Educational research only - not investment advice, an offer, or a trade instruction. Confirm current data and do your own due diligence before acting.
NVDA trades against a final fair-value range of $223.19-$527.51, with the midpoint set by the accepted valuation synthesis rather than earlier draft model outputs.
Combined bear stress — AI capexCapital expendituresCash spent on acquiring or upgrading property, plant, and equipment. Splits into maintenance capex (sustaining current capacity) and growth capex (expanding capacity). digestion (-25% hyperscaler spend in FY2028), 15pp GPU training share loss, and one export control escalation — reduces NVDA FY2028 revenueRevenueRevenue is an income-statement line item used to reconcile revenue to operating profit, pre-tax income, net income, or per-share earnings. It should be compared across periods and against peer disclosure conventions. to ~$270B (vs $508B base). Gross margins compress to 65%, operating marginOperating marginOperating income (EBIT) divided by revenue. Captures profitability after both direct costs and operating expenses but before interest, tax, and non-operating items. to 48%, EPSEarnings per shareNet income divided by weighted-average diluted shares outstanding. The headline accounting earnings figure on a per-share basis. to ~$5.50. Applying 20x mature-semiconductor terminal multipleTerminal multipleThe exit P/E or EV/EBITDA we apply to the final year of an explicit forecast. Anchored to the business's long-run quality and the prevailing risk-free rate. yields a stress fair valueFair valueOur composite estimate of intrinsic per-share value, blended across DCF, exit-multiple, and reverse-DCF methods. Reported as a low/mid/high range to capture model uncertainty. of $110–130. This scenario requires simultaneous occurrence of three adverse catalysts; standalone probability is 5–8%. Near-term balance sheet support (net cash $51.5B, Altman Z 57.3) limits survivability risk.
Hyperscalers cut AI infrastructure spending 20–30% in FY2028 after capacity digestion and disappointing near-term AI ROI. Data center revenue plateaus at $280–300B by FY2028 vs $508B internal valuation cross-check. Revenue growth decelerates to single digits and the stock de-rates to a 20–22x mature semiconductor multiple.
Hyperscaler proprietary chips (TPU, Trainium, Maia) capture 25–30% of AI training compute within three years, compressing NVIDIA ASP and gross margins toward 65%. Pricing power erodes; operating margin falls 1,000bps from the FY2026 peak. The CUDA ecosystem advantage is circumvented rather than replicated.
Further restrictions beyond H20 ban eliminate all advanced AI chip sales to China, reducing addressable market by 15%+ and removing the incremental downside-cycle revenue floor. Geopolitical escalation accelerates customer dual-sourcing mandates globally.
Competitive pricing from AMD MI-series and hyperscaler custom silicon forces ASP reductions. Simultaneously, Blackwell production ramp yields unfavorable mix. Gross margin falls below 68% for two or more consecutive quarters, triggering multiple compression as internal valuation cross-check downgrades the quality premium.
Cash-flow quality is reflected in the OCFOperating cash flowCash generated from the company's core operations after working-capital changes but before capital expenditures. The first line of the cash-flow statement. / net incomeNet IncomeNet Income is an income-statement line item used to reconcile revenue to operating profit, pre-tax income, net income, or per-share earnings. It should be compared across periods and against peer disclosure conventions., accounting-quality, and ROICROICReturn on invested capital. Operating profit (after tax) divided by invested capital. The single best gauge of capital-efficiency. Spread over WACC = economic value created. rows above.
Capital allocation should be evaluated against reinvestment needs, balance-sheet strength, and shareholder returns.
