Nvidia: our view and what our model says about 2030
Our view
Nvidia is the dominant supplier of AI computing systems, still growing about 90% a year at a $400B scale. Fiscal 2028 is backed by guidance and $279B of supply contracts; after that, growth depends on AI spending continuing to compound, and Nvidia increasingly finances the customers who buy its chips. Today's price (about 13× next year's consensus earnings) is consistent with earnings that stop growing after 2028. Our model has growth slowing rather than reversing, so most simulations end above the hurdle; the main way to lose money is an AI-spending downturn.
- Price, Oct 23 close
- $210.96
- Hurdle, end-2030
- $303
- Median value, end-2030
- $440
- P10–P90 range, end-2030
- $175–$1,030
- Simulations beating hurdle
- 70%
- End below today's price
- 16%
Prices: closing price, Oct 23, 2026 · Company data: 10-Q, Jul 26, 2026 · What these numbers mean
Market data as of the Oct 23, 2026 close ($210.96). The hurdle and the probabilities depend on that price. Values per share at Dec 31, 2030, undiscounted, from 50,000 simulations.
Snapshot at the Oct 23, 2026 close
- Market value
- $5.13T
- Enterprise value
- $5.10T
- Price/sales
- 16.9×
- EV/sales
- 16.8×
- P/E
- 26.6×
- Price/FCF
- 40.4×
- Revenue growth
- +83%
- FCF margin
- 42%
- Net cash
- $23.2B
Our calculation from the company's filings (latest: 10-Q for the quarter to Jul 26, 2026, filed Aug 26, 2026); not investment advice. Revenue, earnings and cash flow: 12 months to Jul 26, 2026. Price $210.96. n/m = not meaningful (e.g. negative earnings).
Earlier versions
| Version | Price used | Horizon | Median value | P10–P90 | Beat chance |
|---|---|---|---|---|---|
| Oct 28, 2026THIS VERSION | $210.96 (Oct 23, 2026) | end-2030 | $440 | $175–$1,030 | 70% |
No earlier versions yet. Each row is frozen as published: when we update this view, the previous numbers stay here.
Version 1.15 · Oct 29, 2026
Editors hold this stock. Market data as of the Oct 23, 2026 close ($210.96); latest filings: 10-K for FY2026 and 10-Q for Q2 FY2027 (quarter to Jul 26, 2026). Nvidia’s fiscal year ends in late January (FY2030 runs from about February 2029 to January 2030). Labels: Fact, Company (management claims), Analyst, Our view.
Simulation result (value per share at Dec 31, 2030, undiscounted; 50,000 simulations):
- P10 $175 · P50 $440 · P90 $1,030 (middle half $272–$700; mean $535).
- Hurdle $303 (the end-2030 price that earns 9% a year from today): 70% of simulations beat it (about +19% a year at the median); 16% end below today’s price.
What would move the result
- The AI-spending cycle, by far the biggest driver: about a 16% chance a year of a downturn lasting about five quarters, balanced by booms. With a neutral economy (no recessions or sector cycles), the median would be about $500.
- AI-spending growth after 2028 (about 14%, 7% and 5% in fiscal 2029–31). With a slower path, the median would be about $340–$355.
- Nvidia’s share: up about 15 points in fiscal 2028, as the ~70% guide implies, then flat to slightly down.
- The exit multiple, the second-biggest driver: Nvidia’s own past P/E relative to the S&P 500 (0.88×, range 0.42–1.86×), giving a median of about 20× the following year’s earnings (today: about 18×).
- A Taiwan/TSMC disruption (12% of simulations). Without it, the median would be about $450.
Our view in brief
- Dominant and accelerating. Q2 FY2027 revenue was $96.2B (+106%); about 70% growth is guided for FY2028, limited by supply rather than orders.
- Visible to FY2028, open after. Beyond FY2028 our model trims AI-spending growth to about 7% and 5% in fiscal 2030–31, reflecting Gartner’s warning that AI revenue may fall short of what the spending needs.
- The edge is widening at the system level, but the biggest customers now design credible chips, and Nvidia increasingly wins business with financing.
- Customer financing is the new risk. About $285B of stakes, commitments and guarantees tie Nvidia to its customers; about a quarter of next year’s business comes from AI labs it helps finance.
- Clean accounts, newer flags: receivables and inventory outgrowing revenue, GAAP gains on stakes in customers, and about half of free cash flow going back into the ecosystem.
- Value. Most simulations end above the hurdle; the downside comes almost entirely from an AI-spending downturn.
- Watch Q3 results (Nov 17), the Q4 gross-margin trough, hyperscalers’ 2027 capex budgets and any losses on customer financing.
1. The business in one paragraph
Nvidia sells complete AI data-centre systems (chips, networking and CUDA software), built by partners such as TSMC. Data Center was 93% of Q2 FY2027’s $96.2B of revenue, which grew 106% at a 75% gross margin and 66% operating margin (10-Q), so our model treats Nvidia as one bet: AI-spending growth times Nvidia’s share. The full facts are in Nvidia’s Long View → /stocks/nvda/long-view/
Confidence: high.
2. Where growth comes from (3–5 years)
Growth through FY2028 is about as well underwritten as a forecast can be: about 70% guided growth, backed by $279B of supply contracts. After FY2028, it depends on whether AI spending keeps compounding, and none of that is contracted. Our model follows guidance and consensus to fiscal 2029, then trims growth, so the post-2028 path is the biggest open question in the value.
| FY2025 | FY2026 | FY2027E | FY2028E | FY2029E | FY2030 | |
|---|---|---|---|---|---|---|
| Revenue ($B) | 130.5 | 215.9 | 411.7 | 692 | ~903 | not public |
| Growth | 114% | 65% | 91% | 68% (Company: ~70%) | ~30% | |
| EPS, non-GAAP (consensus) | 4.77* | 9.31 | 15.91 | ~20.5 |
Consensus: Yahoo Finance and MarketScreener, Oct 23, 2026 (*aggregator basis; notes in the facts below).
