Nvidia's 'Three Reasons the Rally Continues' — A Bull Thesis Stress Test
A Yahoo Finance piece argues NVDA's monstrous rally has three legs left to run. The arguments are real, but each resolves at a different time horizon — and each carries a counter.
A Yahoo Finance / Motley Fool column from Keithen Drury this week argues that Nvidia’s monstrous rally still has three reasons to keep running. Drury walks through hyperscaler capex, the 2027 setup, and the forward multiple. Each argument is real. Each one is also worth reading slowly, because they resolve at three different time horizons and they don’t share the same risk profile.
The point of this note is not to render a verdict on Nvidia. We covered the May 20 fiscal Q1 print setup earlier today and have written up the late-April retail bull case and the intraday pullback that followed Meta’s capex raise. What’s worth doing here is taking the bull article’s three arguments one at a time and asking what each one actually proves, what time frame it commits to, and what would need to be true for it to fall apart. That is a different exercise from “is the headline right or wrong.” Headlines bundle. The job is to unbundle.
Bull Argument #1: 2026 Hyperscaler Capex Is “Locked In”
Drury’s first reason is straightforward. The four big hyperscalers have collectively guided to roughly $650 billion in 2026 capital expenditures, with Microsoft alone guided to about $190 billion and Meta’s 2026 range pushed to as much as $145 billion. Both of those numbers were raised, not lowered. That commitment is the proximate driver behind Nvidia’s own +77% year-over-year revenue guide for the May 20 fiscal Q1 2027 print.
This argument is the strongest of the three because it is the closest in time. A guidance number is harder to walk back the deeper you are into a fiscal year, and the customer concentration here is a feature, not a bug, when capex is rising. The setup into May 20 is genuinely de-risked at the consensus revenue line.
What this argument cannot prove: that the back half of 2026 capex actually flows the way the front half did. Hyperscalers retain the right to taper. A guided range is not a contractual commitment. The bull thesis on argument #1 is essentially a thesis about what the next two quarters look like, not what the next two years look like.
Bull Argument #2: 2027 Capex Is Set Up to Grow, Not Plateau
Drury’s second reason stretches the trade out. Alphabet has telegraphed that 2027 capex will increase meaningfully versus 2026. Nvidia management has framed the global data center buildout as a $3 trillion to $4 trillion cycle by 2030, as cited in the Drury column. The implication: the AI infrastructure cycle isn’t peaking in 2026 — the planning horizon stretches to the back end of the decade.
This is where the time-horizon problem starts to bite. A 2030 figure rests on a chain of assumptions: that hyperscaler revenue growth supports the capex, that no recession compresses the planning cycle, that custom silicon does not displace a meaningful share of GPU spend, and that the sovereign AI buildout (which Nvidia management framed in its Q4 FY26 release and accompanying commentary as more than $30 billion in fiscal 2026, more than triple the prior year) keeps compounding from a small base. None of those assumptions is implausible. All of them have to hold for the 2027–2030 leg of the thesis to work.
The counter-pressure is the multi-year ceiling, not the multi-quarter one. Hyperscaler custom silicon — Google’s TPU, Amazon’s Trainium, Meta’s MTIA, Microsoft’s Maia — is a 2028-and-beyond drag on Nvidia’s wallet share, not a 2026 one. We have walked the valuation read across the AI capex complex before, and we have noted Michael Burry’s reservations on the Microsoft side of the trade. Neither argument concludes the cycle is over. Both note that priced-to-perfection setups are most fragile when forward growth merely decelerates rather than reverses.
Bull Argument #3: The Forward Multiple Is No Longer Rich
The third argument is the cleanest in print and the most subtle in practice. Nvidia is trading near 24 times forward earnings versus the S&P 500 at roughly 21.8 times. For a company growing 70%-plus, a low-single-digit-multiple premium is, on its face, modest. The bear objection from 2024 and 2025 — “valuation is stretched” — has narrowed materially.
The catch is that “forward multiple” is doing a lot of work. The ~24x rests on the FY2027 EPS estimate, which sits near $8.34 against $4.77 of adjusted EPS for fiscal 2026. If the May 20 print arrives in line and the Q2 guide is firm, the multiple holds. If the Q2 guide softens by even a few percentage points, the FY27 number gets re-marked, and the same 24x multiple becomes a higher implied multiple before the price has had time to move. Forward P/E is a denominator argument, and the denominator is a forecast.
| Bull argument | Time horizon | What it proves | What it doesn’t |
|---|---|---|---|
| 2026 capex locked in | Near-term (next 2 quarters) | The May 20 setup is de-risked at the revenue line | Whether back-half 2026 capex executes |
| 2027+ capex growth | Multi-year (2027–2030) | The cycle isn’t peaking in 2026 | Whether custom silicon erodes wallet share |
| Forward multiple is reasonable | Spot valuation | Premium has narrowed vs. S&P 500 | Whether FY27 EPS holds if Q2 guides down |
What May 20 Actually Tests
The bull article’s three arguments do not all get adjudicated on the same day. May 20 tests argument #1 directly. It can support or undermine argument #3 depending on how the Q2 guide lands. It cannot resolve argument #2 at all — the multi-year capex cycle is a theory about hyperscaler behavior in 2027, 2028 and beyond, and a single quarter’s print does not settle that. Conflating the three arguments into a single “rally continues” framing collapses the time-horizon distinction that matters most for sizing.
In our view, that distinction is the entire reason headlines like “three reasons” tend to read so cleanly. Three discrete claims operating on three different clocks get bundled into one verdict. Investors looking at the print on May 20 will see one of those clocks tick. The other two keep running on their own schedules, and the question of whether the rally continues is really three separate questions wearing one coat.
What FC Is Watching
A few signals in the May 20 release matter more for unbundling these arguments than the headline beat does:
- Q2 FY27 guide vs. consensus. The single biggest determinant of whether argument #1 strengthens or argument #3 weakens. A guide that brackets consensus de-risks the next quarter. A guide a few points light triggers the FY27 EPS re-mark.
- Customer concentration commentary. The 10-K and any prepared color on hyperscaler custom silicon timing speak to argument #2. Generic optimism is not the same as a roadmap.
- Sovereign AI and physical AI line-item disclosure. The $30 billion-plus sovereign figure and the $6 billion physical AI line management cited in Q4 FY26 commentary need to keep compounding for the multi-year leg to look credible. These are the longest-duration legs of the bull case and the ones with the smallest base today.
None of this is a recommendation. NVDA’s place in any given portfolio depends on tax situation, concentration, time horizon, and a dozen factors that a blog post cannot see. The point is narrower: a “rally continues” headline is a compound claim, and compound claims are easier to evaluate when you separate them into the parts that report on May 20 from the parts that report in 2028.
Disclosures. Ferrante Capital LLC is a Registered Investment Adviser (RIA). This commentary is for informational and educational purposes only and does not constitute investment advice or a recommendation to buy or sell any security. NVDA is referenced as a research subject; Ferrante Capital, its principals, or related accounts may hold or transact in NVDA, broad-market index funds and ETFs, semiconductor sector ETFs, or related securities at any time. Past performance does not guarantee future results.
Forward-looking statements reflect Ferrante Capital’s current analysis based on publicly available data as of the publication date. Actual results may differ materially from any expectations expressed or implied. Earnings forecasts, analyst price targets, capex projections and rate-of-change estimates are inherently uncertain.
Please consult a qualified financial professional before making investment decisions.
