NVIDIA reported its July quarter (FQ2 FY2027) on 26 August. Revenue of $96.2bn rose 106% y/y and 18% q/q, about $4bn above the $91.7bn consensus and a wider beat than the company's habitual ~$2bn. Data Center revenue of $89.0bn grew 117% y/y; a newly disclosed split puts hyperscalers at roughly 55% of the mix, with the balance in neoclouds and enterprise, which grew 25% q/q. Gross margin of 75.0% was in line and non-GAAP EPS of $2.22 beat the $2.07–2.10 consensus. October-quarter guidance of $108bn at the midpoint (+90% y/y) came in some $5bn ahead of the Street, with gross margin guided down to 74.0% against 74.6% expected. Two things dominated the call: management for the first time explicitly guided to roughly 70% revenue growth for the fiscal year ending January 2028, against consensus nearer 40–44%, stressing that even this is limited by supply rather than demand; and supply commitments for FY2028 and beyond more than doubled from $119bn to $279bn, overwhelmingly memory.

Studying the market estimates going forward, there is a wider divergence on the shape of the cycle: one path carries growth uninterrupted through FY2029 to roughly $970bn of revenue; another has revenue peaking near $985bn in FY2029 and then declining to about $935bn and $840bn in FY2030–31, with EPS falling from $22 back toward $18.50. That is the honest range — an unbroken compounding story on one side, a very large and very profitable cyclical peak arriving in roughly three years on the other.

---

NVDA stock chart by TradingView

---

Valuation Dynamics & Market Reaction

The stock closed at $227.98 on 27 August, a market capitalisation of about $5.5trn. Against next full-year (FY2028) EPS that is 17.4x on consensus, and 13–15x on the more bullish estimates of $15.01–17.16. Average EPS growth over the next two years is roughly 33% on consensus and 54–60% on those higher numbers, implying a PEG of 0.3–0.5x against a peer median near 1.0x. Revenue is expected to grow 88–95% this fiscal year, 70–77% next and 32–41% in FY2029, with net income roughly doubling to over $400bn by FY2028.

Against the AI-exposed semis peer group on a consistent consensus basis, NVIDIA trades at 16–17x next-year earnings versus a peer weighted average of 15x and a median of 18x — Broadcom at 18x, Qualcomm 16x, AMD 31x, Marvell 39x — and at 13–14x the following year against a peer average of 11–14x. The fastest-growing, highest-margin name in the group carries an ordinary group multiple. Most telling is what the most of the sell side analysts did: estimates for FY2028 were raised by double digits while target multiples were cut or held flat, on the explicit reasoning that the sheer size of the revenue base leaves less room for upward revisions. The resulting targets contain essentially all earnings growth and no re-rating.

---

Portfolio Positioning & Thesis

The market reaction was positive — up 7.6% after hours and closing 8.7% higher on 27 August, still just below the 52-week high near $236. That was earned: the explicit 70% guidance was genuinely new information and the margin reset removed a live overhang rather than creating one. We agree with the direction of the move but would not read it as the start of a re-rating.

Our view is that the revenue and net income path for the next two to three years is now well understood and broadly agreed; what the market is pricing is what comes afterwards — margin erosion, or share loss to custom silicon and recovering competition — and it expresses that by assigning a progressively lower multiple to a rapidly growing earnings base. The consequence is that the stock should compound more slowly than its earnings. That can still be a respectable return — 50%-plus earnings growth against a multiple drifting from 17x toward the low teens leaves meaningful upside — but it is a different proposition from owning a compounder that re-rates. We hold a small position and are not adding on these results.