NVDA — NVIDIA Corporation

As of 2026-09-13, sentiment is overwhelmingly bullish on fundamentals: the 2026-08-26 Q2 FY27 print showed revenue up 106% YoY to $96.2B, Data Center at $89.0B (+117% YoY, ~93% of the mix), 75% gross margin, and $26B returned to shareholders in one quarter, with management guiding Q3 to $108B and framing demand as supply-constrained. The Street is nearly unanimous - roughly 95% Buy-rated, consensus target $341-345 (range $270-515), with J.P. Morgan at $320 and Piper Sandler initiating Overweight at $300. The one soft spot is price action: the stock is down 5.24% over the past week and 2.59% over the month despite the beat, which reads as a crowded, well-owned name digesting a big run (+21.1% over six months) rather than any deterioration in the story.

Accelerating, not decelerating, and by a wide margin. Revenue has climbed from $18.1B in the quarter ended 2023-10-29 to $96.2B in the quarter ended 2026-07-26 - roughly 5.3x in 11 quarters - with sequential growth of 10-19% every single quarter in that span. YoY revenue growth has inflected upward for four straight quarters: +62.5% (Oct-25), +73.2% (Jan-26), +85.2% (Apr-26), +105.9% (Jul-26). Margins are expanding at the same time: gross margin 72.4% -> 75.0% YoY, and operating margin 60.8% -> 66.2% YoY (and 63.2% -> 65.0% -> 65.6% -> 66.2% sequentially), with diluted EPS going $1.08 -> $2.46 YoY. Two honest caveats. First, the Q3 FY27 guide of $108B implies +89.5% YoY versus +105.9% this quarter - still enormous, but the growth rate's second derivative is about to turn. Second, Q1 FY27 net income ($58.3B) exceeded operating income ($53.5B) on non-operating gains, and Q2 FY27 operating cash flow ($24.1B) was less than half of net income ($59.7B) - a working-capital build, not a demand problem, but a real quality asterisk.

Cheap relative to its own history and its growth rate; merely fair relative to the broad market. As of 2026-09-13 the stock trades at 27.5x trailing EPS of $7.97, versus a five-year average P/E of 59.71x, and at 17.45x trailing sales versus a five-year average P/S of 21.8x - it is roughly half its historical earnings multiple on the P/E line and ~20% below on sales. The TTM PEG is 0.22, which is an artifact of hypergrowth, but even on my more conservative FY2027 EPS estimate of ~$10.40 the forward P/E is ~21x (23.6x on the sell-side's stale $9.255 next-FY number). The counterweight: 23.1x book, 22.8x EV/EBITDA, a 2.2 beta, and 97.2% of analyst targets sitting above the price - there is no absolute-value cushion here, only a growth-adjusted one. Verdict: cheap-to-fair for the growth and margin profile, but the multiple only looks 'cheap' because earnings have been compounding faster than the share price.

This is a multiple-compression story, not an earnings-disappointment story. Over the past six months the stock is up 21.1%, while quarterly EPS roughly doubled ($0.89/$1.08 in the Jan-26/Jul-25 quarters versus $2.39/$2.46 in Apr-26/Jul-26) - meaning the market has been actively de-rating the stock even as estimates rose. The mechanism is straightforward: at $5.3T market cap, the market is refusing to pay the old 60x multiple on numbers this large, and it is pricing in hyperscaler capex cyclicality, China export restrictions, and the risk that 66% operating margins are a peak rather than a floor. The practical read: if revenue and EPS keep beating while the multiple stays at ~21x forward, the stock grinds higher on earnings alone; if the multiple re-rated merely to 30x - still half its five-year average - you'd get ~$310 on FY2027 numbers. The thing that would break this is not a demand miss but a cash-conversion miss: TTM net income exceeded operating cash flow by roughly $58B, so watch the receivables line and the OCF-vs-net-income gap on the November print.

As of 2026-09-13, RSI(14) is 49.75 - dead neutral, neither overbought nor oversold - and the price of $218.29 sits 0.85% below its 20-day SMA (~$220), 2.68% above its 50-day SMA (~$212.6), and 10.54% above its 200-day SMA (~$197.5). That is an intact, long-dated uptrend that has stalled and gone range-bound over the short term: -5.24% over the past week, -2.59% over the month, versus +6.38% over three months and +21.1% over six. The stock is chopping between roughly $212 (50-day) and $236.54 (52-week high) with no oversold entry signal yet. Timing implication: this is a consolidation in an uptrend, so weakness toward the 50-day is a better entry than chasing into $230+; a clean break below $212 (and especially the 200-day near $197.5) would be the first real technical warning that the market is pricing in a capex peak.