The facts behind this
Consensus table notes: *aggregator basis. Nvidia’s non-GAAP measures include stock-based pay from FY2027 on, so older years are not strictly comparable. Estimates: Yahoo Finance (54–57 analysts) and MarketScreener (62 analysts), Oct 23, 2026. FY2029 is thin (one aggregator; another shows $921B). No FY2030 consensus is public; for FY2030 our model assumes growth slows to high single digits (section 7).
The drivers, roughly in order of visibility:
- The Vera Rubin generation (Company). Production shipments began in August 2026. Nvidia says it has purchase orders from every major cloud, AI cloud and system maker, and expects Rubin to be about 20% of Data Center revenue in Q3. Nvidia’s estimated revenue per gigawatt of data-centre capacity rises from about $25B (Blackwell) to about $40B (Rubin), because each rack includes more Nvidia content: CPUs, three types of networking, and Groq inference chips.
- New product lines (Company). Vera CPUs sold on their own (CPU revenue is expected to more than double in FY2028), Groq 3 inference systems (shipping from late Q3), and networking, which already grew 18% sequentially in Q2.
- Customers beyond the big clouds (Company). ACIE (AI clouds, model makers, enterprises and governments) grew 138% year on year. Management says it will be about half of Data Center revenue and that sovereign demand tripled year on year.
- New financing models (Company). Nvidia now guarantees leases, buys back cloud capacity and invests in its customers (section 4). It says AI labs whose purchases it helps finance will be “roughly a quarter” of next year’s business.
Size of the opportunity (independent estimates):
- Dell’Oro (Aug 18, 2026) expects worldwide data-centre capex to pass $3T by 2030, with high-end accelerators the largest share, nearly double its January 2026 forecast.
- Gartner expects nearly $1T of AI infrastructure spending in 2026 and about $6T more in 2027–30. It also warns that AI revenue may reach only about $3.6T against the $8.5T needed to earn a 12% return on that spending.
- Our view: Nvidia’s sales in calendar 2026 (about FY2027’s $412B) are roughly 40% of Gartner’s 2026 figure. Our model’s path to FY2030 needs Nvidia to take a similar 35–50% share of AI infrastructure spending in 2029, depending on whose forecast is used. That is plausible if the build-out continues, but these forecasts are themselves being revised up every few months.
Track record (Fact). Financial guidance has been reliable. Over the last three quarters revenue came in about 5% above the midpoint each time (4.6–5.7%), and gross margin landed on guidance.
| Quarter | Revenue guide (mid) | Actual | vs guide |
|---|---|---|---|
| Q4 FY2026 | $65.0B | $68.1B | +4.8% |
| Q1 FY2027 | $78.0B | $81.6B | +4.6% |
| Q2 FY2027 | $91.0B | $96.2B | +5.7% |
| Q3 FY2027 | $108.0B | due Nov 17 | consensus $109.0B |
Product timelines are less clean:
- Blackwell slipped about a quarter in 2024 because of a design flaw that needed a mask change, which Huang acknowledged as Nvidia’s fault. Rubin arrived on schedule.
- Two newer commitments have softened. The September 2025 plan to invest up to $100B in OpenAI became a $30B equity investment, and Huang said in March 2026 that $100B was “probably not in the cards”. The AI-cloud revenue-share program was partly paused weeks after launch, according to the WSJ (Aug 28, 2026); Nvidia says the model remains in place.
- The 70% FY2028 guide is the first year-ahead revenue guidance in Nvidia’s history, so it is also a new credibility test.
Backlog and pipeline quality (Fact/Our view). Nvidia does not report a backlog, and contracted multi-year revenue (remaining performance obligations) is only $3.2B. Forward visibility comes from management statements and supply contracts instead.
- In October 2025, management said it had $500B of Blackwell and Rubin bookings through calendar 2026, and data-centre revenue over FY2026–27 looks set to exceed that.
- For FY2028, the question is less whether orders exist than who is behind them. About a quarter of next year’s business depends on AI labs that need Nvidia’s credit support, and the largest of them, OpenAI, is not investment grade.
- The 70% guide is limited by supply, not by orders, according to management.
Pending transformative event (Our view): none. The largest pending deal is Hugging Face ($11.9B cash plus up to $1.0B of retention equity, signed Sep 2, 2026, closing expected in the first half of 2027). That is about 0.25% of Nvidia’s market value and adds a platform, not revenue. The event that matters most is product: the Vera Rubin ramp, which began in August 2026.
Confidence: high for FY2027–28; low beyond FY2028.
3. The edge (moat)
The edge is real and arguably widening at the system level, but the biggest customers are now credible chip designers, and Nvidia increasingly uses financing, not just technology, to win business. In our model this is Nvidia’s share: a large gain in fiscal 2028, then flat to slightly down, plus a 25% chance that a big cloud customer moves part of its inference to its own chips.
The facts behind this
Why customers choose Nvidia (Company/Our view):
- It sells the full system: GPU, CPU, NVLink interconnect, scale-out networking, software and reference designs for whole data centres (DSX). Content per gigawatt is still rising.
- One platform runs every model and workload, so a buyer’s investment keeps its value as models change. That makes Nvidia hardware easier to finance and resell.
