Nvidia's business is growing even faster than the trailing figures commonly used to size it suggest, because those figures do not yet include the most recent quarter, which alone brought in more than 80 billion dollars, up 85 percent from a year earlier.
The company keeps roughly 75 cents of gross profit on every sales dollar and turned around 60 cents of every sales dollar into spare cash last quarter, levels almost no company this large has ever sustained.
At the same time, today's price already assumes that pace continues for years without a serious stumble, while big customers like Google, Amazon, and Microsoft are building more of their own chips, sales into China remain mostly blocked even after government approval, and some of the huge deals fueling growth involve Nvidia financing or guaranteeing purchases from the very companies buying its chips, which has drawn public criticism.
Weighing the clear strength of the business against those real risks, the shares look fairly valued today, not a bargain and not obviously stretched.
A quick look at how a few well known investing styles might weigh this stock.
Investors who follow Warren Buffett's approach would notice that Nvidia converts an unusually large share of every sales dollar into cash, with margins near 70 percent and returns on capital few companies ever reach, the kind of moat Buffett has always prized. What would give this style of investor pause is that the business depends on a handful of huge customers and a technology cycle that moves far faster than almost anything Buffett has owned, which cuts against his preference for simple, predictable businesses held for decades.
A Peter Lynch style investor would appreciate that Nvidia's story is easy to explain in one sentence, it sells the chips that make AI work, and that revenue has grown roughly eightfold in three years on real, reported demand rather than hype alone. Lynch also liked checking a stock's growth rate against its price tag, and here he would want to see growth keep outrunning an already very large market value before calling the shares cheap.
Investors who follow Cathie Wood's approach would frame Nvidia as a picks and shovels play on a technology shift they see reshaping the whole economy, pointing to the newer networking and national government business lines as evidence the growth runway extends well beyond chips alone. This style of investor tends to look past near term stock swings and current profitability in favor of how large the eventual market could be, betting the AI buildout is still early.
A Stanley Druckenmiller style investor would treat Nvidia less as a company to hold forever and more as the clearest way to express a view on the AI spending cycle, sizing a position around how much cloud companies say they will spend and stepping back at the first sign that spending is slowing. This approach stays comfortable moving in and out as the trend, or the broader interest rate backdrop that affects how investors value fast growing companies, shifts.
Investors who follow Philip Fisher's scuttlebutt approach would want to talk to customers and engineers about why they keep buying Nvidia's chips instead of cheaper alternatives, pointing to the software tools built around the chips as the kind of switching cost that keeps buyers loyal for years. This style of investor would also watch closely whether big customers building their own AI chips gradually wear down that loyalty over time.
These takes are illustrations of well known investing styles applied to this stock. They are informed guesses, not the actual views, statements, or holdings of the people named.
Nvidia's central claim, that demand for its AI chips is enormous and still growing, holds up well against independently reported numbers: both data center revenue and its newer networking business have grown at rates that back up management's own targets. Two narrower claims look weaker under scrutiny: a clean return to selling into China has not materialized despite government approval, and the company's market position, while still dominant, is being tested harder than bulls suggest by its own biggest customers building their own chips.
Reported results since that claim have tracked it: data center revenue reached about 39 billion dollars in the quarter ended April 2026, up 69 percent from a year earlier, and total company revenue hit 81.6 billion dollars, up 85 percent. Independent Wall Street consensus estimates for 2027 and 2028 combined revenue, compiled separately from the company, land in the same broad range as management's target.
The company's own reported networking revenue reached 31 billion dollars for fiscal year 2026, up 267 percent from the prior year. Independent research firm IDC corroborated this from the buyer's side, finding Nvidia became the largest data center Ethernet switch vendor by revenue in the first quarter of 2026 as the overall market it competes in grew about 40 percent.
Even after Washington approved H200 exports to China in December 2025, Chinese customs blocked the shipments from entering the country the following month, and Nvidia's own finance chief said in February 2026 that the company had booked zero revenue from the approved China shipments.