• Hyperscaler capex cycle: essentially the entire thesis rests on a handful of cloud and frontier-lab buyers sustaining multi-hundred-billion-dollar annual AI infrastructure spend; any moderation in their capex guides hits NVDA first and hardest, and it is a demand pool NVDA does not control.
• US export controls on advanced chips to China: a structural cap on total addressable market plus recurring headline risk; any easing is unmodeled upside, any tightening is immediate downside.
• AI buildout financing and rates: a growing share of the buildout is debt- and lease-financed, and NVDA itself raised long-term debt from $7.5B to $32.4B in the quarter ended 2026-07-26 - tighter credit or higher-for-longer rates would slow orders and raise counterparty risk in the customer base.
• Supply chain capacity (HBM memory, advanced packaging/CoWoS, leading-edge foundry): demand is supply-constrained today, which supports pricing and margins, but a capacity catch-up can flip this market toward oversupply faster than most investors expect in semiconductors.
• Index/sector concentration and AI-sentiment beta: with a 2.22 beta and a $5.3T market cap, NVDA is the de facto bellwether of the entire AI trade - sector rotation or an 'AI bubble' narrative moves the stock independently of its own results.

• Q3 FY27 earnings in late November 2026: verification of the $108B revenue guide, the margin path, and - critically - whether the operating-cash-flow-to-net-income gap narrows as working capital normalizes.
• FY2028 guidance framing: management has pointed to roughly +70% revenue growth for FY2028 (ending Jan 2028); confirmation or upward revision would reset the entire earnings base higher.
• Next-gen platform ramp (Vera Rubin): volume production and customer deployments are expected to carry higher revenue per gigawatt than Blackwell, a structural mix and pricing tailwind if it lands on schedule.
• Potential up to $10B investment tied to Anthropic's IPO: a strategic/customer-financing move that both underwrites demand and signals NVDA is willing to fund its own ecosystem.
• China export policy changes in either direction: easing would add an unmodeled revenue stream; further restriction would remove one, and both would move the stock more than a normal quarter's noise.

• Customer concentration plus extended payment terms: a handful of hyperscalers drive the majority of Data Center revenue, accounts receivable have grown past $63B with extended terms for select investment-grade clients, and a capex pause would hit revenue, cash conversion, and counterparty credit simultaneously.
• Earnings quality / cash conversion: trailing four-quarter net income exceeded operating cash flow by roughly $58B (the single-quarter accruals ratio was 11.12% as of 2026-07-26), so reported EPS is running meaningfully ahead of cash generated - a red flag if working capital reverses.
• Multiple and cyclicality risk: a 2.22 beta means 30-40% drawdowns are a normal feature, not a bug, and the de-rating from 59.7x to 27.5x trailing earnings shows the market is already refusing to pay historical multiples - there is less valuation cushion than the 'cheap vs. its own history' label suggests.
• Competitive and architectural displacement: custom silicon (TPU, Trainium, MTIA and internal hyperscaler ASICs) plus maturing model-efficiency gains could erode share and pricing at the margin; the CUDA software moat is deep but not impregnable at these volumes.
• Concentration of supply and geopolitics: leading-edge fabrication and advanced packaging remain geographically concentrated, leaving the supply chain exposed to Taiwan-related disruption and tariff/export policy shifts.

This is the strongest fundamental growth story in large-cap equities right now, and it is not priced like one: 105.9% YoY revenue growth, 66.2% operating margins, 62.9% ROIC, a 4.59 current ratio, and a 27.5x trailing / ~21x forward earnings multiple against a five-year average P/E of 59.7x. The bull case is simply that earnings keep outrunning the multiple. The honest bear case is not demand - it is cash: net income is running roughly $58B ahead of operating cash flow on a trailing basis, receivables are ballooning with extended terms, and the whole edifice leans on a handful of hyperscaler capital budgets that have never been tested through a downturn. At $218.29 as of 2026-09-13, with RSI at 49.75 and the stock consolidating below its 52-week high of $236.54, I would be a buyer on weakness toward the 50-day (~$212) rather than a chaser, sized for a 2.2-beta stock that can lose a third of its value in a bad quarter without anything actually breaking in the business.

I applied a 30x forward P/E - roughly half NVDA's five-year average P/E of 59.7x and a ~0.6 PEG on forward EPS growth, a haircut I consider warranted for $5.3T of scale, hyperscaler capex cyclicality, and China export restrictions - to my own FY2027 (ending Jan-2027) EPS estimate of ~$10.40, built from H1 FY27 actual GAAP EPS of $4.85 plus ~$5.55 implied by the $108B Q3 guide at a ~62% net margin, which yields ~$312. A cross-check requiring a 2.5-3.0% normalized free-cash-flow yield on roughly $195B of FY2027 FCF brackets $270-325, so I settle at $310 - below the $341-345 Street consensus to reflect the cash-conversion and customer-concentration risks the sell-side is not discounting.