- CUDA and its libraries are the default for developers, and the pending Hugging Face deal adds the main open-model hub.
- Scale in the supply chain. $279B of supply commitments, mostly memory, lock up components that rivals also need.
- It can finance customers: about $50B invested in frontier labs (Company) and $108.5B of guarantees (maximum exposure; the $105B part phases in from FY2029).
Signs the edge is widening: customers are buying Nvidia’s newer products (CPUs, networking, inference systems), revenue per gigawatt is rising, and gross margin stayed at 75% through the Blackwell Ultra ramp.
Signs it is narrowing:
- OpenAI’s own Broadcom-built inference chip, Jalapeño, beat Blackwell on performance per watt in SemiAnalysis tests and was roughly even with Rubin on cost per token. It is still at engineering-sample stage, with broad rollout planned for 2027.
- Google now sells TPUs directly (Anthropic has committed to up to 1 million). AMD’s MI450 racks start shipping to OpenAI (up to 6 GW) and Meta (6 GW) in the second half of 2026. Amazon and Meta have their own chips.
- China is closed to Nvidia’s data-centre products, which Nvidia says is helping Huawei and other local rivals build ecosystems.
- Gross margin is guided down to 72–73% in FY2028 as memory costs rise. Seaport’s Jay Goldberg (Sell) calls the financing deals “another form of discounting.”
Competitors (Fact, 10-K): AMD, Huawei and Intel in GPUs and accelerators; Broadcom and Marvell designing custom chips for Google, Meta, Amazon, OpenAI and others; cloud providers’ own chips; networking rivals (Broadcom, Arista, Cisco). Market share isn’t disclosed. Third-party estimates for 2026 range from about 70% to 85% of AI accelerator revenue, measured inconsistently (low confidence).
Confidence: medium.
4. The worst realistic risk
The worst realistic case is cyclical: AI infrastructure spending pauses in 2028–29, just as Nvidia carries hundreds of billions of supply commitments and large stakes in, and guarantees for, the same customers whose spending has slowed. Our model captures this through the AI-spending cycle and a gross-margin write-down in a downturn, the main source of the losing simulations. Survival is not the test: Nvidia could absorb a bad year without raising money, though it would show up as write-downs and lower buybacks.
Our view: about $285B of stakes, commitments and guarantees tie Nvidia to its customers’ finances, against about $690B of expected FY2028 revenue (gross amounts of different kinds, not an estimate of losses). The customer and contract facts are in Nvidia’s Long View → /stocks/nvda/long-view/
The facts behind this
| Risk | How likely (3–5 yrs) | How bad | Notes |
|---|---|---|---|
| AI capex digestion (spending outruns returns, financing tightens) | Medium | Severe | Gartner’s revenue-shortfall math. Neocloud debt stress in Oct 2026 (CoreWeave bonds). Seaport sees 55 GW announced vs ~30 GW buildable next year. |
| Customer-financed demand unwinds | Low–medium | High | About a quarter of next year’s business is tied to labs Nvidia supports. There is a $105B guarantee for OpenAI’s Ohio site from FY2029. |
| Custom chips take the inference market | Medium | Medium–high | Jalapeño, TPU and Trainium ramp in 2027. Hyperscalers are 55% of Data Center revenue. |
| Supply-commitment write-downs | Medium in a downturn | Medium | $279B committed, $175B of it due in FY2028–29. Precedents: a $4.5B H20 charge (FY2026) and $0.4B for H200. |
| Geopolitics (China, Taiwan, export rules) | Medium | Medium; severe for Taiwan | China data centre is already excluded from guidance. TSMC dependence is near total. |
Legal and regulatory docket (Fact, 10-Q Note 10 and risk factors):
| Case or issue | Status | Accrued | Next date |
|---|---|---|---|
| Securities class action over crypto-era channel inventory (2017–18), N.D. Cal. | Revived by the 9th Circuit, Supreme Court dismissed Nvidia’s appeal (Dec 2024), class certified Mar 25, 2026 | None (“reasonably possible”, not estimable) | Not disclosed |
| Derivative suits on the same facts (N.D. Cal., D. Del., Del. Chancery) | Stayed | None | After the class action |
| China SAMR review of Mellanox merger conditions | Preliminary finding of violation (Sep 15, 2025) | None disclosed | No date |
| Competition inquiries (EU, US, UK, China, Korea, France) | Information requests on supply allocation, investments in model makers and customer agreements | None | None disclosed |
| US export controls and China | Effectively shut out of China data centre. H200 licences granted but Beijing blocks deliveries; 25% US tariff on licensed H200s | $0.4B H200 charge (1H FY2027) | GAIN AI Act and Remote Access Security Act pending in Congress |
| Hugging Face acquisition | Signed Sep 2, 2026 | n/a | Close expected 1H 2027 |
| Tax | IRS examining FY2023–24 | $5.0B unrecognised tax benefits incl. interest (Item 2); $5.6B long-term income tax payable, mostly these | Not disclosed |
Dependence map (Fact unless marked):
- Customers. Three direct buyers, mostly assemblers building for clouds, were 16%, 15% and 13% of first-half revenue. Five customers hold 70% of receivables. Nvidia estimates that “one AI research and deployment company” (OpenAI, by context; Our view) contributed a meaningful share of revenue through cloud partners.
- Suppliers. TSMC makes the chips and does the advanced packaging. Three memory makers (SK hynix, Samsung, Micron) supply high-bandwidth memory, which is now the binding constraint. Racks are assembled in Taiwan and Asia.
- Government. US export licences, the Chinese regulator (SAMR) and Chinese customs, tariffs, and power and permitting for the data centres that house the product.
