Micron vs Nvidia Stock: $1,540 vs $330, Which Returns More?

Micron vs Nvidia is usually framed as a supplier riding its biggest customer’s coattails, with the memory maker as the cheaper, cruder way to own the GPU boom. The latest filings reverse that picture. In its May quarter Micron (MU) booked an 84.6% gross margin, above Nvidia’s (NVDA) 75.0% in the quarter to July, and generated $18.3bn of adjusted free cash flow against Nvidia’s $21.3bn, from a company worth about one-fifth as much (Micron fiscal Q3 2026 release; Nvidia Q2 FY2027 release). So the question of which stock offers higher returns is really a question of durability. The market pays roughly 24 times run-rate earnings for Nvidia and about 7.5 times for Micron because it has watched memory stocks collapse three times in eight years. Our 12-month scenarios give Micron the bigger upside case and Nvidia the better risk-adjusted one.

The information that settles the debate sits in Nvidia’s own CFO commentary. Colette Kress disclosed that Nvidia’s supply commitments jumped from $119bn to $279bn in a single quarter, “primarily related to the procurement of memory” (Nvidia CFO commentary, 26 Aug 2026). Nvidia is paying up front to secure the component that is scarcest, and scarcity rent flows to whoever owns the scarce input. That is why the supplier now out-earns the platform on every margin line. Micron’s new multi-year Strategic Customer Agreements carried only about $5bn of remaining performance obligations at 28 May, which is a tenth of the $50bn Micron expects to bill in its current quarter alone (Micron 10-Q). Until that figure grows, investors are right to price Micron as a cyclical.

What is actually happening: the margin stack has flipped

For most of the AI build-out, the rule of thumb was simple. Nvidia captured the value, and memory makers sold a component into Nvidia’s bill of materials at whatever price the cycle allowed. The last three reporting periods broke that rule. Micron’s fiscal Q3, which ended 28 May, delivered $41.46bn of revenue against $9.30bn a year earlier. GAAP operating margin reached 80.4%. Every one of its four business units posted gross margins between 79% and 87%, including the Mobile and Client unit that sold into consumer devices at 24% a year ago.

Nvidia’s quarter to 26 July was enormous in absolute terms: $96.2bn of revenue, $89.0bn of it from the data centre, and a 66.2% GAAP operating margin. Nvidia’s gross margin rose 2.6 points year on year and is guided down a point for Q3. Micron’s rose 46.9 points and is guided up another 1.4 points to about 86%.

The useful analogy is a gold rush in which the shovel maker’s forge runs short of iron. The miners, here the hyperscalers, AI labs and Nvidia, still earn well, but the iron merchant now names the price. High-bandwidth memory (HBM), the stacked DRAM that sits next to every AI accelerator, is the iron. Nvidia’s Q2 GAAP net income of $59.7bn also flatters its position: it includes $7.8bn of other income, largely gains on equity stakes, which is why GAAP EPS of $2.46 sits above non-GAAP EPS of $2.22.

The customer relationship is less direct than the popular “Micron is an Nvidia proxy” story assumes. Micron’s 10-Q does not mention Nvidia by name once. It says HBM4 is in “high-volume shipments for our lead customer’s platform” and that a single customer accounted for 10% of revenue in the first nine months of fiscal 2026, down from 16% a year earlier. Nvidia’s Q2 release, meanwhile, names a multi-year technology partnership with SK hynix, not Micron. Having tracked Micron through the 2018, 2022 and 2024 downturns, I think the falling customer concentration matters more than any single design win: pricing power spread across a broad customer base is what lifted every unit’s margin at once. Sanjay Mehrotra, Chairman, President and CEO of Micron Technology, framed it the same way: “Micron’s record fiscal Q3 financial results and even stronger outlook for Q4 reflect the strategic value of memory in the AI era” (Micron, 24 Jun 2026).

Quick take: Micron is no longer the low-margin way to own AI. It is currently the higher-margin one. The debate is about how long that lasts, not whether it is real.

How each company is spending the windfall

Capital allocation shows how differently the two management teams read their own cycles. Micron is behaving like a company that remembers its last downturn. It repaid $9.38bn of debt in nine months, prepaying five note and loan tranches in full, which leaves $5.72bn of debt against $30.13bn of cash and investments. Buybacks were modest at $650m under the programme, and diluted share count still rose from 1,125m to 1,145m year on year. Net capital expenditure is guided at about $27bn for fiscal 2026. The first new Boise, Idaho fab is due to produce its first DRAM wafers in mid-calendar 2027, and the first New York fab, where ground was broken in January, supplies from 2030.

