cash flow利潤、SBC與現金流
Gross profit = Revenue × Gross margin EBIT = Gross profit − R&D − other operating expenses Core net income = (EBIT + recurring net non-operating income) × (1 − normalized tax rate) Target-window net income = Σ quarterly net income × overlap days / quarter days Target NTM EPS = Target-window net income / weighted-average diluted shares FCFF = EBIT × (1 − tax rate) + D&A − CapEx − ΔNWC
Core保留SBC经济成本，排除未来投资重估收益和离散税项；不自动等于供应商或公司Non-GAAP。SBC、回购现金、股份稀释和净现金必须使用一致口径，不免费加回或重复扣减。FCFF用WACC折现；股东现金流用权益成本折现。Core retains the economic cost of SBC and excludes future investment revaluation gains and discrete tax items; it is not automatically vendor or company non-GAAP. Use consistent SBC, buyback cash, dilution and net-cash treatment without free add-backs or duplicate deductions. Discount FCFF at WACC and equity cash flows at the cost of equity.Core保留SBC經濟成本，排除未來投資重估收益和離散税項；不自動等於供應商或公司Non-GAAP。SBC、回購現金、股份稀釋和淨現金必須使用一致口徑，不免費加回或重複扣減。FCFF用WACC折現；股東現金流用權益成本折現。
P/E：六年盈利资本化P/E: six-year earnings capitalizationP/E：六年盈利資本化
e₁ = 1; eᵧ₊₁ = eᵧ × (1 + gᵧ₊₁)
bᵧ = gᵧ₊₁ / incremental ROEᵧ; fᵧ = eᵧ × (1 − bᵧ)
kₑ = Rf + β × ERP
P/E = Σ(y=1..6) f_y/(1+k_e)^y
+ e₆(1+g∞)(1−g∞/ROE∞) / [(kₑ−g∞)(1+kₑ)⁶]倍数取决于盈利持续性、再投资回报和权益成本，不是固定行业平均值。增长/增量ROE只在相应再投资假设成立时适用；利润率恢复、效率提升或回购增厚不能全部当作新增资本驱动的增长。需满足权益成本高于永续增长，并检查终值占比与参数敏感性。The multiple reflects earnings duration, reinvestment returns and cost of equity, not a fixed industry average. Growth divided by incremental ROE is valid only under the corresponding reinvestment assumptions; margin recovery, efficiency and buyback accretion are not all capital-driven growth. Cost of equity must exceed perpetual growth; inspect terminal-value dependence and parameter sensitivity.倍數取決於盈利持續性、再投資回報和權益成本，不是固定行業平均值。增長/增量ROE只在相應再投資假設成立時適用；利潤率恢復、效率提升或回購增厚不能全部當作新增資本驅動的增長。需滿足權益成本高於永續增長，並檢查終值佔比與參數敏感性。
同一经营情景决定盈利与倍数One operating scenario drives earnings and the multiple同一經營情景決定盈利與倍數
目标窗口后续盈利增长、增长持续时间、再投资与现金分配采用一致路径。每个历史起点独立使用当时可得利率、风险溢价和Beta；没有历史共识不伪造市场Forward P/E。回购现金分配不能与无成本EPS增厚重复计算。Use a consistent path for earnings after the target window, growth duration, reinvestment and distributions. At each historical origin, use then-available rates, equity risk premium and beta. Missing historical consensus cannot be replaced by fabricated market forward P/E. Do not count buyback distributions and cost-free EPS accretion twice.目標窗口後續盈利增長、增長持續時間、再投資與現金分配採用一致路徑。每個歷史起點獨立使用當時可得利率、風險溢價和Beta；沒有歷史共識不偽造市場Forward P/E。回購現金分配不能與無成本EPS增厚重複計算。
DCF是交叉检查，不是强行平均DCF cross-check, not a forced averageDCF是交叉檢查，不是強行平均