Customer financing: what makes up the ~$285B:
- Payment terms (Fact, 10-Q). Large investment-grade buyers get payment terms of 90 days to a year.
- Equity stakes (Fact, 10-Q and 13F). Nvidia held about $95.6B of equity stakes at Jul 26: $47.7B listed and $47.9B private. They are mostly in customers and partners, though some are not customers (Intel, Synopsys).
- Largest listed stakes (Fact, 13F; market data). Intel, SpaceX (which now owns xAI) and CoreWeave were worth $30.0B, $21.0B and $4.7B at the Jun 30 13F marks; at Oct 23 prices the same shares are worth about $18.7B, $18.1B and $2.6B (about $39B against $56B).
- Private stakes (press). OpenAI (a $30B investment, per press) and Anthropic (up to $10B).
- Commitments (Fact, 10-Q). $25B more of committed investments, $36B of six-year commitments to rent back capacity from AI clouds, and $20B of data-centre leases it plans to pass to third parties.
- Guarantees (Fact, 10-Q and 8-K). $108.5B at maximum exposure.
- The total (Our view): ~$96B of equity stakes (Jul 26 book value; the listed part has fallen since) + $25B of committed investments + $36B of rent-back commitments + $20B of leases to be reassigned + $108.5B of guarantees ≈ $285B.
- What reduces the exposure. The rent-back commitments shrink as the AI clouds resell capacity to third parties. The AI-cloud lease guarantees are partly backed by $712M in escrow and decline over 5–7 years. The $105B SB Energy guarantee for OpenAI’s Ohio site only takes effect as each phase’s lease starts (first in FY2029), is limited to defined portions of lease and power payments, and declines over each 20-year lease; Nvidia also has an option to back about 3.8 GW more.
Product-programme risk (Fact/Company): Rubin is on plan, with volume ramping through the first half of 2027, but supply is expected to stay a bottleneck through FY2028. Groq 3 went from licence (December 2025) to volume shipments in about nine months. The revenue-share program is the one recent initiative that has visibly stumbled.
Survival check (Fact/Our view): passes with a wide margin.
- Cash and marketable debt total $56.6B, with another $47.7B of listed equities at Jul 26 (of which $41.9B under lock-up), against $33.4B of debt, of which $1B is due within a year (10-Q Q2 FY2027).
- Operating cash flow was $74B in the first half. Nvidia raised $25B of bonds in June 2026, which it did not need to.
- Supply commitments for FY2028 alone are $87B, and the guarantees switch on from FY2029.
Confidence: medium–high on facts, medium on probabilities.
5. Financial quality (read the notes)
Period and source: latest quarter (Q2 fiscal 2027, quarter to Jul 26, 2026; 10-Q filed Aug 26, 2026) and the first half of fiscal 2027, compared with the 10-K for fiscal 2026 (year to Jan 25, 2026).
The core numbers are high quality: margins are real, controls are clean, and the adjusted measures became more conservative this year. The newer issues shape our model: we value Nvidia on non-GAAP earnings (which now include stock pay) because GAAP profit is inflated by gains on stakes in customers, and we assume less than all of earnings turns into free cash flow while receivables and inventory build. Buybacks more than offset stock pay, so the share count is falling slowly. The full facts are in Nvidia’s Long View → /stocks/nvda/long-view/
The facts behind this
| Check | Finding | Source | Flag |
|---|---|---|---|
| Revenue vs gross profit growth | Q2: revenue +106%, gross profit +113%. FY2026: +65% vs +57% (H20 charge). | 10-Q (Q2); 10-K (FY2026) | No |
| Receivables vs revenue | Receivables +64% in six months ($38.5B → $63.1B) vs quarterly revenue +41%. DSO 51 → 60 days because of extended payment terms. | 10-Q balance sheet (Jan 25 vs Jul 26); Q1–Q2 FY2027 CFO commentaries furnished with the earnings 8-Ks (DSO) | Yes |
| Inventory vs revenue | Inventory +48% ($21.4B → $31.6B). Raw materials tripled to $11.3B ahead of Rubin. Inventory provisions $1.6B in the first half. | 10-Q balance sheet and inventory note | Watch |
| Stock pay vs free cash flow | First half: $4.0B vs $69.9B (6%). Buybacks more than offset dilution. | 10-Q cash-flow statement | No |
| Changes in “adjusted” definitions | From Q1 FY2027, non-GAAP includes stock pay, which makes it more conservative. Market platforms were recast in Q1, and one customer moved from ACIE to Hyperscale in Q2. | 8-K earnings releases and CFO commentaries, Q1–Q2 FY2027 | Positive, but segments are less comparable |
| Estimate catch-ups | None disclosed | 10-Q | No |
| Capitalised costs | No material capitalised development costs | 10-K, 10-Q | No |
| Goodwill vs equity | Jul 26: $21.1B vs $229B (9%). $14.4B came from the Groq licence-and-hire deal. | Q2 FY2027 10-Q balance sheet; FY2026 10-K, Note 2 (Groq goodwill) | Note |
| Material weaknesses | None. The auditor’s (PwC) critical audit matter is inventory and purchase-commitment provisions. | 10-K FY2026 (auditor’s report) | No |
| Accounting-policy changes | None material | 10-Q | No |
| Receivable sales or factoring | None disclosed | 10-Q | No |
Items from the notes that stood out (10-Q Q2 FY2027 unless marked):
- GAAP profit includes large investment gains. In the first half, $23.7B of pre-tax income (17%) came from gains on equity stakes, mostly unrealised; GAAP net income was $118.0B against $99.5B non-GAAP (8-K earnings releases). The gains come from rising prices of listed holdings (Intel, SpaceX) and from new funding rounds that revalue private holdings ($7.5B of private-stake mark-ups in the first half); many of these companies are also Nvidia customers. $36.9B of the listed stakes are under short-term lock-up, plus $5.0B through December 2027. The largest listed holdings have fallen since Jul 26 (section 4), so second-half GAAP results may show equity losses.