Micron’s structural answer to the cycle is the Strategic Customer Agreements. The 10-Q describes “binding commitments for specific volumes over the multi-year contract terms” that “often include substantial customer deposits.” Contract liabilities tied to them were $422m at 28 May, and agreements signed after that date are not yet in the numbers. Our standalone Micron stock prediction treats the next disclosure of those deposits as the single most important line in the 30 September report.

Nvidia is behaving like a company that expects the build-out to keep compounding and is using its balance sheet to make sure it does. It repurchased 203m shares for $39.8bn in the first half, with $99.3bn of authorisation left, and issued about $24.9bn of new debt in the same period. It has provided guarantees of up to $3.5bn on AI cloud partners’ data centre leases. On 17 August it signed residual value guarantees capped at $105bn on SB Energy’s PORTS-Pike campus in Ohio, where OpenAI is the tenant (Nvidia 8-K, 17 Aug 2026). On 2 September it agreed to acquire Hugging Face for about $11.9bn plus a retention pool of up to $1.0bn (Nvidia 8-K, 2 Sep 2026). Receivables rose from $38.5bn in January to $63.1bn in July, which Nvidia attributes to “extended payment terms on large multi-quarter agreements with certain investment-grade customers.” That is why free cash flow fell to $21.3bn in Q2 from $48.6bn in Q1 while revenue grew 18%.

Micron has its own receivables build, from $9.27bn last August to $31.03bn in May, so neither company is collecting cash as fast as it books sales. The difference is who is extending credit to whom. Nvidia is increasingly financing its customers’ ability to pay for its chips. Jensen Huang, founder and CEO of Nvidia, is unambiguous about why: “The AI infrastructure buildout is at full steam. Vera Rubin, now in full production, was built to power exactly this moment” (Nvidia, 26 Aug 2026).

Quick take: Micron is de-risking its balance sheet into the peak. Nvidia is levering its balance sheet to extend the peak.

Market data: returns, volatility and valuation side by side

The chart is the reason this comparison exists. Micron has returned 23 times as much as Nvidia over 12 months, and it did so with more than twice the volatility. It also fell 39.1% from its 25 June record close to its 29 July low inside the same year, against a worst peak-to-trough of 20.2% for Nvidia. Both stocks fell on 14 September. Micron closed down 5.25% and Nvidia 3.36%, while the VanEck Semiconductor ETF (SMH) fell 4.75% and the iShares Semiconductor ETF (SOXX) 5.63%. Tech stocks slid after leaders including Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman warned about the risks of rapid AI development (Motley Fool, 14 Sep 2026). The 10-year Treasury yield was at 5.03% early on 15 September, up from 4.96% at the prior close (CNBC).

On paper Micron is absurdly cheap. On Nvidia’s fiscal 2027 consensus EPS of $9.30, up 95% on fiscal 2026’s $4.77, Nvidia trades at 22.7 times earnings (stockanalysis.com). Micron trades at 7.5 times the run-rate implied by its own guidance. A price-to-earnings-growth (PEG) ratio makes both look like bargains, which is exactly why PEG breaks at a cyclical peak. The market is not mispricing Micron’s earnings. It is discounting their half-life. The 10-year record explains why:

Micron’s last three downturns averaged a 53.8% fall. Nvidia has suffered one deeper crash, in 2022, but its most recent drawdown was shallower. Social sentiment over the past 30 days leans the same way. One widely watched Micron commentary on YouTube asked whether anything had “actually changed” after the 14 September selloff (Market Signal, 15,107 views), and investors across the SK hynix and Samsung trade are watching the same supply question.

Quick take: Micron’s valuation discount is the price of volatility, not a mistake. It closes only if earnings prove stickier than in 2018, 2022 and 2024.

Regulatory and structural tension

The two stocks face different political exposures. Nvidia’s is export control. Its Q3 guidance assumes no data centre compute revenue from China. The 10-Q says US licences allow “small amounts of H200 products” to go to specific Chinese customers, but “such sales were restricted by the PRC government,” which cost a $0.4bn charge in the first half. Revenue from customers headquartered outside the US was 38% of Q2 revenue, so any further tightening lands on a large base. The Hugging Face deal adds a new layer. Nvidia’s own risk factor warns that restrictions on open-source models, many of which “originated in China,” could have “a material impact” on the platform it is acquiring.

Micron’s exposure is industrial policy. It signed CHIPS Act direct funding agreements for up to $6.1bn for fabs in Idaho and New York, collects a 35% investment tax credit on qualified US manufacturing, and holds a non-binding term sheet for up to $5.5bn from New York State. Those subsidies come with conditions and politics. A January 2026 petition in the Supreme Court of New York challenges the environmental review of the Clay, New York site. Micron also faces Netlist patent suits alleging that its HBM products infringe, now moved to the District of Delaware.