DCF沿用经营预测并独立检查CapEx、营运资本和资本支持。T12重新折现剩余现金流，滚动现金、回购、分红及股份，不能把T0价值简单乘1＋WACC。P/E与DCF共享输入，不是统计独立证据，也不机械各占50%。后续V2.2前瞻运行没有独立完成新的T0/T12 DCF与完整现金—股份桥，不宣传已得到双模型验证。DCF shares operating forecasts while separately checking capex, working capital and capital support. At T12, discount remaining cash flows and roll forward cash, buybacks, dividends and shares; do not multiply T0 value by 1 + WACC. Shared inputs mean P/E and DCF are not statistically independent or automatic 50/50 components. The later V2.2 forward run did not independently complete a new T0/T12 DCF and full cash/share bridge, so dual-model validation is not claimed.DCF沿用經營預測並獨立檢查CapEx、營運資本和資本支持。T12重新折現剩餘現金流，滾動現金、回購、分紅及股份，不能把T0價值簡單乘1＋WACC。P/E與DCF共享輸入，不是統計獨立證據，也不機械各佔50%。後續V2.2前瞻運行沒有獨立完成新的T0/T12 DCF與完整現金—股份橋，不宣傳已得到雙模型驗證。
版本边界与检验要求Version boundaries and evaluation requirements版本邊界與檢驗要求
局部诊断仍保留旧0.85衰减与固定P/E，用来隔离盈利修正；不能称为完整版V2回测。完整方法与前瞻运行另行登记。情景权重属于主观研究判断，不等于触及率、上涨概率或连续分布分位数。The partial diagnostic retains the old 0.85 decay and fixed P/E to isolate earnings repairs; it is not a full V2 backtest. The complete method and forward run are separately recorded. Scenario weights are subjective research judgments, not touch rates, upside probabilities or continuous-distribution quantiles.局部診斷仍保留舊0.85衰減與固定P/E，用來隔離盈利修正；不能稱為完整版V2回測。完整方法與前瞻運行另行登記。情景權重屬於主觀研究判斷，不等於觸及率、上漲概率或連續分佈分位數。
评价要同时看收入/毛利/EPS误差、期末价格误差、最高收盘价触及与路径风险。已用于改模型的旧样本不是新的留出样本；不使用旧版平均达到比例反向抬高目标。Evaluate revenue/margin/EPS errors, endpoint price error, maximum-close threshold attainment and path risk together. Samples used to redesign the model are not new holdouts. Do not use the old mean attainment ratio to inflate targets.評價要同時看收入/毛利/EPS誤差、期末價格誤差、最高收盤價觸及與路徑風險。已用於改模型的舊樣本不是新的留出樣本；不使用舊版平均達到比例反向抬高目標。
来源与发布边界Sources and publication boundary來源與發佈邊界
依据“英伟达估值研究”对话及2026-09-25原始研究稿整理，2026-09-26迁入本站。V1/V2属于该对话的研究版本，不等同于本站P11.1、N2.2或模型族v0.1；不将不同分支的参数和成绩混合。Compiled from the NVIDIA Valuation Research conversation and the original 2026-09-25 report; migrated to this site on 2026-09-26. These V1/V2 labels belong to that research conversation, not the site’s P11.1, N2.2 or family v0.1. Parameters and results from different branches are not blended.依據“英偉達估值研究”對話及2026-09-25原始研究稿整理，2026-09-26遷入本站。V1/V2屬於該對話的研究版本，不等同於本站P11.1、N2.2或模型族v0.1；不將不同分支的參數和成績混合。
模型结构是已有研究的摘要，不代表新的估值运行。本次只刊登已有V1历史检验摘要，不刊登V2单点诊断价格、前瞻目标价或尚未取得的十年回测。未对完整回测底稿重新运行，不声称独立复现。The structures summarize existing research, not a new valuation run. This release includes only the available V1 historical-evaluation summary; it excludes V2 single-origin diagnostic prices, forward targets and unavailable ten-year backtests. Complete backtest working files were not rerun; independent replication is not claimed.模型結構是已有研究的摘要，不代表新的估值運行。本次只刊登已有V1歷史檢驗摘要，不刊登V2單點診斷價格、前瞻目標價或尚未取得的十年回測。未對完整回測底稿重新運行，不聲稱獨立復現。
本次整理数据 SHA-256Compiled data SHA-256本次整理數據 SHA-256: 999064b7916f493d522c086041c7e05fe6ffdcb5accb8cf21d9eeb4eb0a2fb06
内容校验值不是历史公开时间证明或第三方审计认证。研究源稿与对话记录保留在本地；不提供失效的附件链接。The checksum is not proof of historical publication or third-party audit certification. Source reports and conversation provenance are retained locally; no expired attachment links are presented.內容校驗值不是歷史公開時間證明或第三方審計認證。研究源稿與對話記錄保留在本地；不提供失效的附件鏈接。