Khan Capitals | May 2026
Key Takeaways
- Record everything, muted reaction: Nvidia delivered $81.6 billion of fiscal Q1 revenue (up 85 percent year-on-year), $75.2 billion of data centre revenue and guided $91 billion for the current quarter, yet the shares slid roughly 1.5 percent in the after-hours session and again on Thursday’s open. This is the fourth straight quarter where a clear beat-and-raise has produced a negative tape.
- The bar has caught up to the company: Consensus was already running near $79.2 billion of revenue and the buy-side whisper considerably higher. With Nvidia now carrying a 7.17 percent weight in the S&P 500 and a $5.4 trillion market capitalisation, “beating estimates” no longer represents the marginal investor’s bar.
- Capital return signal is the real news: The board authorised an additional $80 billion of share repurchases and lifted the quarterly dividend from $0.01 to $0.25 per share, a 25-fold increase. Nvidia is signalling that internal reinvestment opportunities, while still enormous, no longer absorb the entirety of its cash generation.
- Index concentration risk is now the equity story: Roughly 20 percent of the S&P 500’s year-to-date return and a comparable share of 2026 earnings growth can be traced to Nvidia alone. The Magnificent Seven account for 55 percent of expected Q2 2026 earnings growth at the index level.
- The macro tape is not cooperating: The print landed alongside hawkish April FOMC minutes, 30-year Treasury yields above 5.19 percent and a fresh oil shock driven by Tehran’s hardening uranium position. AI capex growth is meeting tighter discount rates, and the multiple compression debate is back on the table.
A Record Quarter That Nobody Cheered
By any conventional metric, Nvidia’s fiscal first-quarter results, reported after the close on Wednesday 20 May, were extraordinary. Total revenue of $81.6 billion arrived 3.0 percent above the consensus figure of $79.2 billion. Data centre revenue, the segment that institutional investors have spent the past twenty-four months obsessing over, printed at $75.2 billion, up 92 percent year-on-year. Non-GAAP earnings per share of $1.87 cleared the $1.78 estimate. Gross margins of 75.0 percent on a non-GAAP basis came in slightly above guidance.
The forward number was, if anything, more striking. Management guided fiscal second-quarter revenue to $91.0 billion, plus or minus 2 percent, against a Street consensus of $86.84 billion. That is a roughly 4.8 percent positive guide on a base that is already the largest quarterly revenue figure ever reported by a semiconductor company. The implicit message: Blackwell 300 shipments are accelerating, hyperscaler capex is not rolling over and the gating factor remains supply rather than demand.
And yet, the shares fell roughly 1.3 percent in the after-hours session immediately following the release, with the weakness extending into Thursday’s regular trading. This marks the fourth consecutive quarter in which a clear beat-and-raise from Nvidia has resulted in a negative immediate price response. It is a sharp inversion of the pattern that defined the 2023 to 2025 period, when even in-line prints could deliver double-digit one-day moves to the upside.
Inside the Numbers
The internal composition of the data centre line is what professional investors will spend the next two weeks dissecting. Of the $75.2 billion in data centre revenue, hyperscale contributed $37.9 billion, up 115 percent year-on-year but only 12 percent quarter-on-quarter. The AI Clouds, Industrial and Enterprise sub-segment came in at $37.4 billion, up 74 percent year-on-year and a striking 31 percent quarter-on-quarter. Two implications follow. First, hyperscaler order patterns are beginning to lumpify: the linear acceleration of 2024 is giving way to step changes timed around platform transitions. Second, sovereign and enterprise demand is broadening, with Middle Eastern, Indian and European compute build-outs absorbing a meaningfully larger share of the mix than at any prior point.
The capital return announcement deserves separate emphasis. The board authorised an additional $80 billion of repurchase capacity without expiration, on top of an existing programme. During the quarter just reported, Nvidia returned approximately $20 billion to shareholders through repurchases and dividends. The dividend was raised from a token $0.01 quarterly rate to $0.25, a roughly 2,400 percent increase. At an annualised $1.00 per share against the current share price, the implied yield remains well under 1 percent, but the signal is more important than the level: this is no longer a company that requires every cent of internal cash flow to fund growth.
| Metric | Q1 FY27 (Apr 2026) | Q4 FY26 (Jan 2026) | Q1 FY26 (Apr 2025) | YoY change |
|---|---|---|---|---|
| Total revenue | $81.6bn | $68.0bn | $44.1bn | +85% |
| Data centre revenue | $75.2bn | $61.5bn | $39.1bn | +92% |
| Non-GAAP gross margin | 75.0% | 73.5% | 78.4% | -340 bps |
| Non-GAAP diluted EPS | $1.87 | $1.43 | $0.89 | +110% |
| Capital returned to shareholders | ~$20.0bn | n/a (Q4 lower base) | n/a (Q1 FY26 base) | record level |
Why the Stock Did Not Move
The simplest explanation is the right one: positioning. Between the February 2026 lows and the print on 20 May, Nvidia rallied roughly 14 percent, outpacing the S&P 500’s 6.9 percent advance over the same window. Implied volatility going into the print was elevated but term-structure analysis suggested options markets were pricing a one-day move of around 7 percent. The actual move of 1.3 percent down represented a meaningful disappointment for investors who had been long premium into the event.