Independent supply chain researchers tracking the industry project custom AI chips built by Nvidia's largest customers, including Google's TPU, Amazon's Trainium, and Microsoft's Maia, will grow shipments around 45 percent in 2026, nearly three times the growth rate expected for chips sold on the open market like Nvidia's, concentrated in the inference workloads that now make up roughly two thirds of AI computing.
Based on public evidence such as filings, customer announcements, and delivery records.
About seven of every eight dollars in revenue now comes from selling AI computing chips (the current lineup is called Blackwell) and the switches and cables that connect thousands of them together inside data centers, sold mainly to major cloud providers, AI companies, and increasingly national governments building their own AI infrastructure. The rest comes from graphics cards for video game PCs, chips for design and media professionals, and computing platforms for cars.
We'd revisit this call if the price falls below $120.00 or rises above $300.00.
These are review triggers for us to take another look, not price targets or a promise the stock will reach them.
| Metric | NVDA (subject) | Peer median |
|---|---|---|
| EV/Sales | 24.29x | 22.12x |
| EV/EBIT | 40.23x | 67.27x |
| EV/EBITDA | 39.37x | 62.22x |
| P/B | 33.36x | 14.45x |
| Gross Margin | 71.1% | 51.0% |
| ROIC | 71.3% | 7.2% |
| ROE | 76.3% | 15.8% |
| Rule of 40 | N.M. | 21.3% |
Nvidia is the dominant supplier of AI accelerator chips (the Blackwell family) and the networking gear that links them into data center clusters, alongside a smaller gaming GPU business. Fiscal 2026 (ended January 25, 2026) revenue was 215.938 billion dollars, up 65 percent year over year (10-K/8-K). The most recent quarter, Q1 FY2027 (ended April 26, 2026), added 81.615 billion dollars, up 85 percent year over year, with Data Center revenue of 75.2 billion dollars, up 92 percent (NVIDIA 8-K, 2026-05-28). Trailing twelve months through April 26, 2026 (FY2026 total less Q1 FY2026 plus Q1 FY2027): revenue of about 253.491 billion dollars. The scoring engine's revenue_ttm input of 215.938 billion dollars is actually FY2026's fiscal year total, not yet rolled forward to include the Q1 FY2027 print filed in May, understating current trailing revenue by roughly 17 percent, a mismatch that is material even though it falls under the 20 percent threshold for a hard data flag. Correcting it lowers every revenue based multiple used in the comparison, a bullish, not bearish, adjustment to the picture.
GAAP gross margin was 71.1 percent for FY2026 and 74.9 percent in Q1 FY2027. Q1 FY2027 GAAP net income was 58.321 billion dollars on 81.615 billion dollars of revenue, a 71 percent net margin. Data Center is now roughly 92 percent of revenue; Gaming was 16.0 billion dollars for FY2026, up 41 percent, a fraction of Data Center's 193.7 billion dollars.
FY2026 operating cash flow was 102.718 billion dollars and free cash flow 96.575 billion dollars. Q1 FY2027 alone generated 50.344 billion dollars of operating cash flow and 48.554 billion dollars of free cash flow against only 1.757 billion dollars of capital spending, a free cash flow margin near 60 percent. Capital spending this light is unlikely to persist as the Rubin ramp and the company's own compute investments continue.
As of April 26, 2026: cash and equivalents 13.237 billion dollars, short-term debt 1.0 billion dollars, long-term debt 7.47 billion dollars (total debt 8.47 billion dollars), a light balance sheet relative to a company generating over 100 billion dollars of annual free cash flow.