- Part of the cash flow goes back to customers. First-half free cash flow was $69.9B. Net purchases of equity stakes were $35.2B, and the Groq payment was $2.9B, so free cash flow after strategic investments was about $32B, less than half of the headline. Nvidia’s policy is to return at least 50% of free cash flow to shareholders. It returned about 65% in the first half ($39.0B of buybacks in cash plus $6.3B of dividends, against $69.9B of free cash flow; the Q2 CFO commentary reports nearly $26.0B returned in that quarter), in the same half that it issued $25B of bonds.
- Customer financing affects revenue and cash. The longer payment terms are disclosed and limited to investment-grade buyers, but they explain why operating cash flow fell from $50.3B in Q1 to $24.1B in Q2 on higher revenue (10-Qs Q1 and Q2). The rent-back agreements (Nvidia buys cloud capacity from customers that bought its chips) raise an accounting question: under revenue rules, payments to customers reduce revenue unless they buy a distinct service at fair value. The filings do not say how this is treated (open question).
- The guarantees are large but barely appear on the balance sheet. The $3.5B of AI-cloud lease guarantees are recorded as credit derivatives with “not significant” fair value; partners have put $712M in escrow. The $105B SB Energy and OpenAI guarantee starts phasing in from FY2029 and shrinks as OpenAI pays rent (8-K, Aug 17, 2026, and Q2 FY2027 CFO commentary furnished with the Aug 26 earnings 8-K).
- The Groq structure (FY2026 10-K, Note 2). Nvidia agreed total consideration of $17B ($13B paid at closing and $4B, including imputed interest, payable within one year) for a non-exclusive licence plus staff, recording $14.4B of goodwill and $2.5B of developed technology. Our view: the structure avoids buying the company outright, which likely kept it outside merger review. It is a template to watch (Hugging Face, by contrast, is a proposed acquisition under a definitive agreement, subject to closing conditions; September 2026 8-K).
- Calendar effects. FY2027 has 53 weeks, and Q4 has 14 (10-K). This adds roughly 2% to FY2027 growth and slightly flatters the Q4 comparison.
Share-count bridge (Fact): single share class with one vote per share. No convertibles, preferred stock or issued warrants.
| Step | Shares |
|---|---|
| Outstanding, Aug 21, 2026 | 24.1B (24.3B on Feb 20, 2026) |
| Basic weighted average, Q2 | 24.19B |
| + dilutive equity awards | +95M → 24.29B diluted |
| Unvested RSUs and PSUs outstanding | 168M (0.7%) |
| Pro forma: Hugging Face retention equity | up to ~$1.0B (~5M shares at today’s price) |
| Buybacks, first half | −203M shares for $39.8B (average ~$196); $99.3B left under the repurchase programme |
Source: 10-Q Q2 FY2027 (cover page, earnings-per-share note, share repurchases), except the Feb 20, 2026 outstanding share count (FY2026 10-K cover) and the Hugging Face line (8-K, Sep 3, 2026; the share count at today’s price is our estimate).
Confidence: high on facts, medium on interpretation.
6. Valuation today
On consensus numbers Nvidia is priced like a mature, slow-growing company: about 13× next year’s earnings, below Broadcom or TSMC. The market is discounting how long current earnings will last.
What the price implies (Our view): a 9% annual return to Dec 31, 2030 needs a price of about $303 then. At 16× 2031 earnings, that requires EPS of about $19, or about $810B of revenue at a 51% net margin, below the ~$900B consensus for fiscal 2029. So today’s price is consistent with earnings that stop growing after 2028 and slip back.
Our simulation’s median 2031 EPS is about $21.10, above that level. The difference is the multiple: today’s price is about 10× that EPS, while the simulation’s median exit multiple is about 20×.
The facts behind this
Fact: market value is $5.09T on 24.15B shares. Net cash (cash and debt securities minus debt) is $23B, so enterprise value is $5.07T. The equity portfolio ($95.6B at Jul 26, less at Oct 23 prices; about 2% of market value) is excluded from net cash.
| Company (Oct 23 close) | Market cap | EV/sales (TTM) | EV/EBITDA (TTM) | Fwd P/E | Op. margin (TTM) |
|---|---|---|---|---|---|
| Nvidia | $5.09T | 16.7× | 25.2× | 17.4× (13.3× FY2028) | ~65% |
| Broadcom | $1.54T | 17.6× | 30.1× | 18.5× | 49% |
| TSMC | $1.90T | 11.6× | 15.9× | 19.6× | 58% |
| AMD | $0.84T | 20.2× | 87.3× | 46.5× | 16% |
| Marvell | $0.20T | 21.4× | 70.8× | 40.4× | 17% |
Basis: StockAnalysis (S&P Global data), Oct 23, 2026. EBITDA is reported EBITDA, not company-adjusted. Forward P/E is next twelve months; other sources that use the next fiscal year give different figures for Broadcom, TSMC and AMD. TSMC’s $1.90T is StockAnalysis’ figure; on the ADR price it is about $2.17T. Broadcom’s TTM operating margin is 54% on Yahoo’s basis. Medium confidence.