The deeper explanation, articulated by GraniteShares founder William Rhind in the immediate aftermath, is that expectations have caught up to fundamentals: Nvidia is now “the bar” rather than the company that clears it. This is a structural change. For most of the post-ChatGPT era, the buy-side modelled Nvidia’s data centre business by taking sell-side numbers and adding 10 to 20 percent. That arbitrage closed several quarters ago. The whisper figure for Q1 was already in the $80 billion to $82 billion range, and the next-quarter whisper was running close to $90 billion. The official guide of $91 billion is a positive surprise on consensus, but only marginally on the figure that the marginal long had already underwritten.
There is also a margin point worth flagging. Non-GAAP gross margin of 75.0 percent is below the 78.4 percent recorded in fiscal Q1 FY26. The deceleration reflects the ramp of Blackwell 300 (higher initial costs, lower yields) and a richer mix of systems revenue, where Nvidia is increasingly selling complete reference designs rather than discrete GPUs. Management has guided that gross margins will rebuild towards the high 70s through fiscal 2027 as Blackwell yields mature, but the near-term print disappointed bulls who had hoped the margin trough was already behind.
The Macro Cross-Currents
Nvidia’s print did not arrive in a vacuum. The same afternoon, the Federal Reserve released the minutes of its 28-29 April meeting, which made clear that a hike, not a cut, is now the live alternative to a hold. Fed funds futures shifted to pricing roughly a 39 percent probability of a rate hike by year-end, against pricing of eight basis points of cuts as recently as late April. Thirty-year Treasury yields topped 5.19 percent in the immediate aftermath, the highest level since before the 2008 financial crisis. Ten-year yields touched 4.68 percent.
Higher long-end yields matter for Nvidia in two ways. Mechanically, they raise the discount rate applied to the company’s long-duration cash flows, compressing the multiple that any given earnings stream supports. Behaviourally, they alter the relative attractiveness of secular growth versus value: when an investor can clip 5.2 percent in a 30-year Treasury, the hurdle rate for owning a stock trading at 32 times forward earnings rises materially. Nvidia’s price-to-sales multiple, even after the recent slide, sits north of 16 times. That is not unusual for the company’s history, but it is a number that depends on a benign rate environment.
The macro picture darkened further on Thursday morning, when reports emerged that Iran’s supreme leader had issued a directive that the country’s enriched uranium stock must remain inside the Islamic Republic, hardening Tehran’s position in ongoing negotiations with Washington. WTI crude jumped close to 4 percent towards $102, with Brent advancing 3 percent above $108. Energy inflation feeds back into the same hawkish Fed setup that is weighing on the discount rate.
What the Market Is Underappreciating
Three points stand out. First, the rate of change in Nvidia’s hyperscale segment is decelerating even as the absolute level continues to rise. Sequential growth of 12 percent in hyperscale revenue, against 31 percent in the AI Clouds, Industrial and Enterprise sub-segment, reveals that the hyperscaler order pipeline is becoming more lumpy and harder to forecast on a linear basis. This is not a demand problem: it is a customer-concentration problem moving in the right direction, as Nvidia’s customer base broadens away from the four or five hyperscalers that dominated the early 2024 ramp. The implication is that quarter-to-quarter variance is likely to rise, even as the trajectory remains upward.
Second, the dividend reset is more than a token gesture. Lifting the payout 25-fold and authorising an additional $80 billion of buybacks reframes Nvidia as a capital-return story alongside its growth narrative. At current free cash flow generation of roughly $90 billion annualised, Nvidia could repurchase 1.5 percent of its float per quarter at present prices while still funding its capex and R&D programmes. The board’s willingness to commit to that level of capital return suggests internal modelling of demand sufficient to sustain margins, but also a recognition that opportunities to deploy incremental capital at returns above the cost of capital are no longer infinite.
Third, the index concentration risk is rising even as the company-specific risk is, arguably, falling. The Magnificent Seven now account for roughly a third of S&P 500 market capitalisation and 55 percent of expected Q2 2026 earnings growth at the index level. Nvidia alone contributes around 20 percent of the index’s year-to-date return. Investors holding broad equity exposure are, whether they realise it or not, holding a substantial directional position in semiconductors and on the continued viability of the AI capex cycle. A mean-reversion event in Nvidia’s multiple would have first-order consequences for passive-vehicle holders who never explicitly underwrote single-stock risk.
The Capex Question
The largest question that this print does not answer is whether the hyperscalers’ aggregate capital expenditure path, which Khan Capitals has previously sized at $725 billion through 2027, is sustainable. The data centre numbers Nvidia reported are a direct read on what Microsoft, Alphabet, Meta, Amazon, Oracle and the sovereign cloud build-outs are spending today. The Q2 guide implies acceleration. But sell-side models still expect the hyperscaler capex line to flatten in late fiscal 2027 as the initial Blackwell deployments are absorbed, and there is no visibility yet into the demand picture for what comes after.