Data center revenue growth: 92 percent year over year in Q1 FY2027 versus 68 percent for FY2026, an acceleration. Gross margin: 74.9 percent versus a peer group (Broadcom, AMD, Intel, Texas Instruments, Micron, Marvell, Analog Devices) median near 51 percent; Nvidia sits well above every listed peer. Free cash flow margin: about 60 percent in Q1 FY2027 versus roughly 45 percent for FY2026 as a whole, elevated this quarter by unusually low capital spending. AI accelerator market share: independent estimates put Nvidia near 70 to 75 percent of data center AI accelerator revenue, with hyperscaler custom chips (Google TPU, AWS Trainium, Microsoft Maia) near 15 to 20 percent and growing roughly 45 percent in 2026, concentrated in inference workloads that now make up about two thirds of AI compute.
Per the scoring engine's peer set (Broadcom, Micron, AMD, Intel, Texas Instruments, Marvell, Analog Devices), Nvidia's own multiple (about 39x EV/EBITDA) sits below the group's stated median (about 62x), though that median looks skewed high by outliers in a small peer set. Broadcom, the closest AI-exposed comparable, trades near 42 to 45x EV/EBITDA (Gurufocus, 2026-08-19), well below the 62x figure. Anchored to Broadcom specifically, Nvidia is not obviously expensive against its nearest AI peer.
Elevated rates (10-year Treasury near 4.65 percent, FRED) raise the discount rate applied to a long-duration growth story, and Nvidia's stock beta of roughly 2.2 means it moves harder than the market in both directions as rate expectations shift. Hyperscaler capital spending, the main demand driver, has been rising sharply; any pullback in 2027 capex guidance from Microsoft, Google, Amazon, or Meta would flow straight through to Nvidia's order book.
CEO Jensen Huang has sold more than 2.9 billion dollars of stock since mid-2024 under pre-set 10b5-1 trading plans, still holding roughly 860 million shares (about 3.5 percent of shares out, worth over 150 billion dollars). Aggregate insider sales across executives and directors exceed 3.3 billion dollars since early 2024. All disclosed sales follow pre-arranged plans, a normal pattern for an executive team this deep into a multi-year run-up, not clear evidence of a change in view.
Near-term catalyst: Q2 FY2027 results due August 26, 2026, against guidance of 91.0 billion dollars, plus or minus 2 percent. Risks: a Commerce official told Congress in July 2026 that H200 shipments into China remain "very few" even with licenses granted; custom AI chips from Nvidia's largest customers are growing faster than Nvidia's own addressable market; and a set of large financing arrangements with OpenAI and others (up to 250 billion dollars in disclosed lease guarantee talks and 350 billion dollars in chip purchase talks in 2026) has drawn public criticism questioning how much of the reported demand is independently funded.
The bull case would be falsified if Q2 FY2027 data center growth decelerates well below the 92 percent just reported, if hyperscaler 2027 capital spending guidance (typically given on Q4 2026 earnings calls) flattens, or if China shipments stay stuck at token volumes despite approvals. Disconfirming facts already on record: China H200 shipments remain minimal months after approval; custom silicon from Google, Amazon, and Microsoft is growing faster than the merchant GPU market; TSMC wafer and packaging capacity may cap Rubin shipments near 200,000 to 300,000 units in 2026 regardless of demand; and the scale of vendor financing style deals has drawn scrutiny from independent press.
Revisit this analysis if the stock trades outside roughly 115 to 230 dollars, if Q2 FY2027 data center growth (reported August 26, 2026) comes in below about 50 percent year over year, or if China shipments remain effectively zero through year end despite continued approvals.
Mechanical model output, a screen, not a fair value.
Applying a range of price-to-free-cash-flow multiples (15x bear, 27x base, 38x bull) to Q1 FY2027's annualized free cash flow run rate (48.554 billion times four) lands a base case almost exactly at the current price, with the bull case matching the sell-side consensus target range; this is a scenario read on value versus price, not a price target.
The mechanical screen and the research view reach the same verdict.
Confidence = 0.7 × research sub-score + 0.3 × mechanical sub-score. Mechanical sub-score rewards the Layer 1 screen for agreeing with Layer 2's direction (corroboration), not just for being conviction=NORMAL in isolation.