Against its own history: Clime’s Jason Teh notes that Nvidia’s forward P/E has more than halved in two years, to about 17× (Morningstar, Oct 21, 2026). The stock is up about 16% over 52 weeks but about 13% below its high of about $243 (press report).
Analyst views. The Street is almost uniformly positive: 59 of 62 analysts rate it a buy, 2 a hold and 1 a sell (StockAnalysis and Yahoo, Oct 2026; the two split Strong Buy and Buy differently). The average 12-month target is $329 (median $315; range $180–$515).
- Recent reiterations cluster near $300: Morgan Stanley (Joseph Moore, $300, Oct 22), Citi (Atif Malik, $315), Goldman Sachs (James Schneider, $300). Raymond James’ Simon Leopold said $1T of revenue in FY2029 “seems possible”.
- The dissenter, Seaport’s Jay Goldberg (Sell, $180), argues that competition is forcing Nvidia to finance its sales, and that the industry may build only about 30 GW of the 55 GW announced for next year. He concedes the next few months should be strong.
- Short interest is 1.3% of the float.
How today's price compares with our simulation
Reverse arithmetic: Dec 31, 2030 is 4.19 years away; EPS of about $19 at 16× means net income of about $415B.
At a 16× multiple, today’s price is consistent with earnings that stop growing after 2028 and slip back (about $810B of 2031 revenue). Our simulation’s median value at Dec 31, 2030 is about $440 a share (middle half of simulations: $272–$700; mean $535). That is above the hurdle (the end-2030 price that would earn a 9% yearly return from today’s price) of $303. In 70% of simulations the stock beats the hurdle; in 16% it ends below today’s price.
Confidence: medium.
7. Value in 3–5 years, as scenarios
Our simulation puts Nvidia’s median value at Dec 31, 2030 at about $440 a share, above the $303 hurdle, and 70% of simulations beat it. The weak quarter of outcomes (P25 $272) ends between today’s price and the hurdle, and 16% end below today’s price. Almost all of that downside comes from an AI-spending downturn (full range in the Our view box at the top).
Method: Valued directly at end-2030.
Implied end-2030 multiple: our end-2030 value implies 21.9× (P25–P75 15.4–31.3×) fiscal 2032 (Feb 2031–Jan 2032) earnings, the basis of our exit multiple; today: 18.1×. Also on fiscal 2032 (Feb 2031–Jan 2032) sales: 9.5× (P25–P75 6.3–13.6×); today: 9.7×.
Hurdle result: the median is above the hurdle ($440 vs $303). That is an annualised return of about +19% at the median (P25 +6%, P75 +33%).
Key assumptions. Our model uses 56 inputs in total, plus shared settings for the economy and the AI-spending cycle; the main ones are shown here.
- AI-spending cycle: about a 16% chance a year (10–24%) of a downturn lasting about five quarters, balanced by booms.
- AI-spending growth: about 30% in fiscal 2028, then about 14%, 7% and 5% in fiscal 2029–31 (the later cuts reflect Gartner’s revenue-gap warning).
- Nvidia’s share: up about 15 points in fiscal 2028, flat in fiscal 2029, then down about 1 point a year as AMD and others compete.
- Gross margin: 72–73% in fiscal 2028, as guided, drifting toward about 69% by fiscal 2031, with an extra 1–8 point write-down in a downturn.
- Exit multiple: Nvidia’s own past P/E relative to the S&P 500 (80%, with 20% peers, no fade toward peers), moved by the market’s multiple and by each simulation’s growth versus today’s expectations.
The most influential events:
- An AI-spending downturn (the cycle above): the largest driver of the range.
- A hyperscaler moves part of its inference to its own chips: a 25% chance, from fiscal 2029, costing 1–6 points of share and up to 3 points of gross margin.
- A Taiwan/TSMC disruption: cuts revenue by 20–70% when it hits and lowers the exit multiple by up to 15%.
- China data-centre sales reopen: a 20% chance, adding 2–7% to revenue.
- Tighter US export rules: a 20% chance, taking up to 6% off revenue.
Sensitivities:
| Value per share at Dec 31, 2030 (undiscounted) | P10 | P25 | P50 | P75 | P90 | Mean | P(beat 9%/yr) | P(loss) |
|---|---|---|---|---|---|---|---|---|
| With a neutral economy (no recessions or sector cycles) | $255 | $350 | $500 | $695 | $880 | $540 | 83% | 4% |
| Without the Taiwan/TSMC tail event | $180 | $282 | $450 | $715 | $1,045 | $550 | 72% | 14% |
| Excluding retained cash | $150 | $243 | $410 | $666 | $989 | $505 | 65% | 20% |
How our simulation works, and the median path
In short: our simulation puts the median value at Dec 31, 2030 at about $440 a share against a $303 hurdle, with a wide range ($272–$700 for the middle half): 70% of simulations beat the hurdle and 16% end below today’s price. Whether AI spending turns down before then is the single biggest driver.
How it works:
- It runs 50,000 versions of the quarters from now to the end of 2030.
- In each, the inputs (AI-spending growth, Nvidia’s share, margins, the exit multiple, a downturn’s timing, depth and length, and events such as China reopening or a Taiwan disruption) are drawn from ranges, each with a stated source.
- Its path to Nvidia’s fiscal 2029 (Feb 2028–Jan 2029) stays within 5% of guidance and consensus. The value at end-2030 uses the following year’s (2031) earnings.
In the median simulation:
- Revenue is about $945B in fiscal 2030 (Feb 2029–Jan 2030) and about $960B in fiscal 2031.
- EPS rises from about $20.43 in fiscal 2029 to about $21.10, as growth slows to single digits.