The deeper tension is that subsidies are designed to add supply, and supply is what ends memory cycles. Washington is paying Micron to build the capacity that, by 2027 and beyond, could compress the pricing that produced an 84.6% margin. The 14 September selloff added a demand-side worry: if AI leaders themselves are calling for a slower pace of development, the capex that funds both companies’ order books becomes a policy question as well as a commercial one. For a sense of how much institutional money sits in the same trade, see our look at BlackRock’s holdings of Nvidia, Micron and AMD.

The call: Micron vs Nvidia 12-month scenarios

Both sets of levels run to September 2027 and are anchored to the 14 September 2026 closes. Each bear case uses the same method: repeat the stock’s own recent drawdown from its record close.

Micron bull, $1,540: the analyst median target and 12.4 times the $124 annualised EPS implied by Q4 guidance. It requires the 30 September report to show Strategic Customer Agreement deposits large enough to convince investors that fiscal 2027 margins stay above 75%. Base, $1,050: 8.5 times run-rate EPS, a modest re-rating as HBM4 volumes ramp but before the Idaho supply arrives. Bear, $560: the average 53.8% drawdown of Micron’s last three memory downturns applied to the $1,213.56 record close. It arrives if DRAM contract prices roll over in 2027 as new capacity lands.

Nvidia bull, $330: close to the $327.65 average analyst target, and 35.5 times fiscal 2027 consensus EPS, which needs Vera Rubin and the $500bn-plus third-party financing platforms to keep revenue growth above 50% into fiscal 2028. Base, $250: today’s 22.7 times multiple on earnings about 18% above the fiscal 2027 consensus. Bear, $150: a repeat of the 36.9% drawdown of early 2025 from the $235.74 record close, or 16 times fiscal 2027 EPS, triggered by a capex pause or a receivables problem at a large customer.

The verdict: Micron has the larger upside case (+66.7% against +56.4%), but Nvidia offers the better 12-month return profile on both absolute expected value (+16.1% against +13.6%) and risk-adjusted terms. Dividing expected return by volatility gives 0.42 for Nvidia and 0.17 for Micron. Micron is the higher-return bet only for investors who can sit through another 40% drawdown, and only if the contracts prove the cycle has changed.

What would change my mind: if Micron’s 30 September filing shows remaining performance obligations climbing from about $5bn towards the size of a full quarter’s revenue, backed by cash deposits, I would move its bear probability down to 15% and Micron would lead on both measures. If Nvidia’s November report shows receivables again growing faster than revenue, its bear probability goes up. For the full single-stock ranges, see our Nvidia stock prediction and the memory names on our list of stocks that could be the next SanDisk.

FAQ: Micron vs Nvidia

Is Micron or Nvidia the better stock for higher returns?

On our 12-month scenarios Micron has the bigger upside case, +66.7% to $1,540 against +56.4% to $330 for Nvidia. Nvidia has the higher probability-weighted return, +16.1% against +13.6%, with less than half of Micron’s 80.7% volatility. Micron suits investors who accept memory-cycle drawdowns. Nvidia suits those who want AI exposure with a shallower bear case.

Why is Micron so much cheaper than Nvidia on earnings?

Micron trades at about 7.5 times the run-rate earnings implied by its fiscal Q4 guidance, while Nvidia trades at 22.7 times fiscal 2027 consensus EPS. The gap reflects history. Micron’s shares fell 53.7%, 49.8% and 57.8% in its last three memory downturns, so investors discount peak earnings that have repeatedly proved temporary.

Does Micron depend on Nvidia as a customer?

Less than the Micron vs Nvidia narrative suggests. Micron’s 10-Q never names Nvidia. It says one customer accounted for 10% of revenue in the first nine months of fiscal 2026, down from 16% a year earlier, and that HBM4 ships in volume for an unnamed lead customer’s platform. Nvidia’s latest release names a multi-year partnership with SK hynix.

When do Micron and Nvidia next report earnings?

Micron reports fiscal Q4 2026 results on Wednesday 30 September at 2:30 p.m. Mountain time, guiding $50.0bn of revenue and $31.00 of non-GAAP EPS. Nvidia’s Q3 fiscal 2027 quarter ends in late October; it has guided $108.0bn of revenue but had not announced its November report date by 15 September.

Why did Micron and Nvidia fall on 14 September 2026?

Semiconductors slid after tech leaders, including Anthropic’s Dario Amodei and OpenAI’s Sam Altman, warned about the risks of rapid AI development. Micron fell 5.25% to $924.03 and Nvidia 3.36% to $210.96, while the SOXX semiconductor ETF lost 5.63%. A 10-year Treasury yield near 5% is a further headwind for richly valued growth stocks.

This article is analysis, not investment advice. Scenario levels and probabilities are the author’s estimates based on public filings and market data as of 14 September 2026. Share prices can fall as well as rise, and investors can lose capital.