Two lead indicators bear watching over the next two earnings cycles. First, the proportion of hyperscaler capex committed to GPU purchases as opposed to data centre construction, power and networking. As compute density rises and Nvidia’s systems revenue mix grows, this ratio is shifting in Nvidia’s favour, but the headline capex number can mislead. Second, the depreciation life that hyperscalers apply to AI infrastructure. Microsoft and Meta both moved their assumed lives from four years to six years in 2024; any rollback would dramatically affect free cash flow and the appetite for new spend.
Investor Implications
Equities
The base case for institutional equity positioning is that Nvidia’s earnings trajectory remains the cleanest expression of the AI build-out, but the marginal return on additional dollars allocated to the single name is now meaningfully lower than it was twelve or eighteen months ago. Cross-sectional positioning may benefit from rotating part of the Nvidia exposure into adjacent beneficiaries (advanced packaging, high-bandwidth memory, optical interconnects, liquid cooling, power infrastructure) where consensus models still embed a less mature multiple. Within the Magnificent Seven cohort itself, the divergence between Nvidia’s data centre engine and the weaker fundamental trajectories at Apple and Tesla, examined separately in Apple’s post-Cook transition and Tesla’s Q1 capex surprise, argues for active expression rather than passive cap-weighted exposure.
Fixed income
The same hawkish minutes and supply-driven yield rise that pressured Nvidia’s multiple have a direct implication for fixed income positioning. With 30-year Treasuries above 5.19 percent, the term-premium component of long-end yields has rebuilt to levels not seen since the pre-financial-crisis period. Investors may wish to consider that this is a regime where curve steepeners can carry positively while expressing a view on persistent inflation pressure, and where credit spreads, which remain compressed relative to historical norms at this stage of a hiking cycle, look poorly compensated for the macro tail. The investment-grade corporate index has not yet absorbed the implications of a Fed funds rate that could be higher by year-end than it is today.
Cross-asset positioning
The most underappreciated cross-asset implication is the convergence between the AI capex cycle and the energy complex. Each incremental gigawatt of AI compute requires firm baseload power, and the supply response from nuclear, natural gas turbines and grid-connected storage is not keeping pace. The recent move in Brent above $108, while geopolitically driven in the short term, sits on top of a structurally tighter electricity market in the United States and Europe. Allocators considering inflation hedges may find that the strongest expression sits at the intersection of energy infrastructure and the power generation supply chain, rather than in traditional commodity beta. Within commodity beta itself, gold’s continued advance, examined in our 2026 geopolitical portfolio framework, reflects the same structural concerns about fiscal sustainability that are now visible in the long end of the curve.
Conclusion
Nvidia’s fiscal first-quarter results are, on the numbers, the strongest set of figures any semiconductor company has ever reported. Revenue growth of 85 percent on a base of $44 billion, gross margins above 75 percent and a $91 billion forward guide represent operating performance that, on a historical analogue, would have produced double-digit one-day gains. The fact that the print produced a price decline instead is the more important signal. It tells us that the marginal long is no longer pricing surprise: they are pricing perfection, and partial perfection now disappoints.
For Khan Capitals’ framing, the print marks the moment when Nvidia transitions from a single-name growth thesis to an index-level concentration risk. The capital-return announcement, the dividend reset, the broadening of the customer mix and the maturing margin profile all argue that the company itself is becoming more durable. But the same factors that make it more durable make it less explosive, and a market that has spent two years pricing explosive returns must now adjust. The transition will not be smooth; the broader market, where Nvidia’s weight is now structural, will feel the adjustment alongside it.
Sources: Nvidia Form 8-K press release, fiscal Q1 FY27; Nvidia CFO commentary, fiscal Q1 FY27; CNBC, Nvidia earnings takeaways, 20 May 2026; Reuters via Investing.com, Nvidia $80 billion buyback announcement; Al Jazeera, Nvidia record profit report, 21 May 2026; US News, FOMC April minutes summary, 20 May 2026; MacroMicro, Magnificent Seven share of S&P 500 market cap; Benzinga, Mag 7 share of S&P 500 earnings growth.
Related Reading: this analysis sits alongside our earlier examination of the $725 billion hyperscaler capex split, which laid out the underlying spending cycle that Nvidia’s data centre line measures; the Tesla Q1 capex surprise, which began the Magnificent Seven earnings reckoning; the PPI shock that put records above 7,400 against a 39 percent hike bet, which captured the macro setup into the Nvidia print; the quiet Russell 2000 leadership beneath the all-time highs, which speaks to the index concentration question; and our analysis of Apple after Cook, which examines the divergent fundamental trajectories within the Magnificent Seven cohort. For the next leg of this story, see our analysis of Dell’s $51.3 billion AI server backlog. The week when this dynamic finally bit, with more than $1 trillion erased from the chip complex in a session, is examined in The $1 Trillion AI Semiconductor Selloff. For the strategic backdrop to Apple’s position, see our analysis of the Apple Siri Gemini deal.
Leave a Reply