- The value rests on a multiple of about 20× (14–30× for the middle half) applied to those earnings, a little above today’s ~18×.
- Our view: our model trims AI-spending growth after the consensus years (to about 7% in fiscal 2030 and 5% in fiscal 2031), reflecting Gartner’s warning that AI revenue may fall far short of what the spending needs.
Confidence: the simulation is arithmetic, so confidence applies to the inputs: medium through 2027, low for 2029–31. The AI-spending cycle and exit-multiple settings are judgement calls in our model.
8. Management and governance
Nvidia is founder-led but not founder-controlled, and Huang’s 30-year record of anticipating platform shifts is exceptional. The gaps are a long-tenured board and a pay plan that does not reward capital discipline just as capital deployment has become central; in our model, that matters through how much free cash flow returns to shareholders rather than going into stakes and guarantees.
Insiders (Fact, Form 4s; Nov 2025–Oct 2026): no open-market purchases, and Huang sold only 150,000 shares. Worth watching: director Mark Stevens’ large sales after the Q2 report and Kress’s plan change ahead of earnings.
The facts behind this
- CEO (Fact). Jensen Huang, 63, co-founder and CEO since 1993, owns 870.6M shares (3.58%). FY2026 pay was $36.3M. CFO Colette Kress has been in the role since 2013.
- Credibility events (Fact). The SEC fined Nvidia $5.5M in 2022 over its disclosure of crypto-mining demand, and the related class action was certified in 2026. Blackwell’s 2024 design flaw was acknowledged. Other events include the shrinking of the OpenAI “$100B” plan and the partial pause of the revenue-share program (section 2). On the other side: consistent beats against guidance, and delivery on the $500B bookings claim.
- What pay rewards (Fact, proxy). Cash bonuses depend on revenue; single-year PSUs on non-GAAP operating income (defined to exclude stock pay); multi-year PSUs on three-year total shareholder return relative to the S&P 500. When H20 export controls hit, the FY2026 stretch goals were automatically lowered (revenue goal to $160B from $190B; operating-income goal to $96B from $120B). Results then far exceeded them, and maximum payouts were earned. Our view: nothing in the plan charges for the ~$96B equity portfolio, the commitments or the guarantees, or rewards return on capital or free cash flow per share.
- Board and voting (Fact). Single share class, annual elections, proxy access, and the right for holders of 15% to call a special meeting. Long tenures: Tench Coxe and Harvey Jones have served since 1993, Brooke Seawell since 1997, and Mark Stevens since 2008 (he also served from 1993 to 2006). Lead director Stephen Neal, 77, drew the most votes against at the June 2026 meeting (about 2.2B, roughly 13% of votes cast). Suzanne Nora Johnson joined in July 2026.
- The finance and sales teams were refreshed in 2026: a new chief accounting officer from Intel (May) and a new head of field operations from Microsoft (August).
Insider sales and trading plans in detail:
- Huang sold only 150,000 shares (October 2025, under his 2025 plan), and no new Huang plan was disclosed through Q2 FY2027. His other filings are estate-planning trust transfers.
- Director Mark Stevens sold about 6.9M shares (about $1.5B), including about $818M in September 2026 after the Q2 report. His filings are not marked as plan sales.
- Ajay Puri sold 1M shares ($183M) under a plan in January–March 2026, before his June retirement notice.
- Kress ended a 500,000-share plan on Apr 10, 2026, with no sales made under it, six weeks before the strong Q1 report. She adopted a new 375,439-share plan on Jun 16, ten weeks before the Q2 report and the 70% guide. The new plan covers only shares from future award vestings, net of tax withholding (10-Q Item 5).
- Timing like this is legal and not evidence of misuse, but governance analysts track plan terminations ahead of earnings.
- Director Tench Coxe’s March 2026 plan (8M shares) is solely for gifts to charitable donor-advised funds.
Confidence: high.
9. Signposts
The first test is Q3 results (Nov 17) and whether the FY2028 guide holds. The bigger tests for our model are hyperscalers’ 2027 capex budgets (the AI-spending cycle drives most of the range), the custom-chip ramp and any losses on customer financing.
Eight signposts: what on track and broken look like
| # | Signpost | On track | Broken | When |
|---|---|---|---|---|
| 1 | Q3 FY2027 results | Revenue ≥$108B; gross margin ~74%; FY2028 ~70% reaffirmed; DSO back toward the low 50s | Guide trimmed; receivables keep outgrowing revenue | Nov 17, 2026 |
| 2 | Q4 FY2027 and the FY2027 10-K | Gross margin bottoms at 71–72% as guided; disclosure of the revenue tied to financed customers | Margin below 71%; new commitment or guarantee growth with no disclosure | Late Feb 2027 |
| 3 | Hyperscaler 2027 capex budgets | Top five near the ~$1.3T Nvidia cites | Cuts or “digestion” language | Late Jan–Feb 2027 |
| 4 | Rubin ramp | Majority of Data Center revenue by mid-2027; Rubin Ultra on its 2027 schedule | Yield or supply slips | Q4 FY2027–Q2 FY2028 |
| 5 | Custom chips | Jalapeño and TPU deployments stay incremental | A major lab or cloud moves inference spending off Nvidia | Through 2027 |
| 6 | Financing exposure | OpenAI and Anthropic strengthen credit (IPOs, ratings); no guarantee calls | Neocloud defaults; guarantee or rent-back losses | Ongoing; SB Energy phases from FY2029 |
| 7 | China and export rules | H200 deliveries resume or SAMR closes the case without penalty | Penalties or tighter US rules (GAIN AI Act, Remote Access Security Act) | Open-ended |
| 8 | Hugging Face | Closes, platform stays open | Blocked or conditions imposed | 1H 2027 |
10. Open questions
The biggest unknown is how much FY2027–28 revenue goes to customers Nvidia finances (stakes, guarantees, rent-backs, extended terms); “roughly a quarter” of next year is the only figure given.
Six more open questions
- How are the AI-cloud rent-back payments accounted for: as costs, or as reductions in revenue?
- How much of the $279B of supply commitments is non-cancellable, and at what point would a slowdown trigger charges?
- Which customer moved from ACIE to Hyperscale in Q2, and which “frontier AI lab” gets about 2 GW of credit support?
- Do the FY2027 pay metrics now include stock pay, in line with the new non-GAAP definition?
- Was the revenue-share program changed after the reported pause?
- Long-range consensus: no FY2030 figures are public, and the FY2029 figures come from one or two aggregators.
Five-line summary
- Nvidia is the dominant AI-systems supplier. Revenue was $96B in Q2 FY2027 (+106%), with $108B guided for Q3 and about 70% growth guided for FY2028, limited by supply.
- Growth through FY2028 is underwritten by orders and $279B of supply contracts. Beyond that it depends on AI spending continuing to compound, and about a quarter of next year’s business comes from labs Nvidia helps finance.
- The core accounts are clean and the adjusted measures became more conservative. The flags are receivables and inventory outgrowing revenue, GAAP gains on stakes in customers, and about half of free cash flow going back into the ecosystem.
- At $210.96 (about 13× consensus EPS for fiscal 2028, Feb 2027–Jan 2028), the price implies earnings stop growing after 2028. Our simulation puts value at Dec 31, 2030 at $272 / $440 / $700 a share (P25 / P50 / P75; mean $535). The hurdle is $303 (the end-2030 price that would earn 9% a year from today’s price). 70% of simulations beat it and 16% end below today’s price. The AI-spending cycle is the main swing factor.
- Watch Q3 results (Nov 17), the gross-margin trough in Q4, hyperscalers’ 2027 capex budgets, and any disclosure or losses tied to the guarantees and customer financing.
Sources
Sources (accessed Oct 25, 2026): 19 filings, releases and data sources
- Nvidia 10-K FY2026 (Feb 25, 2026)
- 10-Qs Q1 and Q2 FY2027 (May 20, Aug 26, 2026)
- 8-K earnings releases and CFO commentaries Q4 FY2026–Q2 FY2027
- 8-Ks Jan 23, Mar 6, Apr 27, May 8, Jun 18 (notes offering), Jun 30, Jul 2, Aug 17 (SB Energy) and Sep 3, 2026 (Hugging Face)
- DEF 14A (May 12, 2026)
- 13F-HR (Aug 14, 2026)
- Forms 4, Nov 2025–Oct 2026 (SEC EDGAR)
- Q2 FY2027 earnings call transcript (FactSet CallStreet via Nvidia IR, Aug 26, 2026)
- StockAnalysis (NVDA statistics and forecast; AMD, AVGO, TSM, MRVL statistics, Oct 23 close)
- Yahoo Finance and MarketScreener (NVDA estimates)
- Dell’Oro Group (Aug 18, 2026)
- Barron’s, Invezz and Benzinga (Seaport, Oct 2026)
- CNBC (Aug 26, 2026; Mar 4, 2026; Feb 27, 2026)
- WSJ via The Star and DatacenterDynamics (revenue-share pause, Aug 28, 2026)
- SemiAnalysis via CNBC and The Decoder (Jalapeño)
- AMD newsroom (Advancing AI 2026)
- Morningstar Australia (Teh, Oct 21, 2026)
- Reuters via Investing.com (Anthropic IPO talks, Sep 2026)
- Gartner (Oct 2026)
Updates
- : Nov 12: Jensen Huang resigned as CEO; Colette Kress is interim CEO. We'll review this view after the Nov 17 results.
- : Nvidia's board said on Nov 8 that it is opening an independent investigation, overseen by a committee of independent directors with outside counsel, and that it has confidence in Huang and his leadership team. Nvidia shares closed down 4.3% at $178.67 on Nov 9. For our view: on what had been disclosed by then, we left our model's assumptions (section 7) unchanged. Section 8 links governance to how much free cash flow returns to shareholders rather than going into stakes and guarantees. The board said it would "provide further updates as appropriate"; Nvidia reports third-quarter results on Nov 17. Editors hold Nvidia shares. (Reworded Nov 12 to the board's statements.)
- : Corrected the valuation snapshot: cash and investments now include short-term marketable securities ($56.6 billion, previously $22.4 billion), so it shows net cash of $23.2 billion instead of net debt of $10.9 billion; enterprise value ($5.10 trillion, previously $5.14 trillion) and EV/sales (16.8×, previously 17.0×) updated. Prices and dates unchanged.
About Nvidia: Website (external site)Investor relations (external site)SEC filings (external site)Wikipedia (external site)
+ 6 more
Accessed Oct 25, 2026: Nvidia Q2 fiscal 2027 CFO commentary · Nvidia 8-K, Aug 17, 2026 (SB Energy guarantee) · Nvidia 8-K, Sep 2026 (Hugging Face)
Accessed Nov 9, 2026: Nvidia: statement from the board of directors (Nov 8, 2026) · Nvidia: conference call for Q3 fiscal 2027 results (Nov 17)
Accessed Nov 12, 2026: Nvidia 8-K, Nov 12, 2026 (CEO transition)
Quoting this? You may quote short excerpts under our reuse policy. Please credit WhyMyStockMoved.com and link to this page.